WO2012021759A2 - Distribution de puissance redondante parallèle - Google Patents

Distribution de puissance redondante parallèle Download PDF

Info

Publication number
WO2012021759A2
WO2012021759A2 PCT/US2011/047496 US2011047496W WO2012021759A2 WO 2012021759 A2 WO2012021759 A2 WO 2012021759A2 US 2011047496 W US2011047496 W US 2011047496W WO 2012021759 A2 WO2012021759 A2 WO 2012021759A2
Authority
WO
WIPO (PCT)
Prior art keywords
power
automatic transfer
parallel
set forth
switch
Prior art date
Application number
PCT/US2011/047496
Other languages
English (en)
Other versions
WO2012021759A3 (fr
Inventor
William Pachoud
Steve Chapel
Original Assignee
Zonit Structured Solutions Llc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zonit Structured Solutions Llc filed Critical Zonit Structured Solutions Llc
Priority to CA2807996A priority Critical patent/CA2807996C/fr
Publication of WO2012021759A2 publication Critical patent/WO2012021759A2/fr
Publication of WO2012021759A3 publication Critical patent/WO2012021759A3/fr

Links

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J9/00—Circuit arrangements for emergency or stand-by power supply, e.g. for emergency lighting
    • H02J9/04—Circuit arrangements for emergency or stand-by power supply, e.g. for emergency lighting in which the distribution system is disconnected from the normal source and connected to a standby source
    • H02J9/06—Circuit arrangements for emergency or stand-by power supply, e.g. for emergency lighting in which the distribution system is disconnected from the normal source and connected to a standby source with automatic change-over, e.g. UPS systems
    • H02J9/061—Circuit arrangements for emergency or stand-by power supply, e.g. for emergency lighting in which the distribution system is disconnected from the normal source and connected to a standby source with automatic change-over, e.g. UPS systems for DC powered loads
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J1/00—Circuit arrangements for dc mains or dc distribution networks
    • H02J1/10—Parallel operation of dc sources
    • H02J1/102—Parallel operation of dc sources being switching converters

Definitions

  • Embodiments of the present invention relate to design and operation of data centers and, in particular, to parallel redundant distribution of power including such distribution of power to critical equipment such as in medical contexts or in data center environments.
  • Data centers have a specific set of issues that they must face in relation to power supply and management.
  • Traditional techniques in this area were developed from prior industrial electrical practice in a time when a typical data center held very small numbers of mainframe computers and the change rate was low.
  • data centers often contain tens of thousands of electronic data processing (EDP) devices with high rates of change and growth.
  • EDP electronic data processing
  • Data centers are also experiencing rapidly growing power capacity demands driven, for example, by central processing unit (CPU) power consumption that is currently increasing at a rate of approximately 1.2 annually.
  • CPU central processing unit
  • Traditional techniques were not adopted to cope with these change rates, and data centers are therefore having great difficulty in scaling to meet those needs.
  • the branch distribution circuit is the area where most incidents that result in a loss of power to a receptacle typically occur. Indeed, this is where people tend to make changes in the types and amounts of load. Possibly the most common cause of electrical failure, then, is the branch circuit breaker being tripped by a person plugging in a load that exceeds the capacity of the circuit.
  • Power monitoring devices e.g., via plugstrips with amperage meters or Power Distribution Units (PDUs), wall mounted or free-standing units which contain distribution circuit breakers that are connected to power whips that power equipment racks on the data center floor
  • PDUs Power Distribution Units
  • wall mounted or free-standing units which contain distribution circuit breakers that are connected to power whips that power equipment racks on the data center floor
  • sudden changes in power draw can cause sudden problems, which would not be easily remedied by such devices.
  • data center staff or users can trip circuit breakers when they install new equipment, potentially causing service interruptions, which may not be detected using power monitoring devices in time to prevent the issue.
  • Another factor that contributes to power distribution issues is that many models of EDP equipment have only one power supply, and therefore one power cord.
  • the present invention relates to improved parallel distribution of power in various contexts including in data center environments.
  • the invention relates to providing improved automatic transfer switches, for switching between two or more power sources (e.g., due to power failures such as outages or power quality issues), as well as associated power distribution architecture, components and processes.
  • the noted utility involves an equipment rack having a number of ports for receiving equipment, e.g., where each port may have a height of 1 u, it will be appreciated that some equipment may occupy multiple ports.
  • the equipment rack system includes a number, N, of EDPs mounted in at least some of the ports of the rack and a number, S, of independently operating ATSs.
  • Each of the ATSs is configured to received input power from first and second external power sources, to detect a power failure (e.g., a power outage or unacceptable power quality) related to the first external source, and to automatically switch its output power feed to be coupled to the second external source when a power failure related to the first external source is detected.
  • a power failure e.g., a power outage or unacceptable power quality
  • the noted equipment rack system has a switch density ratio defined as S/N.
  • the equipment rack system has a switch overhead ratio, defined as a ratio of the number of ports occupied only by ATSs to the number of ports collectively occupied by the ATSs and the EDPs is less than S/(N+1).
  • the switch density ratio may be at least 1/4 (and more preferably at least 1/2), and the switch overhead ratio may be less than 1/5 (and more preferably, less than 1/8).
  • a rack may have a height of more than 30 u's and even more than 50 u's in some cases.
  • the present invention enables high switching densities while occupying no more than 2 of the ports with ATSs.
  • multiple ATSs e.g., 12 or more ATSs
  • a utility for use in supplying redundant parallel power to electronic data processing system.
  • the utility involves a number of automatic transfer switches disposed within an enclosure sized to fit within a single standard equipment rack space.
  • Each of the ATSs is configured to receive a first power feed from a primary power source disposed external to the enclosure, receive a second parallel power feed from a secondary power source disposed external to the enclosure, detect a power failure on the first parallel power feed, and automatically switch an output power feed from being electrically coupled with the first parallel power feed to be electrically coupled with the second parallel power feed when the power failure in the first parallel power feed is detected.
  • a number of ATSs can be disposed in a single rack space.
  • the enclosure has a height of no more than approximately 1.5u and may have a height of 1 u, i.e., no more than about 1.75 inches.
  • each ATS preferably has a power density of at least about 2 kilowatts per 10 cubic inches. In this manner, substantial switching capacity can be provided within the spatial-envelope of one or two u's of a standard rack.
  • multiple ATSs may be disposed in a single housing that may occupy less than two, for example, one u of rack space. For example, 12 or more ATSs, each having a power density of 2 kilowatts can be contained in 1.5 u's of rack space or less.
  • Such embodiments allow for elimination of plugstrips along side equipment in the racks, thus, allowing for narrower racks and more efficient use of data center floor space.
  • FIG. 1 shows a example of a power distribution topology
  • FIG. 2 shows a power distribution designed for a typical data center
  • FIG. 3 shows how power efficiencies vary with load changes for double conversion UPS units used in data centers
  • FIG. 4 shows illustrates the increase in large data centers that use large numbers of servers in recent years;
  • FIG. 5 shows illustrates a switch in accordance with the present invention;
  • FIG. 6 shows examples of a hydra cord;
  • FIG. 7 shows a system diagram of an illustrative micro- ATS, according to various embodiments.
  • FIG. 8A shows a circuit diagram of an illustrative power supply subsystem in context of an illustrative "A" & "B" power switching subsystem for use in some embodiments of a micro-ATS;
  • FIG. 8B shows illustrative detail of the 15 -volt power supply as normally supplied by HV through a set of resistors
  • FIG. 9 shows a circuit diagram of an illustrative "A" power voltage range detect subsystem for use in some embodiments of a micro-ATS
  • FIG. 10 shows a circuit diagram of an illustrative "A" power loss detect subsystem for use in some embodiments of a micro-ATS;
  • FIG. 11 shows a circuit diagram of an illustrative "B" power synchronization detection subsystem for use in some embodiments of a micro-ATS;
  • FIG. 12 shows a circuit diagram of an illustrative "A"/"B" synchronization integrator subsystem in context of the "B" power synchronization detection subsystem and the "A" power loss detect subsystem for use in some embodiments of a micro-ATS;
  • FIG. 13 shows a circuit diagram of an illustrative timing control subsystem for use in some embodiments of a micro-ATS
  • FIG. 14 shows a circuit diagram of an illustrative "A” & "B" power switching subsystem for use in some embodiments of a micro-ATS;
  • FIG. 15 shows a circuit diagram of an illustrative disconnect switch subsystem for use in some embodiments of a micro-ATS
  • FIG. 16 shows a circuit diagram of an illustrative output current detect subsystem for use in some embodiments of a micro-ATS
  • FIG. 17 shows a circuit diagram of an illustrative piezoelectric device driver subsystem for use in some embodiments of a micro-ATS
  • FIG. 18A shows a power distribution topology having an ATS disposed in the root nodes of the topology
  • FIG. 18B shows another illustrative a power distribution topology having an ATS disposed further downstream, in the distribution nodes of the topology;
  • FIG. 18C shows yet another illustrative a power distribution topology having an ATS disposed even further downstream, in the leaf nodes of the topology
  • FIG. 18D shows still another illustrative a power distribution topology having ATSs disposed still further downstream at the EDP equipment, in the end leaf nodes of the topology;
  • FIG. 19 shows an illustrative traditional power distribution topology, according to some prior art embodiments.
  • FIG. 20 illustrates an efficiency versus load graph for a typical double-conversion UPS unit
  • FIG. 21 shows an illustrative power distribution topology, according to various embodiments.
  • FIGS. 22A and 22B show illustrative parallel micro-ATS modules, according to various embodiments.
  • FIG. 23 shows an illustrative power distribution topology that includes a rack-mounted parallel micro-ATS module, according to various embodiments.
  • Section 1 discusses the issues involved in data center power distribution and inventive solutions to those problems and Section 2 discusses detailed methods that can be used to construct an automatic transfer switch with the characteristics needed to build the inventive solutions described in and associated rack/data center architectures. It should be noted that the detailed methods described can also be used for other purposes than constructing automatic transfer switches.
  • the branch distribution circuit is the area where most incidents that result in a loss of power to a receptacle occur.
  • the cause is simple, this is where people make changes in the types and amounts of load.
  • the most common cause of electrical failure is the branch circuit breaker being tripped by a person plugging in a load that exceeds the capacity of the circuit.
  • each branch circuit must usually only be loaded to 75% of its capacity, to account for the "inrush load” that occurs during a cold start, when all of the connected electronic data processing (EDP) equipment is powering up simultaneously, spinning up fans, disk drives, etc. This is the highest load scenario, and if not accounted for, it will trip the branch circuit breaker when it happens.
  • EDP electronic data processing
  • Power monitoring is often used (via plugstrips with amperage meters or Power Distribution Units (wall mounted or free-standing units which contain distribution circuit breakers that are connected to power whips that power equipment racks on the data center floor) to determine the current power draw.
  • Power Distribution Units wall mounted or free-standing units which contain distribution circuit breakers that are connected to power whips that power equipment racks on the data center floor
  • equipment rack to describe both of the terms “equipment rack” and “equipment cabinet”, which are often used for 2-post vs. 4- post racks.
  • a method of increasing power reliability is to use an automatic transfer switch (ATS) at various points in the power distribution topology to provide for automatic failover from a primary power source to a backup power source. This is typically done at one of three points in the power distribution topology, the panelboard on the wall, where the branch circuits originate, the end of the branch circuit in the rack where the power is fed to plugstrips or between the plugstrip and the EDP equipment being powered.
  • ATS automatic transfer switch
  • FIG. 1 shows an example power distribution topology.
  • the root(s) of the graph(s) of the power distribution topology are "downstream" of the core power infrastructure.
  • the root(s) start at a UPS unit or a power distribution panelboard.
  • the power distribution panel is either a root or a distribution layer node.
  • the branch circuits originate at the power distribution panel and end at the equipment racks.
  • power is distributed via plugstrips (which are confusingly also called power distribution units (“PDU”) a term that is often applied to the panelboard.).
  • PDU power distribution units
  • the plugstrips may have circuit breakers in them, also.
  • panelboards and plugstrips we will use the terms panelboards and plugstrips.
  • ATS switches can be used with panelboards and will switch every branch circuit in a given panelboard to a secondary power source when the primary power source fails.
  • the primary design issue with this methodology is if the ATS at the panelboard fails, many EDP devices will be deprived of power.
  • a typical panelboard has a capacity of 225KVA, and 84 or 96 circuit breaker stations. This can power approximately up to 40 racks via 28-96 branch circuits (depending on the type and number of branch circuits and the average number of watts used per rack). Having 40 racks go down due to ATS failure in a data center is a major hit that can have very serious service impacts. This type of failure has happened in numerous data centers.
  • a relay has two primary loss factors, the contact area of the relay and if the relay requires power to keep it in the "on" state, where it is conducting current.
  • the shape and material of the contacts is carefully chosen and engineered to minimize resistance across the contacts, yet minimize or prevent arcing across the contacts when they are switching. Also, since some arcing may occur in some circumstances, the contacts must be designed to minimize the possibility of the arc "welding" the contacts shut, which is very undesirable.
  • the contacts are mounted (usually on an armature) so that they can be moved to accomplish their switching function.
  • the contact mass, range of motion, mechanical leverage and force used to move the armature are all relay design issues.
  • the range of motion is dictated by the gap needed between the contacts to minimize arcing at the maximum design current level. As the maximum design current is increased, the gap must also increase.
  • the mass of the contact must be accelerated by the force applied to the armature, which has a practical limit. These factors impose a limit on the amount of current that can be sent through a pair of contacts and still maintain an acceptable transfer time for EDP equipment.
  • EDP equipment CBEMA guidelines recommend a maximum of approximately 20 milliseconds of power outage for continued operation of modern switched power supplies. If the mass of the armature and contact gap are too large, the relay transfer time exceeds this time limit.
  • Well designed relay based automatic transfer switches have a loss factor of about 0.5 % or less. They also have power supplies to power their internal logic that typically use in the range of 12-20 watts in operation. b. Solid State ATS
  • switches use solid state semiconductors to accomplish switching between their input power sources and their output load. They can switch faster than relay based switches, because they use semi-conductor based switching, not mechanical relays. However, the semi-conductors have a loss factor and the efficiency of this type of switch is less than that of a relay based switch, typically around 1%. Also, they are usually less reliable, unless they are built with redundant internal failover capability, which makes them much more expensive. Again, they also have power supplies to power their internal logic that typically use in the range of 12-200 watts or more in operation, depending on the size of the transfer switch, and the level of redundancy offered by the switch.
  • Rack space in a data center is expensive.
  • the data center infrastructure of generators, UPS units, power distribution, raised floor, computer room cooling, raised floors, etc. is a very large capital investment and a large ongoing operational expense.
  • 1U of rack space in a standard 42U equipment cabinet is 2.5% of the space available in that rack.
  • Putting rack mounted automatic transfer switches in large numbers in equipment racks uses a lot of rack space, which represents a loss of space that can be used for EDP equipment. This is very undesirable, which is one reason it is not done.
  • the transfer switch(s) that are upstream of the UPS units are part of the core power infrastructure not the power distribution. Automatic transfer switching is done in the core infrastructure to insure continuity of connection to a valid power source, such as utility power grid feeds or generators.
  • the transfer time of relay based switches that can handle the power capacities required in the core infrastructure is too slow to avoid (a time of 20 milliseconds or less is recommended for EDP equipment by CBEMA guidelines) shutdown by connected EDP equipment for the reasons described earlier. This is why transfer switches of this type are placed upstream of the UPS units where the brief power outages that these switches create on transfer are covered by the UPS units.
  • the Zonit ⁇ 8TM is very small (4.25"xl .6"xl" ⁇ 10 cu. inches), very efficient ( ⁇ 2V @ maximum load loss) and requires no rack space, since it can be self-mounted on the back of each EDP device or incorporated in the structure of the rack outside the volume of the rack used to mount EDP equipment or in rack mounted plugstrips or in a in-rack or near-rack Power Distribution Unit, due to its very small form-factor. It should be noted that the ⁇ 8TM is small enough that it could be integrated into EDP equipment also.
  • This small form factor also helps enable the usage of 24" outside-to-outside width EDP equipment cabinets, which have two key advantages, they fit exactly on 2'x2' raised floor tiles which makes putting in perforated floor tiles to direct air flows easy, since the racks align on the floor tile grid and they saves precious data center floor space.
  • NEMA equipment racks are not standardized for overall rack width, and the narrower the rack is, the more racks can be fit in a given row length. For example a 24" rack will save 3" over the very common 27" width racks and that represents one extra rack for each 8 equipment racks in a row. This is now practical with modern EDP equipment, since almost all models now utilize front to back airflow cooling.
  • the ⁇ 8TM allows efficient, cost-effective and rack space saving per device or near per device (ratios of 1 ⁇ TM to 1 EDP device or 1 ⁇ TM to a low integer number of EDP devices) highly parallel and highly efficient auto-switched power distribution methods to be utilized. It should be pointed out that the ratio of ⁇ TM units to EDP equipment can be selected to optimize several interrelated design constraints, reliability, cost and ease of moving the EDP device in the data center. The 1 to 1 ratio maximizes per device power reliability and ease of moving the device while keeping it powered up.
  • Typical data centers use a power distribution design as shown below in FIG. 2.
  • UPS Uninterruptible Power Supply
  • flywheel UPS flywheel UPS devices
  • the best double conversion UPS units used in data centers have power efficiencies that vary as their load changes as shown in FIG. 3. They typically average 85-90% efficiency, flywheel UPS units average -94% efficiency at typical load levels. This level of efficiency was acceptable when power costs were stable, relatively low and the climate impacts of carbon based fuels was not fully appreciated. Power is now quickly changing from an inexpensive commodity to an expensive buy that has substantial economic and environmental costs and key implications for national economies and national security.
  • a traditional UPS powered data center more typically has efficiencies in the 88-92% range, because no data center manager wants to run his UPS units at 100% capacity, since there is no margin for any needed equipment adds, moves or changes.
  • the load between the UPS units is commonly divided so that each has approximately 1/2 the load of the total data center.
  • neither UPS can be loaded above 50%> since to be redundant, either UPS must be able to take the full load if the other UPS fails. This pushes the UPS efficiency even lower, since each unit will usually not be loaded up above 40-45%) so that the data center manager has some available UPS power capacity for adds, moves and changes of the EDP equipment in the data center.
  • FIG. 4 illustrates an important point.
  • the number of very large data centers that house extremely high numbers of servers has been on the increase for the last five years or more.
  • the server deployment numbers are huge.
  • Commodity servers have great pressure to be cost competitive, especially as regards their initial purchase price. This in turn influences the manufacturers product manager to choose the lowest cost power supply solution, potentially at the expense of best power efficiency, an issue that has impacts that will be discussed further below.
  • FIG. 4 Data Center Size and Server Counts
  • Typical Modern EDP power supplies are almost all auto-ranging (accept 110-240V input) and all switched (Draw on the Alternating Current [AC] input power for just a short period of time and then convert this energy to Direct Current [DC], then repeat). Power supplies of this type are more resistant to power quality problems, because they only need to "drink" one gulp at a time, not continuously. If the input AC power voltage range is controlled within a known range, they will function very reliably. They do not require perfect input AC waveforms to work well. All that is necessary is that they receive sufficient energy in each "gulp" and that the input power is within the limits of their voltage range tolerance. This makes it possible to use a data center power distribution system that is much more efficient than a fully UPS supplied power system at a very reasonable capital expense.
  • TVSS Unit will filter transient surges and spikes, but it does not compensate for long periods of input power over-voltage, these are passed through.
  • the data center power system we are discussing must deal with out of range voltage (since modern power supplies are not damaged by under-voltage but will shutdown) by switching to the conditioned UPS power if the utility line power voltage goes out of range.
  • Voltage sensing and auto-switching could be put in at other points in the data center power system, but for the reasons discussed earlier, the options we present are the most feasible.
  • Step-down transformers of this type are available and can be ordered from utility companies. They have a set of taps on their output coil and automatically switch between them as needed to control their output voltage to a specified range. Step-down transformers of this type of this type are not usually deployed for cost reasons by utility companies, but they can be specified and retrofitted if needed.
  • the second place in the data center power system that over-voltage protection can be implemented is at an ATS in the power distribution topology. This can be done at the ATS at a panelboard or at an ATS at the end of a branch circuit, or at an ATS at the device level. The last is what we chose for reasons that are discussed later. It should be noted that a semiconductor based ATS could be used upstream of the UPS, but this is very expensive and the results of a failure of the ATS are potentially catastrophic, all of the powered EDP units could have their power supplies damaged or destroyed if the ATS unit fails to switch. This is a large downside to chance.
  • EDP devices share the load among all available power supplies in the device. It is possible to build an EDP device that switches the load between power supplies, so that only one or more are the active supplies and the others are idle, but as described earlier, this is rarely done for both cost and reliability reasons. To insure that multi-power supply EDP devices draw on only filtered utility line power if it is available and switch to the UPS if it is not, each of the secondary power supply units needs to be auto-switched between the utility line and UPS unit(s). Otherwise, the UPS unit will bear a portion of the data center load, lowing the overall efficiency of the power distribution.
  • Utility line power is outside the control of the data center operator. It can be affected by weather, equipment faults, human error and other conditions. It can fail intermittently which poses a potential hazard to the core data center power infrastructure. If utility power goes on & off intermittently and the timing of the on-off cycles is within a certain range, auto-switching between the utility line source and the generator (even filtered by the UPS units) can result in harmonic reinforcing power load surges being imposed on the generator. This can happens as follows: i. The utility line power fails
  • the generator stabilizes and is switched into the system, feeding the UPS v.
  • the utility line power returns, then goes off again
  • the generator stabilizes and is switched into the system, feeding the utility line power side of the system. Note: This is done in preference to feeding through the UPS in order to maintain redundant feeds to the racks w/ EDP equipment.
  • the generator overshoots the high voltage cutoff value of the highly parallel ATS units and they switch back to UPS removing the load from the generator. vii. The generator then throttles back and its output voltage returns to normal levels. viii. The highly parallel ATS units switch back to the generator, causing it to sag again
  • Steps vi-viii repeat and can cause a harmonic reinforcing power load surge to build up and destabilize the generator.
  • the solution that we have selected for these problems is to auto-switch at the device or near device level in the power distribution topology. This solution has a number of benefits over other methods of auto-switching which will be discussed, but requires an automatic transfer switch with specific characteristics.
  • the chosen auto-switch needs to have the following qualities.
  • Standard NEMA racks are only standardized in one dimension, the width of the equipment that is mounted in the rack as reflected in the spacing of the vertical mounting flanges used in the rack (and the spacing of the fastener holes in those vertical mounting flanges). They are not standardized for total overall width, height or depth.
  • the depth is generally limited to what the projected maximum depth of the mounted EDP equipment will be.
  • the current maximum depth of most EDP equipment is around 36" with a few exceptions.
  • the overall width of the rack is dependant on what kinds of cabling, power distribution devices and sometimes cooling devices the rack designer may want to support mounting on the sides of the rack, outside the volume occupied by EPD equipment mounted in the rack.
  • the NEMA standard equipment width that is most commonly used is 19".
  • ATS should mount in a zero-U fashion or otherwise be integrated into or near the rack without using rack space that could be used to mount EDP equipment. It could be integrated into EDP equipment directly. It could also be integrated into plugstrips or in-rack or near-rack power distribution units such as the Zonit Power Distribution Unit (ZPDU), which trades a slight amount of rack space usage against access at the rack to the circuit breakers controlling power to the plugstrips in the racks. In this case the ATS function must be integrated into every sub-branch output of the ZPDU, so that each one is auto-switched. This is a potentially worthwhile trade-off to some data center managers. As discussed earlier, rack space is very expensive. It is not cost effective to use it for device level ATS units.
  • ZPDU Zonit Power Distribution Unit
  • the Zonit ⁇ TM has all of the needed qualities. Its design is specific to that set of requirements, and incorporates patent pending means of accomplishing each of those requirements.
  • the ⁇ TM is designed to always use the primary power source if it is available and of sufficient quality
  • the ⁇ TM monitors voltage on the primary input and switches to the secondary source if it is out of range. It switches back to the primary source when it returns to the acceptable range and is stable.
  • the ⁇ TM switches from A to B in 14-16 milliseconds. Faster switching times are can be achieved, but the times chosen maximize rejection of false conditions that could initiate a transfer.
  • the B to A transfer times are approximately 5 milliseconds once initiated. This is possible because most B to A transfers occur after A power has returned and therefore the ⁇ TM can pick the time to make the transfer, both power sources are up and running.
  • the ⁇ TM "spreads" the load on the source being transferred. This is by design. A population of ⁇ TM units will have a small degree of variability in their timing of transfers from the B source to the A source, which is not much in real time but is significant in electrical event time. This variance "spreads" the load being transferred as seen by the power source, for example a generator or UPS unit. This is because the load appears as a large number of ⁇ 8TM transfers in a time window to the power source. This is beneficial to generators and UPS units, since it distributes a large number of smaller loads over a period of time, thus reducing the instantaneous inrush. This is another advantage of our power distribution method.
  • the ⁇ 8TM waits a specified time constant when on B power before transferring back to A power (unless B power fails, then it transfers immediately to A power).
  • the time is selected to be outside of the normal response time characteristics of most typical generators. This prevents Harmonic Reinforcing Load Surges because the generator has time to adapt to the load change and stabilize its output.
  • the ⁇ 8TM is a very small form factor. It can implemented as a self mounting device that fits onto a 1U EDP device as shown in U.S. Patent Application Serial No. 12/569,733, which is incorporated herein by reference. It can be deployed as a true "zero U” solution.
  • the ⁇ 8TM is extremely efficient, using less than 100 milliwatts when on the primary power source in normal operational mode.
  • the ⁇ TM is very inexpensive to make. It design is such that its expected useful lifetime is 20 years or more and this could easily be extended to 35 years+ by using components with longer lifetimes, which would raise the cost slightly but could be a worthwhile tradeoff.
  • the ⁇ TM is moved from each generation of deployed EDP equipment to the next deployed generation and will return a very low annual amortized cost over its expected service lifetime.
  • the ⁇ TM is very reliable. This is a requirement and also a consequence of designing the device for a very long service lifetime.
  • ATS units in parallel at the device or near device level vs. ATS switches closer to the root of the power distribution helps to address these limitations and increase power distribution efficiency. This is because many ATS units working in parallel have a cumulative relay contact area that is much greater than is feasible to put in a higher capacity relay based ATS unit regardless of where that unit is placed in the power distribution topology.
  • the ATS units in parallel also can easily have a quick enough transfer time because they use smaller relay contacts with quicker transfer times.
  • the ⁇ 8TM is a more efficient than traditional ATS units of the same power handling capacity by a factor of 10 or more. This is a required characteristic to make highly parallel auto-switched power distribution practical. Otherwise the net result would be to consume more power not less, regardless of the capital expense of the switching units used.
  • the space in a data center equipment rack or cabinet is very expensive, a point that we covered earlier.
  • the ⁇ 8TM does not consume any rack space and is small enough to be integrated into the rack structure outside of the volume in the rack where EDP equipment is mounted.
  • a large server farm in a data center often will consist of many "pizza- box" servers in a rack with perhaps a network switch. Each server may use ⁇ 3-6 watts of 120V power. This means that a 15A ATS can only handle 2-4 servers. If the ATS units are 1U rack mounted devices, then using the median value of 3 servers per 15A ATS, 25% of the rack space devoted to servers would be consumed by ATS devices! This is too inefficient a use of expensive rack space to be practical.
  • a method and apparatus are provided for distributing power via receptacles (or hard-wired output cords) as is shown in FIG. 5.
  • This apparatus has two power inputs, one from an "A” source, the other from a “B” source.
  • the amperage of the "A” and “B” power sources can be chosen to match the number of auto-switched output receptacles and their anticipated average and/or maximum power draw.
  • the apparatus takes input power from the "A” and "B” sources and distributes it to a number of Zonit Micro Automatic Transfer Switch Modules contained in the enclosure (either as separately deployed modules or modules combined onto one or more printed circuit boards) of the apparatus.
  • the "A” and “B” power sources may be single phase, split-phase or three-phase, but in a preferred instantiation, both would be identical.
  • Each of the Zonit Micro Automatic Transfer Switch Modules feeds an output receptacle (or hard wired power cord) located on the face of the enclosure of the apparatus.
  • Two or more circuit breakers may be provided to allow disconnecting the unit electrically from the branch circuits that feed it.
  • Additional "Virtual Circuit Breaker" control switches and indicators may also be included to provide a means to disconnect end-user equipment from individual Zonit Automatic Transfer Switch modules.
  • the apparatus can be mounted within the rack or on top of it or on its side.
  • the Zonit ⁇ 8 modules can be constructed as separate or combined circuit boards to optimize ease and cost of manufacture.
  • the enclosure takes up rack space, by eliminating the need for in rack plugstrips, which are usually mounted vertically in the rack, data center floor space can be optimized as follows.
  • the NEMA standard equipment width that is most commonly used is 19".
  • a multi-head power cord "hydra cord” can be provided.
  • This cord can be hardwired into the apparatus shown in FIG. 5 or receptacles can be provided. An example is shown in FIG. 6.
  • the number of output heads on the hydra cord can be varied to match the desired average power output (or summed power output to a chosen set of end-user devices) to each connected end-user device.
  • the length and gauge of the hydra power cord (both the main feed section and the separate feeds to each "hydra head”) can be optimized to minimize electrical transmission losses and power cord tangle by optimizing the cord lengths for each hydra cord to supply power to a particular set of equipment positions in the equipment rack.
  • a set of appropriately sized hydra cables can be used to feed each equipment location in the rack at whatever interval is desired, such as one uniform equipment mounting space "1U” of 1.75 vertical inches.
  • the pattern of which heads from which hydra cords feed which "U" positions in the rack can also be varied to control which power phase and source feed each "U” positions for whatever reason is desired, for example to balance power phase usage. This is an example usage of the technology described in US Patent 6,628,009 the contents of which are incorporated herein as if set forth in full.
  • locking power cord technologies can be used to improve the security of power delivery.
  • the apparatus could for example, be equipped with standard NEMA L5-15 locking receptacles for 120V service or NEMA L6-15 receptacles for 200V+ service. Other locking receptacle types could be used.
  • the "hydra cord head" on the output cords can be equipped with IEC locking technologies (IEC C13 and C19 would be the types most commonly used in IT equipment) using the technologies described in PCT Applications PCT/US2008/057149 and PCT/US2008/05048 the contents of which are incorporated herein as if set forth in full.
  • the apparatus described can be incorporated in the technologies described in US Patent 6,628,009 and PCT Applications PCT/US2008/057140 and PCT/US2010/050550 the contents of which are incorporated herein as if set forth in full.
  • This apparatus incorporates the parallel auto-switching functionality in an instantiation of the the Zonit Power Distribution System a novel implementation of which is shown in U.S. Patent No. 6,628,009.
  • a preferred instantiation would adopt the rule "if one phase of a polyphase power source fails, all phases switch to the alternate power source". To do this the logic each of the single-phase automatic transfer switches would be modified to achieve this functionality.
  • An additional possible refinement would be to use two (or more) uATS modules to monitor each of the phases of the polyphase power source.
  • the two (or more) modules would act as primary and backup logic for determining when to switch the relays (of which there would normally only be on set of relays per power phase of the polyphase power source) from the A to the B power source and back from B to A.
  • Another possibility is to use more than two modules and use a majority approach to decide when to switch the relays (of which there would normally only be on set of relays per power phase of the polyphase power source) from the A to the B power source and back from B to A.
  • An example would be to use three modules and set the logic such that at least two of three agree to switch power sources before the switch can be made.
  • the advantage of these "multi-uATS" approaches is that they eliminate single points of failure in the polyphase switching apparatus.
  • UPS unit is at idle and will only be used if the primary unit fails.
  • the UPS units must be designed to handle this type of load being immediately placed on them, but almost all modern UPS units can do this. The result is an increase of -3-5% in the efficiency of the data center, a useful improvement. It should be noted that while only one pair of UPS units is discussed here the methodology scales to larger data centers that have many UPS units deployed in pairs for redundancy.
  • Every single power supply (or corded) EDP device will be connected via a ⁇ 8TM to the A and B UPS units.
  • Every dual or N+l power supply EDP device will have one power supply connected to the A UPS via a normal power cord, the second or all other N+l power supplies will be connected to the A and B UPS units via ⁇ 8TM units. This insures that when the A UPS unit is available it takes all of the load and when it is not the B UPS carries the load.
  • the ⁇ 8TM units can be deployed in one to one per device ratios, or low integer number ratios, respecting ⁇ 8TM power capacity limits.
  • EDP devices share the load among all available power supplies in the device. It is possible to build an EDP device that switches the load between power supplies, so that only one or more are the active supplies and the others are idle, but as described earlier, this is rarely done for both cost and reliability reasons. To insure that multi- power supply EDP devices draw on only filtered utility line power if it is available and switch to the UPS if it is not, each of the secondary power supply units needs to be auto- switched between the utility line and UPS unit(s). Otherwise, the UPS unit will bear a portion of the data center load, lowering the overall efficiency of the power distribution, which is undesirable.
  • the micro-ATS (e.g., the Zonit ⁇ 8TM) is very small (e.g., 4.25-inches x 1.6-inches x 1-inch, or less than 10 cubic inches) and very efficient (e.g., less than 0.2 volts at maximum load loss).
  • Certain implementations use no rack space, as they are self-mounted on the back of each EDP device, incorporated in the structure of the rack outside the volume of the rack used to mount EDP equipment, incorporated in rack-mounted plugstrips, or incorporated in an in-rack or near-rack Power Distribution Unit (i.e., any of which being possible due to the small form-factor of the micro-ATS).
  • the micro-ATS is small enough to be integrated directly into the EDP equipment itself.
  • Various embodiments of micro-ATSs are described herein, including their various components.
  • the micro-ATS embodiments are described as switching between two separate power sources, "A" and "B.”
  • the A and B power sources are single-phase sources.
  • polyphase power sources are connected. Where polyphase power sources are connected, polyphase embodiments of micro-ATSs are used. Substantially the same components (e.g., circuits) described herein with reference to the single-phase implementations are applicable to the polyphase implementations.
  • polyphase embodiments can be implemented as multiple single-phase micro-ATS units acting in parallel, with additional functionality provided for synchronizing certain of the control circuits so that they act together across the multiple ATS units to handle switching and return from one polyphase source to the other polyphase source and back.
  • Various embodiments of polyphase micro-ATSs can also have different conditions under which to switch power sources. For example, given three phase power with X, Y, and Z "hot" leads, a fault on any of three might be considered reason to switch from the A to the B polyphase source. To return to the A polyphase source, it may be desirable to ensure first that all three hot leads are present, stable, and of sufficient power quality on the A source.
  • FIG. 7 a system diagram of an illustrative micro-ATS 700 is shown, according to various embodiments. As illustrated, the micro-ATS 700 is connected to an "A" power source 760 and a "B" power source 765, and uses its various components to provide output power 770 to one or more devices or distribution topologies (e.g., to one or more EDP devices in a branch circuit of a data center).
  • the micro-ATS 700 is connected to an "A" power source 760 and a "B" power source 765, and uses its various components to provide output power 770 to one or more devices or distribution topologies (e.g., to one or more EDP devices in a branch circuit of a data center).
  • the micro-ATS 700 includes a power supply subsystem 705, an "A" power voltage range detect subsystem 710, an "A” power loss detect subsystem 715, a “B” power synchronization detection subsystem 720, an "A"/"B” synchronization integrator subsystem 725, a timing control subsystem 730, an "A” & “B” power switching subsystem 735, an output current detect subsystem 740, a disconnect switch subsystem 745, and a piezoelectric device driver subsystem 750.
  • Embodiments of the power supply subsystem 705 include innovative ways of powering control circuitry of the micro-ATS 700.
  • Embodiments of the "A" power voltage range detect subsystem 710 determine if the power being supplied to the micro-ATS 700 is in a desired (e.g., predetermined) voltage range.
  • Embodiments of the "A” power loss detect subsystem 715 determine when the "A" power being supplied to the micro-ATS 700 has been lost using desired discrimination characteristics.
  • Embodiments of the "B" power synchronization detection subsystem 720 measure the timing of the alternating current waveform of the "B" power.
  • Embodiments of the "A"/"B" synchronization integrator subsystem 725 provide synchronization of "B" to "A” transfers at zero voltage crossing times and "A" to “B” integration functionality.
  • Embodiments of the timing control subsystem 730 control when the selected power source to the micro-ATS 700 is switched, either the "A” source to the “B” source or from the “B” source to the “A” source, and can handle over-current condition switching and relay sequencing.
  • Embodiments of the "A” & “B” power switching subsystem 735 control actual switching between the "A” and “B” power sources in either direction to change which supply is acting as the input power source to the micro-ATS 700.
  • Embodiments of the output current detect subsystem 740 detect and measure presence and various characteristics of output current from the micro-ATS 700, and can, in some embodiment, mimic characteristics of a fuse so that the micro-ATS 700 can protect itself without blowing actual physical fuses (i.e., which must be replaced).
  • Embodiments of the disconnect switch subsystem 745 disconnect a secondary power source from the power supply when it is not in use.
  • Embodiments of the piezoelectric device driver subsystem 750 implement innovative techniques for driving piezoelectric or other devices.
  • each component is described below as performing particular functionality in the context of the micro-ATS 700. It will be appreciated that other configurations are possible in which similar or identical functionality can be implemented using other components, combinations of components, etc. Further, in some cases, values are given for components such as resistors and capacitors, etc. and ranges are given for current, voltage, and/or other power characteristics. These values and ranges are intended to add clarity to illustrative examples and should not be construed as limiting the scope of embodiments.
  • FIG. 8A shows a circuit diagram 800 of an illustrative power supply subsystem 705a in context of an illustrative "A" & “B" power switching subsystem 735 for use in some embodiments of a micro-ATS 700.
  • the micro-ATS 700 is connected to an "A" power source 760 and a "B" power source 765.
  • the power supply subsystem 705a performs a number of functions, including power conditioning (e.g., current limiting and power clean-up).
  • the source power for the power supply subsystem 705 a is acquired from the center taps of RY3 H ("Hot") and RY1 N ("Neutral”).
  • the source of power for the ATS Power Supply is from the "A" side when the output of the micro-ATS 700 is on the “A” Side, and on the "B” Side when transferred to the "B” side.
  • Components of the "A” & “B” power switching subsystem 735 serve as the automatic transfer switch for the micro-ATS 700 power supply subsystem 705a.
  • the current available from output power is limited by R3.
  • ZD 10 limits the full wave rectified output of the bridge BR4 to 150V peak. ZD 10 guarantees that CI does not exceed its rated voltage. CI stores enough charge to allow HV ("High Voltage") operations of the relays during transitions between the "A" power source 760 and the "B” power source 765.
  • Bridge BR6 normally blocks A2, as will be discussed more fully below.
  • GC On When a fault occurs, GC On will be pulled almost down to Common, and positive power will be available at U5 LED via ZD11. This turns on the U5 transistor.
  • the U5 transistor and Q3 form a Darlington pair that shorts the bridge BR6, allowing A2 to drive HV through diode D9.
  • A2 also drives LV ("Low Voltage") through diode D10.
  • C20 provides filtering and storage for LV.
  • the 15-volt power supply is normally supplied by HV through R63, R64 and R65. These three resistors drop the voltage and limit the current for Zener diode ZD2.
  • ZD2 regulates the voltage for the comparator and other electronics on the control board.
  • C17 further filters out the 15-volt signal.
  • A2 provides power for the 15V supply through LV.
  • HV is pulled down to 45V in this condition due to current limiting capacitor C16 mentioned earlier.
  • Embodiments of the power supply subsystem 705, such as the embodiments illustrated and described with reference to FIGS. 8A and 8B, provide a number of innovative features.
  • One such feature is that some embodiments of the power supply subsystem 705 acts as a transformer-less, very high-efficiency power supply.
  • the circuit produces low and high- voltage DC power from multiple AC inputs that is suitable for powering both low-voltage control circuitry and higher power relays. It does so very efficiently and with a minimum of expensive analog parts.
  • some embodiments of the power supply subsystem 705 provide capacitor current limiting for very low-power usage. As illustrated above, power consumption of the power supply subsystem 705 is limited by use of a capacitor. This can efficiently limit the power supply capacity to a desired value, thereby providing maximum efficiency and low power consumption.
  • embodiments of the power supply subsystem 705 provide optical isolation to reduce or even eliminate cross currents
  • embodiments of the control and synchronization subsystems e.g., the "B" power synchronization detection subsystem 720, the "A"/"B" synchronization integrator subsystem 725, the timing control subsystem 730, etc.
  • some embodiments of the power supply subsystem 705 provide power control relays (e.g., illustrated as RY1, RY2 and RY3) to direct the source power to the output of the micro-ATS 700 as well as provide the internal source selection (transfer switch function) for powering the micro-ATS 700 power supply subsystem 705.
  • some embodiments of the power supply subsystem 705 use an optically isolated disconnect circuit to prevent cross-source currents when the micro-ATS 700 is in the over-current fault mode. In this mode, there is no power delivered to the output, and thus, power must still be delivered to the micro-ATS 700 control and relay drive circuitry. As illustrated above, this can be accomplished via the A2 775 power path, and controlled by BR6 and optical isolation control U5.
  • FIG. 9 shows a circuit diagram of an illustrative "A" power voltage range detect subsystem 710a for use in some embodiments of a micro-ATS 700.
  • Embodiments of the "A" power voltage range detect subsystem 710a receive "A" power 760 nodes AH and AN.
  • the "A” power 760 is full wave rectified by bridge BR2.
  • C3 is used to limit the current available in this and the "A" power loss detect subsystem 715, as will be described below.
  • the "A" power voltage range detect subsystem 710a can generate the A (ON), A (COM), and CQ18 signals to the "A" power loss detect subsystem 715.
  • the illustrated "A" power voltage range detect subsystem 710a includes over- voltage detection and under-voltage detection functionality.
  • D18 half-wave rectifies the "A" power 760 and drives a ladder comprised of R14 and R27, which charges C4.
  • ZD6 and Q36 will begin to conduct drawing current through resistors R7, R74, R6, R5, and Rl .
  • This will turn on Q34, which will pull up the voltage on C4 through R17.
  • This will latch Q35, Q36, and Q34 in an on state.
  • This also will pull current through R8, thereby turning on the over- voltage indicator, LED2.
  • C2 will be charged and ZD5 will conduct.
  • Q32 and Q31 will conduct, turning off Q37 and Q38. This turns off A (ON). In the illustrated embodiment, this will tend to occur at around 135 VAC at AH, though other over-voltage thresholds can be set as desired.
  • Under-voltage functionality detects when "A" power 760 is below a desired low-voltage threshold. As illustrated, increasing diode D18 will charge capacitor C5 via the ladder formed by R16 and R30. When the charge on C5 reaches a preset level (illustrated as lOOVAC, but other levels can be set as desired), ZD8 starts to conduct through R31 and R58. Ql and Q2 will start to turn on, pulling current through R10, D 12 and R9. This turns on Q37 and Q38 applying power to A (ON). This in turn drives current through R35 and Dl 1, which turns Ql and Q2 on harder and adds hysteresis to the voltage on C5.
  • lOOVAC preset level
  • Embodiments of the "A" power voltage range detect subsystem 710 provide a number of innovative features.
  • One such feature is that some embodiments of the "A" power voltage range detect subsystem 710 provide very high efficiency. As illustrated, using high impedance components and switching off of the power delivery to the "A" power loss detect subsystem 715 as the technique for initiating and holding the voltage fault conditions is very efficient and consumes a minimum amount of power.
  • the "A" power voltage range detect subsystem 710 also provides all power to the "A" power loss detect subsystem 715, and has current limiting provided by C3. Use of a capacitor for current limiting minimizes power consumption by returning unused current to the source on each half cycle instead of wasting it as heat as in a traditional resistor limiting technique.
  • the over- voltage detect functionality uses a single capacitor value (C4) to determine the delay for detecting an over-voltage condition.
  • C4 uses a single capacitor value (C4) to determine the delay for detecting an over-voltage condition.
  • C4 uses a single capacitor value (C4) to determine the delay for detecting an over-voltage condition.
  • C5 uses a single capacitor value (C5) to determine the delay for accepting the A input voltage for start-up of the micro-ATS 700, and can be easily adjusted for various requirements.
  • some embodiments of the "A" power voltage range detect subsystem 710 provide an easily programmed thresholds for "A" under- voltage detect, "A" voltage OK, and "A" over-voltage detect. Under voltage assumes that the voltage was at one point acceptable, and that it is now lower than desired. As illustrated, a single resistor value (R35) controls the difference between the acceptable value and the low voltage shut down point. Similarly, the "A" voltage OK threshold can be programmed via a single resistor value change (R16), and the "A" over- voltage threshold can be programmed via a single resistor value change (R14).
  • FIG. 10A shows a circuit diagram of an illustrative "A" power loss detect subsystem 715a for use in some embodiments of a micro-ATS 700.
  • Embodiments of the "A" power loss detect subsystem 715 handle fundamental operation of the primary power (“A" power 760) detect and delay portions of the micro-ATS 700.
  • This circuit in the uATS is to detect the presence of AC power on the "A" side, and hold connection to that power source in the relay section later described.
  • This circuit also has the delay control that prevents returning to a power after a transfer for about 5 seconds. This prevents unnecessary transfers if the "A" side power source is intermittent. In addition, this circuit also rejects many conditions, such as outages shorter than 4 ms, momentary sags, etc., of power that would otherwise cause false transfers.
  • the core functionality is achieved by rectifying the AC power via a current limiting .22uf capacitor.
  • the rectified power is mildly filtered in C6, but the primary function of C6 is to control the amount of hold-over current present during AC outages at the zero crossing of the AC line. Otherwise, the uATS would disconnect from the "A" side at every AC crossing, 120 times a second.
  • This capacitor is also largely responsible for determining the time for a minimum outage before releasing the latch that controls the "A" side connection, and transferring to the Alternate Power Source, ("B" side).
  • the transistor pair of Q 17 (PNP) and Q18 (NPN) act as a SCR connected pair.
  • the thyristor is a four-layer, three terminal semiconductor device, with each layer consisting of alternately N-type or P-type material, for example P-N-P-N.
  • the main terminals, labeled anode and cathode, are across the full four layers, and the control terminal, called the gate, is attached to p-type material near to the cathode.
  • SCS Silicon Controlled Switch
  • This sub-circuit has a time constant of about 5 seconds and provides the delay function on start-up necessary to prevent rapid transfers if source power is intermittent.
  • a unique feature of using this dual transistor SCR emulation is that access to the collector of Q17 allows the secondary function of the "SCR” pair by allowing supplemental current to be presented to the base of Q15.
  • the timing capacitor C8 Upon successful latching of the "SCR” pair, the timing capacitor C8 is reset nearly to zero voltage by the conduction of Q15. This prepares the timing circuit for the next off-to-on cycle.
  • Another feature of this circuit is that access to the base of Q17 allows insertion of a transient suppression filter and programmed current release point determined by resistors R13, R26 and C7. This is necessary because the release point of the "SCR” pair must be below the on threshold of the optical isolator and subsequent amplifier circuit. In other words, it is important that the release point be determined by the "SCR” un-latching rather than the gain of the optical coupler and amplifier. Additional components include R19, which depletes C6 at a known rate, R21 which guarantees full discharge of C 8 on initial startup, and LED 5 (Green), an indicator for the user interface to show that A power is on and is selected for the source of delivery to the output of the uATS.
  • Another unique feature of this design is it's extremely low power consumption. Since this circuit must operate at all times when the primary power ("A" side) is being delivered to the load, minimization of power consumption was of high importance. No external power supplies are required, and power through LED 5, and the Optical Isolator LED, is determined primarily by the current limiter .22uf capacitor (C3), and the 56 K pass resistor R2. Other values of Resistance and capacitance could be selected to further reduce normal function power consumption, but these values are selected for maximum noise immunity and lowest power consumption in this application.
  • FIG. 5 demonstrates the initial electrical activity very shortly after application of power to the circuit.
  • Figure 6 represents the condition just at the conduction threshold of ZDl . At this point the base of Q18 now has voltage being applied to it relative to the emitter.
  • FIG. 7 shows the condition of current flow a few microseconds after the base current starts to flow, and current starts to flow through LED5, the optical coupler LED, R13 and the base of Q17.
  • FIG. 8 shows this state of this circuit during the discharge cycle of C8, and just before the final state of the circuit prior to normal functional operation of the uATS which is the state of delivering "A" Side power to the output.
  • FIG. 9 shows the normal operating state of the uATS while on the "A" Side power source, the primary power source. The vast majority of the uATS operating time should be in this mode.
  • FIG. 10 shows the u ATS shortly after the loss of "A" Side Power. The circuit continues to operate for a short period by extracting the remaining charge from C6.
  • FIG. 11 illustrates this transient condition.
  • FIG. 12 shows the condition after the "SCR” pair, Q17 and Q18 have released, and are no longer conducting, and hence the LED 5 and the optical isolator diode are also no longer conducting. At this point the events controlled by the timing and synchronization circuits are initiating the transfer to the "B" side of the relays in the A/B relay switching circuit.
  • the final electrical activity is the removal of residual current from the base of Q18. Also, residual charge in C6 and C8 is depleted via resistors R19 and R21, respectively. Hence, the circuit is ready for the return of "A" power when it occurs.
  • FIG. 11 shows a circuit diagram of an illustrative "B" power synchronization detection subsystem 720a for use in some embodiments of a micro-ATS 700.
  • "B" power 765 is received as BH and BN, and is full-wave rectified by bridge BR5. The current is limited by R57.
  • the output of the bridge BR5 drives the diode in opto-transistor Ul . As illustrated, the transistor on Ul will turn off when the output of the bridge BR5 is less than a preset threshold (illustrated as approximately 6 volts, though other values can be set if desired). This derives the zero-voltage crossing of "B" power 765, which can then be used for zero-voltage synchronization.
  • a preset threshold illustrated as approximately 6 volts, though other values can be set if desired
  • Embodiments of the "B" power synchronization detection subsystem 720 provide a number of innovative features.
  • One such feature is that some embodiments of the "B" power synchronization detection subsystem 720 provide loss detection as a direct function of power availability.
  • the "B" power synchronization detection subsystem 720 is powered by the source it is detecting. If the power fails, the circuit ceases to detect that source and it fails to provide the optically isolated control to the timing control subsystem 730 via the "B" power synchronization detection subsystem 720.
  • This provides a fail-safe design, in that if "A" power 760 fails, the remainder of the micro-ATS 700 circuits default to transferring to "B" power 765.
  • the circuit has a power envelope detect methodology.
  • C6 is continuously charged by the incoming power and continuously discharged by the A Loss Detect circuits and by the LED in the optical isolator U2.
  • the capacitor C6 must discharge its energy to allow the transfer.
  • One result is that it stores the energy during each half-cycle allowing easily programmed delay timing between the time that power fails and the time that initiation of transfer actually occurs.
  • Another result is that unwanted glitches are filtered where they would otherwise cause false transfers.
  • under-frequency detection is provided without additional components.
  • the capacitor C6 will fail to store enough charge from cycle to cycle, and the latch (formed by Q17 and Q18) will release and allow transfer of the load to the "B" power 765 side.
  • FIG. 12 shows a circuit diagram 1200 of an illustrative "A"/"B" synchronization integrator subsystem 725a in context of the "B" power synchronization detection subsystem 720 and the "A” power loss detect subsystem 715 for use in some embodiments of a micro-ATS 700.
  • Embodiments of the "A"/"B" synchronization integrator subsystem 725a provide synchronization of "B" to "A” transfers at zero voltage crossing times and the "A" to “B” integration function. For the sake of clarity, functionality will be described during start-up, during a transfer from “B” power 765 to "A” power 760, and during a transfer from "A” power 760 to “B” power 765.
  • Embodiments of the "A'V'B" synchronization integrator subsystem 725 provide a number of innovative features.
  • One such feature is that some embodiments of the "A'V'B" synchronization integrator subsystem 725 provide very high efficiency, for example, because of their use of high impedance components to reduce size and minimize power consumption.
  • Another such feature is that some embodiments of the "A"/"B" synchronization integrator subsystem 725 act as a multi-function circuit to minimize component count.
  • embodiments combine the functions of synchronizing the return from "B" power 765 to "A” power 760, and the timing control for the gap between disconnecting the "A” power 760 before connecting the "B” power 765 to the outputs. By combining these functions, parts count can be minimized and the overall size of the finished product can be reduced. Yet another such feature is that some embodiments of the "A"/"B" synchronization integrator subsystem 725 provide power off delay timing during transfer from "B" power 765 to "A” power 760. The transition time power off delay is accomplished via an easily programmed capacitor value (C9).
  • C9 is the integration storage capacitor that supplies threshold ramp signal to Ul, the timing comparator. Adjustment of C9 changes the delay between the A disconnect relay (RY2), and changing state of the Gatekeeper (RY3) and Neutral (RY1) relays.
  • FIG. 13 shows a circuit diagram of an illustrative timing control subsystem 730a for use in some embodiments of a micro-ATS 700.
  • Comparators in embodiments of the timing control subsystem 730 control switching functions of the relays and activation of the warning indicators and buzzer.
  • the ladder formed by R45, R46, and R47 defines voltages VI and V2.
  • Integrator (C9) is the output of the "A"/"B" synchronization integrator subsystem 725, and is high when the micro-ATS 700 is using "A" power 760 and low when the micro-ATS 700 is using "B" power 765.
  • Pin 2 of the comparator U3a is the Relay On (Low) signal and drives the emitter of Q29. Pin 2 will subsequently apply HV power to the A relay RY2 (RY On).
  • Pin 1 of the comparator U3b is the GC Shunt Drive (Low) and drives the emitter of Q14. Pin 1 will subsequently ground the other side of the A relay (RY2) when asserted. This applies the full HV power (150 Volts) across the A relay (RY2), which can assures fast operation of the A relay (RY2). Times Tl and T2 are controlled by VI and V2, and by the rising and falling slope of the Integrator (C9).
  • Embodiments of the timing control subsystem 730 participate in "A" power 760 to "B" power 765 transitions according to the following technique.
  • the Integrator (C9) will start to ramp down to VI . HV power will then be applied to RY2. Ground is already on the other side of the RY2 coil. RY2 will then disconnect from the "A" power 760 side.
  • the signal GC Shunt Drive will go high releasing the ground from RY2 and allowing current to flow through RY2 to RY1 and RY3. This connects "B” power 765 and Neutral Out to the output power 770 node. Time Tl assures that "A" power 760 is released before "B" power 765 is connected.
  • Embodiments of the timing control subsystem 730 participate in "B" power 765 to "A” power 760 transitions according to the following technique.
  • RY2 is grounded, removing power from the RY1 and RY3 relays. This connects the Neutral Out (NO) to "A" Side Neutral (AN) and Hot Out (HO) to the RY2 relay, which is open at this point.
  • HV power is removed from the RY2 relay, thus connecting "A" Side Power to the Hot Out (HO).
  • HV and “Common 2" are the outputs of the HV power supply. During normal operation, Common and Common 2 are connected together by the Darlington transistor Q22.
  • Embodiments of the timing control subsystem 730 participate in over-current control according to the following technique.
  • a predefined warning point e.g., between 12 and 13 Amps
  • an indicator is illuminated to warn the user that this is the maximum advisable limit for continuous current.
  • V3 and V4 are defined by the ladder resistors R34 and R36 and diode D7 as discussed above.
  • Output current is detected in the output current detect subsystem 740, and an analog voltage is generated there and sent to the timing control subsystem 730.
  • the slope of the "Load Current Sensed Signal" (LCSS) is a function of applied current to the attached load and time.
  • V4 is set to be equivalent to approximately a 12-Amp load on the output of the micro-ATS 700 (or another desired value).
  • pin 13 of comparator U3d will go low. This will cause a loss of conduction through ZD7 guaranteeing that Q22 will turn off when U3d pin 13 goes low, which, in turn, will energize the indicator LED (LED4).
  • R37 and D13 will cause V4 to drop slightly increasing the separation between V4 and the LCSS.
  • C25 also serves to smooth out the difference between V4 and the LCSS.
  • the micro-ATS 700 output current exceeds the pre-defined limits (e.g., substantially like a 14.5-Amp fast-blow fuse)
  • an indicator is illuminated that indicates the micro-ATS 700 disconnected the load from the source, and a buzzer sounds to warn the user that the micro-ATS 700 has disconnected the load.
  • VB virtual circuit breaker
  • V3 and V4 are defined by the ladder resistors R34 and R36 and diode D7, as discussed above.
  • the LCSS will be detected by the timing and control comparators U3C and U3D. If it represents a current of greater than 15A for 10 seconds, the "Fault (Low)" signal will be generated, latching Q23 and charging CI 2. This will protect fuses Fl and F2. If the overload is removed and switch SWl is pressed, the charge on C12 will be transferred to C13. Q23 will be turned off and power will be restored to the output. If, however, the overload has not been removed, then C15 is still charged, so that "Fault Low” will be regenerated, and Q23 will be turned back on. Repeated pressing of the reset switch (SWl) will charge up C13 from C12 and nothing will happen. This prevents an already hot Fl or F2 from getting repeated hits when SWl is pressed. Thus the virtual fuse protects the internal real fuse.
  • Embodiments of the timing control subsystem 730 provide a number of innovative features.
  • One such feature is that some embodiments of the timing control subsystem 730 provide very high efficiency.
  • Embodiments use very low-power components and high- impedance circuits to minimize power consumption.
  • Another such feature is that some embodiments of the timing control subsystem 730 provide high voltage control.
  • the connection of the low-voltage, low-power control section to the high-voltage relay- power control section is accomplished via an innovative coupling using a variation of the grounded base configuration, where the bases of Q14 and Q29 are referenced to the +15-Volt power supply.
  • the emitters of those transistors are connected to the open collector outputs of U3. Since these outputs are only current sink to the common supply and that they also cannot be exposed to voltages greater than the plus supply, these voltage amplifier transistors (Q14 and Q29) provide that voltage amplification with as few components as is possible.
  • timing control subsystem 730 provide power savings in LED illumination of the indicator LEDs (LED 4 and LED 1).
  • the improvement in efficiency is not so much the savings of power when the LEDs are illuminated, as this is the time when there is either a fault or a undesired condition, and is not the predominant operating condition of the micro- ATS 700.
  • the fact that the innovative way of powering these LEDs eliminates the need for an additional power supply, and the attendant losses associated with such an addition, is an improvement in efficiency. Both of these LEDs are powered by current through the rest of the circuitry that is utilized, regardless of the conditions of the LEDs otherwise. The current is being passed through LED 4 from the current utilized to operate the timing control subsystem 730.
  • FIG. 14 shows a circuit diagram of an illustrative "A" & "B" power switching subsystem 735 for use in some embodiments of a micro-ATS 700.
  • the "A” & “B” power switching subsystem 735 controls when the selected power source to the micro- ATS 700 is switched, either from “A” power 760 to "B" power 765 or “B” power 765 to “A” power 760.
  • Embodiments also control over-current condition switching and relay sequencing, using RY1, RY2, and RY3 to control the flow of power from the "A” power 760 and “B” power 765 inputs to the output power 770 node. Both "A" power 760 and "B” power 765 can be protected by fuses Fl and F2, respectively.
  • FIG. 15 shows a circuit diagram of an illustrative disconnect switch subsystem 745a for use in some embodiments of a micro-ATS 700.
  • This switch is used to disconnect the alternate power source (via node A2 775) from the power supply subsystem 705 when not required.
  • application of the alternate power source occurs only in one condition.
  • the relays of the "A" & "B" power switching subsystem 735 are configured with the A relay (RY2) in the disconnect (i.e., energized) position, and the Gatekeeper relay (RY3) and the Neutral relay (RY1) in the disabled (i.e., non- energized) position.
  • the base of Q4 is pulled negatively by the output of the timing control subsystem 730 and the voltage amplifier via the signal GC On 825. Current through the base of Q4 to the emitter thus clamps off current through the Gatekeeper relay (RY3) and the Neutral relay (RY1).
  • RY Power On 820 is active to energize the via the emitter and base of Q4 to GC On 825, which is pulled to round.
  • the A relay (RY2) is on (i.e., energized), and the Gatekeeper relay (RY3) and the Neutral relay (RY1) are held off. It is worth noting that, in this condition, there is no source of power available to the output power 770 node.
  • the reset switch SW1 in the Timing Control Subsystem can be connected to have additional functionality.
  • two switches are used, one of which is the existing "reset” button which re-initiates power connection to the output after a fault condition, and one is an additional switch and associated latching circuit, used to clear the latching circuit if it is in the "off mode and turn it back on.
  • the "off switch will essentially "trip" the Virtual Circuit Breaker, by initiating the clear state (ie "off) condition without sounding the piezoelectric buzzer. This is done by connecting the off switch controlled latch to the control node between the present fault detect circuit and the relay control circuit.
  • the mode of connection is commonly know as "wired OR".
  • Embodiments of the disconnect switch subsystem 745 provide a number of innovative features.
  • One such feature is that some embodiments of the disconnect switch subsystem 745 provide very high efficiency. As illustrated, the use of three relays in series, rated at 48 volts each, can allow the application of directly rectified AC mains voltage to the relays. This eliminates additional power conversion circuitry, thus reducing parts counts as well as increasing efficiency.
  • Another such feature is that some embodiments of the disconnect switch subsystem 745 use current limiting capacitor (CI 6) when only the A relay (RY2) is activated continuously, which allows use of the otherwise unused "normally open” contact of that relay, and further reduces parts count.
  • CI 6 current limiting capacitor
  • disconnect switch subsystem 745 provide relay sequencing when all three relays release simultaneously to prevent arching.
  • the coupling of the fly-back suppression diode D48 on RY2 to the return path rather than directly across the relay provides a slight contact timing delay between the A relay (RY2) losing power and the Gatekeeper relay (RY3) and Neutral relay (RYl). This is accomplished by the anode of D48 being connected to the common rail instead of the more traditional connection to the relay coil.
  • the A relay (RY2) has a higher impedance for the fly-back current to sink into, as the sinking current is also going through RYl and RY3.
  • FIG. 16 shows a circuit diagram of an illustrative output current detect subsystem 740a for use in some embodiments of a micro-ATS 700.
  • Embodiments of the output current detect subsystem 740a detect and measure the presence and various characteristics of the output current from the automatic transfer switch. This also circuit tends to mimic characteristics of a fuse, but only slightly below the thresholds of 15-Amp Fast Blow physical fuses that may be used in the "A" power 760 and "B" power 765 inputs to the micro-ATS 700. This allows the micro-ATS 700 to protect itself without blowing actual physical fuses, which must be replaced.
  • the neutral out of the micro-ATS 700 has a current sense transformer on it.
  • This transformer has a diode bridge, formed by D28, D29, D30, and D31, which full- wave rectifies the "Load Current Sensed Signal" (LCSS, described above with reference to the timing control subsystem 730).
  • C22 and R48 filter out the higher frequencies of the AC current and provide proper impedance loading of the current transformer.
  • Rl 1 and Thermistor RT1 provide thermal compensation so the micro-ATS 700 can remain accurate over a wide range of temperatures.
  • C14, R50, D32, D3, C15, R51, and R75 can be configured and selected to effectively emulate the time VS current opening threshold of a 14.5-Amp fast-blow fuse, with the time part of the curve advanced to open about 33% faster than the equivalent 15-Amp fast-blow fuse.
  • Embodiments of the output current detect subsystem 740 provide a number of innovative features.
  • One such feature is that some embodiments of the output current detect subsystem 740 employ a combination of capacitors and resistors that result in an analog representation of the characteristics of a fast blow fuse.
  • the principal characteristics of the timing of a fuse are as follows: it can carry significant over-current for a short period of time; and, after some time, the fuse will "blow" at or near the rated current of the fuse. This is primarily controlled by the thermal characteristics of the fuse material itself, and how much mass is being heated to the melting point by the applied current.
  • a pseudo two-pole filter is formed by these components to emulate the desired characteristics.
  • the circuit reduces the parts count to the minimum by having the two halves of the two poles interact with each other by allowing current from the first pole to charge the second pole without the reverse.
  • the second pole is discharged by a controlled discharge path through R32.
  • This single point of discharge can thus be used to bias both poles of the filter and alter the overall curve of the response time of this circuit without significantly changing the shape of the curve.
  • This is useful for adjusting the "rating of the fuse" by a simple one component change, specifically R32.
  • the programmed threshold of maximum current rating of the electronic, or virtual, fuse circuit breaker
  • RT1 and Rl 1 are thermister that has a negative temperature coefficient. As temperature goes up, the resistance goes down, in a predictable fashion. By placing these two components in series across the current sense transformer, the load impedance presented to that transformer is affected by temperature. The selection of values for the Thermistor RT1 and Rl 1 result in the overall performance of the virtual circuit breaker effectively mimicking its electro-mechanical equivalent.
  • FIG. 17 shows a circuit diagram of an illustrative piezoelectric device driver subsystem 750a for use in some embodiments of a micro-ATS 700.
  • Embodiments of the piezoelectric device driver subsystem 750a include an innovative implementation of a power driver for driving a piezoelectric buzzer (or similar device).
  • "NOT FAULT” goes low, it provides a ground path for the oscillator formed by Q25, Q27, R40, R42, R44, R39, CIO, and CI 1.
  • Q25 or Q27 When power is applied, either Q25 or Q27 will turn on. If Q25 turns on, the voltage on both sides of CIO will drop, thereby providing a low to the base of Q27 and turning Q27 off. Then CIO will charge up via R44. When the voltage at the base of Q27 reaches approximately 0.6 volts, Q27 will turn on. Both sides of CI 1 will drop. Q25 will turn off, and so on, back and forth.
  • Q21 and Q26 are emitter followers. As the collector of Q25 charges up when Q25 is off, Q23 emitter will follow its base driving the piezoelectric buzzer. When Q25 turns on it will sink connect from the piezo through D2, and similarly for Q26, D4, and Q27.
  • Embodiments of the piezoelectric device driver subsystem 750 such as the
  • embodiments illustrated and described with reference to FIG. 17, provide a number of innovative features.
  • One such feature is that the circuit provides high efficiency with a low parts count. As illustrated, high efficiency is achieved by combining the oscillator and amplifier necessary to drive the piezo into a power oscillator, and the combination also tends to minimize parts count. In this configuration, nearly all the current expended in the circuit for both the oscillator function and the power amplifier function is applied to the piezo crystal for conversion to acoustic energy.
  • the various circuits and other embodiments described above form novel embodiments of micro-ATSs 700 and provide a number of features.
  • One such feature is ultra-low power consumption. Use of a transformer-less power supply reduces overall loss in converting from 120 VAC mains voltage to DC voltages required for internal circuits.
  • Another such feature is that virtually no power is consumed on the non-connected side (e.g., on the "B" power 765 side when the load is connected to "A” power 760). Yet another such feature is that the use of optical isolation in the power supply virtually eliminates cross currents between the "A” power 760 and "B” power 765 inputs. Still another such feature is that arc suppression is provided at the disconnection of the "B" power 765 side on the contacts of the affected relays by timing the break of the relay contacts at the zero crossing of the current presented to those contacts.
  • Additional features are realized by selections of certain circuit components and/or topologies.
  • One such feature is that ultra low power consumption can be achieved by utilizing a capacitor current limit on the AC input.
  • the selected capacitor (0.22uF) limits 60 Hz to about 10 milliamps into the 48-volt load. At 50 Hz, the limit is 8 milliamps, both within operating range of the connected relays.
  • Another such feature is that use of the same path of current in the "A" power loss detect subsystem 715 for illuminating the "on A" indicator, activating the optical isolator link to the
  • Another such feature is that power delivery from the "A" power 760 side to the Load, the predominant path of power for the majority of use with regards to time, is via the Normally Closed position of the routing relays. This eliminates the current necessary to activate the relays.
  • Another such feature is that conversion of the AC mains voltage directly to DC without the use of a transformer, or power conversion, for the purpose of minimizing power consumption, is made possible by the arrangement of the relays when activated. Three relays are connected in series when activated to transfer the load to the "B" power 765 side. This action allows the use of high voltage (AC Mains Voltage directly rectified to 150 VDC). This design methodology allows the transformer-less power supply to be implemented.
  • Another such feature comes from use of the Normally Open contact of the A disconnect Relay (RY2) as an AC Mains diversion path for and alternate power source during a fault condition. In this condition, only the A relay is activated. The alternate path described allows a second rectifier and a capacitor current limit on the AC Mains, so the current rating of the coil of the single activated relay (RY2) is within operating range of 8 - 10 milliamps.
  • Another such feature is that use of total circuit pass current on the low voltage timing control subsystem 730 to energize the indicator LED (LED 4) that indicates approaching the over current condition, and shunting the LED to the off condition when not in use, eliminates the need for an auxiliary power supply for this device.
  • SCR simile requires very little current to cause to latch into a conduction state. This latched state will continue until passing current through the pair ceases. When it ceases, the latch will disconnect and not allow current to flow until re-initiated.
  • the otherwise inaccessible second PN junction of the SCR simile is accessible, thus allowing the action at this point to also affect Q15 at a certain point in the sequence to be beneficial.
  • the Q15 is the timer reset for delaying the return to "A" power 760 from an "A" loss state. This must be reset every time the SCR simile goes into conduction.
  • the avalanche activation of the SCR simile at that time can be utilized to force the reset of the timing capacitor C8 via Ql 8.
  • An optically coupled synchronization pulse and level from the "B" power 765 source is used for two functions: to synchronize transfer from the "B" power 765 to the "A” power 760 at the zero-crossing of the "B” power 765; and to provide a source of signal that indicates the loss of the "B” power 765 and thus force the connection to the "A” power 760, regardless of the state of other circuits that normally would influence transfers to the "B” power 765 side.
  • the optically coupled sync signal holds off the signal from the optical coupler that indicates that "A" power 760 is ready to be transferred to until the sync pulse appears (via Q19, Q20, and the Ul opto-coupler transistor).
  • the release of the hold on the signal from Q33 allows the circuit to initiate the integrate function for the comparator, U3 to utilize to complete the transfer.
  • the release of the hold function then latches the sync pulses out so no additional pulses can be present via bias on the base of Q19.
  • Another such feature arises from the unique piezoelectric device driver subsystem 750.
  • This arrangement of Q21, Q26, Q25, and Q27 forms a transistor pair oscillator.
  • Q25 and Q27 form the basis of an astable oscillator with R39, R40, R42, R44, CIO, and CI 1 forming the components that determine oscillation.
  • the innovative addition of D2, D4, Q21, and Q26 turn the oscillator into a power oscillator capable of driving piezo components bilaterally (e.g., one side of the Piezo is at +15, while the other is at Common; then it switches, and the opposite is true). This maximizes the output to the Piezo for a given power supply voltage, at minimum parts count and minimum power dissipation.
  • This same power oscillator can be used for other applications such as driving a miniaturized switching power supply, or for signal source in a small tester. Numerous applications exist for a miniature, low power, very low cost power oscillator. Various changes, substitutions, and alterations to the techniques described herein can be made without departing from the technology of the teachings as defined by the appended claims. Moreover, the scope of the disclosure and claims is not limited to the particular aspects of the process, machine, manufacture, composition of matter, means, methods, and actions described above. Processes, machines, manufacture, compositions of matter, means, methods, or actions, presently existing or later to be developed, that perform substantially the same function or achieve substantially the same result as the corresponding aspects described herein may be utilized.
  • a method is described to implement extremely efficient data center power distribution using highly parallel auto-switched power at ratios of 1 ⁇ TM to 1 EDP device or 1 ⁇ TM to a low integer number of EDP devices) as described earlier. This method is also much more reliable than other methods of using ATS devices for power distribution.
  • ATS technology used to implement the highly parallel auto-switched data center power distribution has a sufficiently small form factor, it is possible to implement the methodology without using any rack space that could otherwise be used for mounting EDP equipment by placing the ATS units in various locations as follows: 1. Mounted on and fitting within 1U or near the EDP equipment being powered
  • Each sub-branch output of the ZPDU would normally be auto-switched in this case.
  • the described example ⁇ TM is sufficiently small to be placed in these locations.
  • Embodiments provide techniques for delivering auto-switched power using multiple automatic transfer switches (ATSs) to end-user equipment mounted in data center racks.
  • ATSs automatic transfer switches
  • the design and construction is facilitated by the optional use of hydra cords to create a power distribution system in the rack that allows the dimensions of the rack to be optimized to maximize the efficiency of the use of data center floor space. Accordingly, airflow can be optimized in the equipment rack by minimizing the number of power cables needed and their routing in the rack.
  • Certain embodiments further include other power distribution technologies that unite auto- switching capabilities with power phase load balancing.
  • Typical modern data centers can have power distribution networks that include thousands of branch distribution circuits.
  • Precise loading and/or demands of the various electronic data processing (EDP) equipment is often unknown to the data center personnel. Accordingly, changes to the loading of these circuits can cause electrical failures, for example, when branch circuit breakers are tripped by personnel plugging in a load that exceeds the capacity of the circuit. Further, it may be important to maintain loading of each branch circuit at or below about 75% of its capacity to account for "inrush loads" that can occur during a cold start, which may be the highest load scenario and can otherwise trip the branch circuit breakers when it happens.
  • EDP equipment in a data center can be undesirable for a number of reasons.
  • One reason is that these EDP devices can support mission-critical or life safety goals, for which even a short interruption in functionality can be catastrophic.
  • Another reason is that inter-dependencies of modern IT infrastructures and their applications are quite complex and may not always be fully appreciated.
  • a single EDP device may provide an underlying service that nobody realized was associated with that device, and a power loss can cause larger business functions that depend on the affected service to be adversely and expensively impacted.
  • Yet another reason that downtimes or failures of EDP equipment in a data center can be undesirable is that restarting an IT infrastructure and the applications that run on it successfully, from either a cold-start or intermediate state, can be very site-specific and unpredictable.
  • Restoring proper functioning of the functionality may involve extensive human intervention, including manual reboots or service stop/starts. Further, EDP equipment downtimes can be significant, can be difficult to diagnose and fix, and can cause corruption of service configurations or data in some instances.
  • EDP equipment may be particularly vulnerable to electrical failures at the branch circuit.
  • many models of EDP equipment have only one power supply, and therefore one power cord.
  • a failure of the input power to the power cord interrupts power to the power supply of the equipment.
  • One technique for improving power distribution reliability for EDP equipment is to offer configurations having multiple power supplies.
  • interruption of the input power can cause failure (or downtime) of the equipment when the device can only be plugged into one power source at a time.
  • Implementations with multiple power supplies can have additional limitations.
  • One limitation is that additional power supplies raise the cost of the device and can be associated with minimum additional amounts of power loss associated with running those additional supplies.
  • Another limitations is that power supplies, especially the types typically used in EDP equipment, function most efficiently in a relatively narrow band of their output load rating (e.g., 70 - 85%). If an EDP device is offered in single and dual power supply configurations, the product manager may not wish to stock two different models of power supply with different load range optimizations, since that can represent additional expense to the manufacturer. This often means that the product manager will choose to stock the single power supply optimized configuration, which is optimized for perhaps a 70% load. When this power supply is specified in a dual power supply configuration, it will then be running at around 50% of its optimal loading, thereby appreciably lowering its efficiency.
  • a different approach to addressing the issue of power distribution reliability is to implement EDP equipment with auto-switching power plugstrips.
  • those plugstrips are bulky and expensive, and they are usually mounted horizontally in data equipment racks, which can take up valuable rack space (they tend to take even more rack space with two input power plugs connected to two different power sources).
  • enabling auto-switched A-B power delivery to EDP equipment even having only a single power supply can appreciably increase the uptime of such equipment (though not typically to the level of dual power supply configurations). In many cases, this gain in reliability can be enough to meet the service level availability goals for the applications that the EDP equipment supports.
  • single power cord devices can realize significant gains in uptime reliability when connected to two independent power sources. These sources can be specified to gain the optimum cost benefit to reliability gain ratio.
  • implementations include data centers with uniform, redundant A-B power distribution, such as that supplied by the Zonit Power Distribution System.
  • One illustrative configuration includes two independent uninterrupted power sources (UPSs). This is often considered the highest reliability scenario, and is a common data center configuration.
  • UPSs independent uninterrupted power sources
  • Another illustrative data center configuration includes one line power source and one UPS. This type of configuration may be used as a cost savings measure when the total power usage in the data center exceeds the UPS capacity or to achieve higher power usage efficiency, since UPS units have a loss factor for the power they supply.
  • the EDP equipment in the data center can be connected to run off of line power and only selected mission critical equipment is connected to the UPS as a backup power source. The majority of equipment is just line powered with a power conditioner module to stop input surges. For example, if the site steps down industrial power delivery from 480V to a more standard 208V, then a transformer is already in-line and any further power conditioning may not be necessary.
  • UPS units both condition the power that goes through them and, if power is lost, use batteries to deliver backup power. Their capacity is therefore twofold: the sustained amperage they can condition; and how many minutes of power they can deliver at a set load percentage, usually 100% for rating purposes.
  • the amperage capacity is a function of the design of the UPS.
  • the battery capacity is a combination of the amperage capacity and the number of batteries that are connected to the UPS. The battery capacity can be changed using external battery packs so the battery runtime of the UPS is increased to the desired target level.
  • Still another illustrative data center configuration includes two independent line power sources from two different power grids. In this configuration, only line power sources are used, but they are delivered on different distribution legs. This can reduce risk by insuring that two power branch distribution circuits must trip their breakers before power is lost to the equipment.
  • ATSs automatic transfer switches
  • Some embodiments described herein provide additional features. Some embodiments provide space-efficient parallel ATS deployments in data center rack configurations. These embodiments can provide alternate techniques to maximize the efficiency of usage of data center floor space and allow the deployment of the maximum number of equipment racks in a data center environment. Other embodiments provide
  • ATSs and/or power cords to minimize power cable routing and/or airflow issues in data center equipment racks (e.g., "2 -post” and/or cabinet (“4-post") equipment racks).
  • data center equipment racks e.g., "2 -post” and/or cabinet (“4-post") equipment racks.
  • Still other embodiments incorporate locking power cord technologies at one or both ends of power cords for more secure power delivery, for example in data centers located in seismically active geographies (e.g., California).
  • micro-ATSs small ATSs
  • control logic to construct a large capacity, fast, efficient, and relatively low cost ATS.
  • micro-ATSs incorporate functionality described in PCT Application No. PCT/US2008/057140, U.S. Provisional Patent Application No. 60/897,842, and U.S. Patent Application No.
  • micro-ATSs 12/569,733, all of which are fully incorporated herein by reference for all purposes. Embodiments of the micro-ATSs will be discussed below, followed by embodiments for deploying the micro-ATSs to provide redundant power distribution.
  • the micro-ATS (e.g., the Zonit ⁇ 8TM) is very small (e.g., 4.25-inches x 1.6-inches x 1-inch, or less than 10 cubic inches) and very efficient (e.g., less than 0.2 volts at maximum load loss).
  • Certain implementations use no rack space, as they are self-mounted on the back of each EDP device, incorporated in the structure of the rack outside the volume of the rack used to mount EDP equipment, incorporated in rack-mounted plugstrips, or incorporated in an in-rack or near-rack Power Distribution Unit (i.e., any of which being possible due to the small form-factor of the micro-ATS).
  • the micro-ATS is small enough to be integrated directly into the EDP equipment itself.
  • micro-ATSs are described herein, including their various components.
  • the micro-ATS embodiments are described as switching between two separate power sources, "A" and "B."
  • the A and B power sources are single-phase sources.
  • polyphase power sources are connected. Where polyphase power sources are connected, polyphase embodiments of micro-ATSs are used. Substantially the same components (e.g., circuits) described herein with reference to the single-phase implementations are applicable to the polyphase implementations.
  • polyphase embodiments can be implemented as multiple single-phase micro-ATS units acting in parallel, with additional functionality provided for synchronizing certain of the control circuits so that they act together across the multiple ATS units to handle switching and return from one polyphase source to the other polyphase source and back.
  • Various embodiments of polyphase micro-ATSs can also have different conditions under which to switch power sources. For example, given three phase power with X, Y, and Z "hot" leads, a fault on any of three might be considered reason to switch from the A to the B polyphase source. To return to the A polyphase source, it may be desirable to ensure first that all three hot leads are present, stable, and of sufficient power quality on the A source.
  • FIG. 7 a system diagram of an illustrative micro-ATS 700 is shown, according to various embodiments. As illustrated, the micro-ATS 700 is connected to an "A" power source 760 and a "B" power source 765, and uses its various components to provide output power 770 to one or more devices or distribution topologies (e.g., to one or more EDP devices in a branch circuit of a data center).
  • the micro-ATS 700 is connected to an "A" power source 760 and a "B" power source 765, and uses its various components to provide output power 770 to one or more devices or distribution topologies (e.g., to one or more EDP devices in a branch circuit of a data center).
  • the micro-ATS 700 includes a power supply subsystem 705, an "A" power voltage range detect subsystem 710, an "A” power loss detect subsystem 715, a “B” power synchronization detection subsystem 720, an "A"/"B” synchronization integrator subsystem 725, a timing control subsystem 730, an "A” & “B” power switching subsystem 735, an output current detect subsystem 740, a disconnect switch subsystem 745, and a piezoelectric device driver subsystem 750.
  • Embodiments of the power supply subsystem 705 include innovative ways of powering control circuitry of the micro-ATS 700.
  • Embodiments of the "A" power voltage range detect subsystem 710 determine if the power being supplied to the micro-ATS 700 is in a desired (e.g., predetermined) voltage range.
  • Embodiments of the "A" power loss detect subsystem 715 determine when the "A" power being supplied to the micro-ATS 700 has been lost using desired discrimination characteristics.
  • Embodiments of the "B" power synchronization detection subsystem 720 measure the timing of the alternating current waveform of the "B" power.
  • Embodiments of the "A"/"B" synchronization integrator subsystem 725 provide synchronization of "B" to “A” transfers at zero voltage crossing times and "A" to “B” integration functionality.
  • Embodiments of the timing control subsystem 730 control when the selected power source to the micro- ATS 700 is switched, either the “A” source to the “B” source or from the “B” source to the “A” source, and can handle over-current condition switching and relay sequencing.
  • Embodiments of the "A” & “B” power switching subsystem 735 control actual switching between the "A” and “B” power sources in either direction to change which supply is acting as the input power source to the micro-ATS 700.
  • Embodiments of the output current detect subsystem 740 detect and measure presence and various characteristics of output current from the micro-ATS 700, and can, in some
  • Embodiments of the disconnect switch subsystem 745 disconnect a secondary power source from the power supply when it is not in use.
  • Embodiments of the piezoelectric device driver subsystem 750 implement innovative techniques for driving piezoelectric or other devices.
  • each component is described below as performing particular functionality in the context of the micro-ATS 700. It will be appreciated that other configurations are possible in which similar or identical functionality can be implemented using other components, combinations of components, etc. Further, in some cases, values are given for components such as resistors and capacitors, etc. and ranges are given for current, voltage, and/or other power characteristics. These values and ranges are intended to add clarity to illustrative examples and should not be construed as limiting the scope of embodiments.
  • FIG. 8A shows a circuit diagram 800 of an illustrative power supply subsystem 705a in context of an illustrative "A" & “B" power switching subsystem 735 for use in some embodiments of a micro-ATS 700.
  • the micro-ATS 700 is connected to an "A" power source 760 and a "B" power source 765.
  • the power supply subsystem 705a performs a number of functions, including power conditioning (e.g., current limiting and power clean-up).
  • the source power for the power supply subsystem 705 a is acquired from the center taps of RY3 H ("Hot") and RY1 N ("Neutral”).
  • the source of power for the ATS Power Supply is from the "A" side when the output of the micro-ATS 700 is on the “A” Side, and on the "B” Side when transferred to the "B” side.
  • Components of the "A” & “B” power switching subsystem 735 serve as the automatic transfer switch for the micro-ATS 700 power supply subsystem 705a.
  • the current available from output power is limited by R3.
  • ZD 10 limits the full wave rectified output of the bridge BR4 to 150V peak. ZD 10 guarantees that CI does not exceed its rated voltage. CI stores enough charge to allow HV ("High Voltage") operations of the relays during transitions between the "A" power source 760 and the "B” power source 765.
  • Bridge BR6 normally blocks A2, as will be discussed more fully below.
  • GC On When a fault occurs, GC On will be pulled almost down to Common, and positive power will be available at U5 LED via ZD11. This turns on the U5 transistor.
  • the U5 transistor and Q3 form a Darlington pair that shorts the bridge BR6, allowing A2 to drive HV through diode D9.
  • A2 also drives LV ("Low Voltage") through diode D10.
  • C20 provides filtering and storage for LV.
  • the 15-volt power supply is normally supplied by HV through R63, R64 and R65. These three resistors drop the voltage and limit the current for Zener diode ZD2. ZD2 regulates the voltage for the comparator and other electronics on the control board. C17 further filters out the 15-volt signal.
  • A2 provides power for the 15V supply through LV. HV is pulled down to 45V in this condition due to current limiting capacitor C16 mentioned earlier.
  • the power supply subsystem 705 acts as a transformer-less, very high-efficiency power supply.
  • the circuit produces low and high- voltage DC power from multiple AC inputs that is suitable for powering both low-voltage control circuitry and higher power relays. It does so very efficiently and with a minimum of expensive analog parts.
  • some embodiments of the power supply subsystem 705 provide capacitor current limiting for very low-power usage. As illustrated above, power consumption of the power supply subsystem 705 is limited by use of a capacitor. This can efficiently limit the power supply capacity to a desired value, thereby providing maximum efficiency and low power consumption.
  • embodiments of the power supply subsystem 705 provide optical isolation to reduce or even eliminate cross currents
  • embodiments of the control and synchronization subsystems e.g., the "B" power synchronization detection subsystem 720, the "A"/"B" synchronization integrator subsystem 725, the timing control subsystem 730, etc.
  • some embodiments of the power supply subsystem 705 provide power control relays (e.g., illustrated as RY1, RY2 and RY3) to direct the source power to the output of the micro-ATS 700 as well as provide the internal source selection (transfer switch function) for powering the micro-ATS 700 power supply subsystem 705.
  • power control relays e.g., illustrated as RY1, RY2 and RY3
  • FIG. 9 shows a circuit diagram of an illustrative "A" power voltage range detect subsystem 710a for use in some embodiments of a micro-ATS 700.
  • Embodiments of the "A" power voltage range detect subsystem 710a receive "A" power 760 nodes AH and AN.
  • the "A” power 760 is full wave rectified by bridge BR2.
  • C3 is used to limit the current available in this and the "A” power loss detect subsystem 715, as will be described below.
  • the "A" power voltage range detect subsystem 710a can generate the A (ON), A (COM), and CQ18 signals to the "A" power loss detect subsystem 715.
  • the illustrated "A" power voltage range detect subsystem 710a includes over- voltage detection and under-voltage detection functionality.
  • D18 half-wave rectifies the "A" power 760 and drives a ladder comprised of R14 and R27, which charges C4.
  • ZD6 and Q36 will begin to conduct drawing current through resistors R7, R74, R6, R5, and Rl .
  • This will turn on Q34, which will pull up the voltage on C4 through R17.
  • This will latch Q35, Q36, and Q34 in an on state.
  • This also will pull current through R8, thereby turning on the over- voltage indicator, LED2.
  • C2 will be charged and ZD5 will conduct.
  • Q32 and Q31 will conduct, turning off Q37 and Q38. This turns off A (ON). In the illustrated embodiment, this will tend to occur at around 135 VAC at AH, though other over-voltage thresholds can be set as desired.
  • Under-voltage functionality detects when "A" power 760 is below a desired low-voltage threshold. As illustrated, increasing diode D18 will charge capacitor C5 via the ladder formed by R16 and R30. When the charge on C5 reaches a preset level (illustrated as lOOVAC, but other levels can be set as desired), ZD8 starts to conduct through R31 and R58. Ql and Q2 will start to turn on, pulling current through R10, D 12 and R9. This turns on Q37 and Q38 applying power to A (ON). This in turn drives current through R35 and Dl 1, which turns Ql and Q2 on harder and adds hysteresis to the voltage on C5.
  • lOOVAC preset level
  • Embodiments of the "A" power voltage range detect subsystem 710 provide a number of innovative features.
  • One such feature is that some embodiments of the "A" power voltage range detect subsystem 710 provide very high efficiency. As illustrated, using high impedance components and switching off of the power delivery to the "A" power loss detect subsystem 715 as the technique for initiating and holding the voltage fault conditions is very efficient and consumes a minimum amount of power.
  • the "A" power voltage range detect subsystem 710 also provides all power to the "A" power loss detect subsystem 715, and has current limiting provided by C3. Use of a capacitor for current limiting minimizes power consumption by returning unused current to the source on each half cycle instead of wasting it as heat as in a traditional resistor limiting technique.
  • the over- voltage detect functionality uses a single capacitor value (C4) to determine the delay for detecting an over-voltage condition.
  • C4 uses a single capacitor value (C4) to determine the delay for detecting an over-voltage condition.
  • C4 uses a single capacitor value (C4) to determine the delay for detecting an over-voltage condition.
  • C5 uses a single capacitor value (C5) to determine the delay for accepting the A input voltage for start-up of the micro-ATS 700, and can be easily adjusted for various requirements.
  • some embodiments of the "A" power voltage range detect subsystem 710 provide an easily programmed thresholds for "A" under- voltage detect, "A" voltage OK, and "A" over-voltage detect. Under voltage assumes that the voltage was at one point acceptable, and that it is now lower than desired. As illustrated, a single resistor value (R35) controls the difference between the acceptable value and the low voltage shut down point. Similarly, the "A" voltage OK threshold can be programmed via a single resistor value change (R16), and the "A" over- voltage threshold can be programmed via a single resistor value change (R14).
  • FIG. 10 shows a circuit diagram of an illustrative "A" power loss detect subsystem 715a for use in some embodiments of a micro-ATS 700.
  • Embodiments of the "A" power loss detect subsystem 715 handle fundamental operation of the primary power (“A" power 760) detect and delay portions of the micro-ATS 700.
  • This circuit in the uATS is to detect the presence of AC power on the "A" side, and hold connection to that power source in the relay section later described.
  • This circuit also has the delay control that prevents returning to a power after a transfer for about 5 seconds. This prevents unnecessary transfers if the "A" side power source is intermittent. In addition, this circuit also rejects many conditions, such as outages shorter than 4 ms, momentary sags, etc., of power that would otherwise cause false transfers.
  • the core functionality is achieved by rectifying the AC power via a current limiting .22uf capacitor.
  • the rectified power is mildly filtered in C6, but the primary function of C6 is to control the amount of hold-over current present during AC outages at the zero crossing of the AC line. Otherwise, the uATS would disconnect from the "A" side at every AC crossing, 120 times a second.
  • This capacitor is also largely responsible for determining the time for a minimum outage before releasing the latch that controls the "A" side connection, and transferring to the Alternate Power Source, ("B" side).
  • the transistor pair of Q 17 (PNP) and Q18 (NPN) act as a SCR connected pair.
  • the thyristor is a four-layer, three terminal semiconductor device, with each layer consisting of alternately N-type or P-type material, for example P-N-P-N.
  • the main terminals, labeled anode and cathode, are across the full four layers, and the control terminal, called the gate, is attached to p-type material near to the cathode.
  • SCS Silicon Controlled Switch
  • This sub-circuit has a time constant of about 5 seconds and provides the delay function on start-up necessary to prevent rapid transfers if source power is intermittent.
  • a unique feature of using this dual transistor SCR emulation is that access to the collector of Q17 allows the secondary function of the "SCR” pair by allowing supplemental current to be presented to the base of Q15.
  • the timing capacitor C8 Upon successful latching of the "SCR” pair, the timing capacitor C8 is reset nearly to zero voltage by the conduction of Q15. This prepares the timing circuit for the next off-to-on cycle.
  • Another feature of this circuit is that access to the base of Q17 allows insertion of a transient suppression filter and programmed current release point determined by resistors R13, R26 and C7. This is necessary because the release point of the "SCR” pair must be below the on threshold of the optical isolator and subsequent amplifier circuit. In other words, it is important that the release point be determined by the "SCR” un-latching rather than the gain of the optical coupler and amplifier.
  • Additional components include R19, which depletes C6 at a known rate, R21 which guarantees full discharge of C 8 on initial startup, and LED 5 (Green), an indicator for the user interface to show that A power is on and is selected for the source of delivery to the output of the uATS.
  • Another unique feature of this design is it's extremely low power consumption. Since this circuit must operate at all times when the primary power ("A" side) is being delivered to the load, minimization of power consumption was of high importance. No external power supplies are required, and power through LED 5, and the Optical Isolator LED, is determined primarily by the current limiter .22uf capacitor (C3), and the 56 K pass resistor R2. Other values of Resistance and capacitance could be selected to further reduce normal function power consumption, but these values are selected for maximum noise immunity and lowest power consumption in this application.
  • FIG. 5 demonstrates the initial electrical activity very shortly after application of power to the circuit.
  • Figure 6 represents the condition just at the conduction threshold of ZD1. At this point the base of Q18 now has voltage being applied to it relative to the emitter.
  • FIG. 7 shows the condition of current flow a few microseconds after the base current starts to flow, and current starts to flow through LED5, the optical coupler LED, R13 and the base of Q17.
  • FIG. 9 shows the normal operating state of the uATS while on the "A" Side power source, the primary power source. The vast majority of the uATS operating time should be in this mode.
  • FIG. 10 shows the u ATS shortly after the loss of "A" Side Power. The circuit continues to operate for a short period by extracting the remaining charge from C6.
  • FIG. 11 illustrates this transient condition.
  • FIG. 12 shows the condition after the "SCR” pair, Q17 and Q18 have released, and are no longer conducting, and hence the LED 5 and the optical isolator diode are also no longer conducting. At this point the events controlled by the timing and synchronization circuits are initiating the transfer to the "B" side of the relays in the A/B relay switching circuit.
  • the final electrical activity is the removal of residual current from the base of Q18. Also, residual charge in C6 and C8 is depleted via resistors R19 and R21, respectively. Hence, the circuit is ready for the return of "A" power when it occurs.
  • FIG. 11 shows a circuit diagram of an illustrative "B" power synchronization detection subsystem 720a for use in some embodiments of a micro-ATS 700.
  • "B" power 765 is received as BH and BN, and is full-wave rectified by bridge BR5. The current is limited by R57.
  • the output of the bridge BR5 drives the diode in opto-transistor Ul . As illustrated, the transistor on Ul will turn off when the output of the bridge BR5 is less than a preset threshold (illustrated as approximately 6 volts, though other values can be set if desired). This derives the zero-voltage crossing of "B" power 765, which can then be used for zero-voltage synchronization.
  • a preset threshold illustrated as approximately 6 volts, though other values can be set if desired
  • Embodiments of the "B" power synchronization detection subsystem 720 provide a number of innovative features.
  • One such feature is that some embodiments of the "B" power synchronization detection subsystem 720 provide loss detection as a direct function of power availability.
  • the "B" power synchronization detection subsystem 720 is powered by the source it is detecting. If the power fails, the circuit ceases to detect that source and it fails to provide the optically isolated control to the timing control subsystem 730 via the "B" power synchronization detection subsystem 720.
  • This provides a fail-safe design, in that if "A" power 760 fails, the remainder of the micro-ATS 700 circuits default to transferring to "B" power 765.
  • the circuit has a power envelope detect methodology.
  • C6 is continuously charged by the incoming power and continuously discharged by the A Loss Detect circuits and by the LED in the optical isolator U2.
  • the capacitor C6 must discharge its energy to allow the transfer.
  • One result is that it stores the energy during each half-cycle allowing easily programmed delay timing between the time that power fails and the time that initiation of transfer actually occurs.
  • Another result is that unwanted glitches are filtered where they would otherwise cause false transfers.
  • under-frequency detection is provided without additional components.
  • the capacitor C6 will fail to store enough charge from cycle to cycle, and the latch (formed by Q17 and Q18) will release and allow transfer of the load to the "B" power 765 side.
  • FIG. 12 shows a circuit diagram 1200 of an illustrative "A"/"B" synchronization integrator subsystem 725a in context of the "B" power synchronization detection subsystem 720 and the "A" power loss detect subsystem 715 for use in some
  • Embodiments of a micro-ATS 700 Embodiments of the "A"/"B" synchronization integrator subsystem 725a provide synchronization of "B" to “A” transfers at zero voltage crossing times and the "A" to “B” integration function. For the sake of clarity, functionality will be described during start-up, during a transfer from “B” power 765 to "A” power 760, and during a transfer from "A” power 760 to “B” power 765.
  • Embodiments of the "A'V'B" synchronization integrator subsystem 725 provide a number of innovative features.
  • One such feature is that some embodiments of the "A'V'B" synchronization integrator subsystem 725 provide very high efficiency, for example, because of their use of high impedance components to reduce size and minimize power consumption.
  • Another such feature is that some embodiments of the "A'V'B" synchronization integrator subsystem 725 act as a multi-function circuit to minimize component count.
  • embodiments combine the functions of synchronizing the return from "B" power 765 to "A” power 760, and the timing control for the gap between disconnecting the "A” power 760 before connecting the "B” power 765 to the outputs. By combining these functions, parts count can be minimized and the overall size of the finished product can be reduced. Yet another such feature is that some embodiments of the "A'V'B" synchronization integrator subsystem 725 provide power off delay timing during transfer from "B" power 765 to "A” power 760. The transition time power off delay is accomplished via an easily programmed capacitor value (C9).
  • C9 is the integration storage capacitor that supplies threshold ramp signal to Ul, the timing comparator. Adjustment of C9 changes the delay between the A disconnect relay (RY2), and changing state of the Gatekeeper (RY3) and Neutral (RY1) relays.
  • FIG. 13 shows a circuit diagram of an illustrative timing control subsystem 730a for use in some embodiments of a micro-ATS 700.
  • Comparators in embodiments of the timing control subsystem 730 control switching functions of the relays and activation of the warning indicators and buzzer.
  • the ladder formed by R45, R46, and R47 defines voltages VI and V2.
  • Integrator (C9) is the output of the "A"/"B" synchronization integrator subsystem 725, and is high when the micro-ATS 700 is using "A" power 760 and low when the micro-ATS 700 is using "B" power 765.
  • Pin 2 of the comparator U3a is the Relay On (Low) signal and drives the emitter of Q29. Pin 2 will subsequently apply HV power to the A relay RY2 (RY On).
  • Pin 1 of the comparator U3b is the GC Shunt Drive (Low) and drives the emitter of Q14. Pin 1 will subsequently ground the other side of the A relay (RY2) when asserted. This applies the full HV power (150 Volts) across the A relay (RY2), which can assures fast operation of the A relay (RY2). Times Tl and T2 are controlled by VI and V2, and by the rising and falling slope of the Integrator (C9).
  • Embodiments of the timing control subsystem 730 participate in "A" power 760 to "B" power 765 transitions according to the following technique.
  • the Integrator (C9) will start to ramp down to VI . HV power will then be applied to RY2. Ground is already on the other side of the RY2 coil. RY2 will then disconnect from the "A" power 760 side.
  • the signal GC Shunt Drive will go high releasing the ground from RY2 and allowing current to flow through RY2 to RY1 and RY3. This connects "B” power 765 and Neutral Out to the output power 770 node. Time Tl assures that "A" power 760 is released before "B" power 765 is connected.
  • Embodiments of the timing control subsystem 730 participate in "B" power 765 to "A” power 760 transitions according to the following technique.
  • RY2 is grounded, removing power from the RY1 and RY3 relays. This connects the Neutral Out (NO) to "A" Side Neutral (AN) and Hot Out (HO) to the RY2 relay, which is open at this point.
  • HV power is removed from the RY2 relay, thus connecting "A" Side Power to the Hot Out (HO).
  • HV and “Common 2" are the outputs of the HV power supply. During normal operation, Common and Common 2 are connected together by the Darlington transistor Q22.
  • Embodiments of the timing control subsystem 730 participate in over-current control according to the following technique.
  • a predefined warning point e.g., between 12 and 13 Amps
  • an indicator is illuminated to warn the user that this is the maximum advisable limit for continuous current.
  • V3 and V4 are defined by the ladder resistors R34 and R36 and diode D7 as discussed above.
  • Output current is detected in the output current detect subsystem 740, and an analog voltage is generated there and sent to the timing control subsystem 730.
  • the slope of the "Load Current Sensed Signal" (LCSS) is a function of applied current to the attached load and time.
  • V4 is set to be equivalent to approximately a 12-Amp load on the output of the micro-ATS 700 (or another desired value).
  • pin 13 of comparator U3d will go low. This will cause a loss of conduction through ZD7 guaranteeing that Q22 will turn off when U3d pin 13 goes low, which, in turn, will energize the indicator LED (LED4).
  • R37 and D13 will cause V4 to drop slightly increasing the separation between V4 and the LCSS.
  • C25 also serves to smooth out the difference between V4 and the LCSS.
  • the piezoelectric device driver subsystem 750 will be turned on via the negative supply path.
  • Q23 will be turned on by R41. Q23 will pull the LCSS to 15 volts, latching the fault signal low. Q23 is discussed further with reference to the piezoelectric device driver subsystem 750.
  • the micro-ATS 700 output current exceeds the pre-defined limits (e.g., substantially like a 14.5-Amp fast-blow fuse)
  • an indicator is illuminated that indicates the micro-ATS 700 disconnected the load from the source, and a buzzer sounds to warn the user that the micro-ATS 700 has disconnected the load. This provides "virtual circuit breaker" (VCB) function of the micro-ATS 700.
  • VVB virtual circuit breaker
  • V3 and V4 are defined by the ladder resistors R34 and R36 and diode D7, as discussed above.
  • the LCSS will be detected by the timing and control comparators U3C and U3D. If it represents a current of greater than 15A for 10 seconds, the "Fault (Low)" signal will be generated, latching Q23 and charging CI 2. This will protect fuses Fl and F2. If the overload is removed and switch SWl is pressed, the charge on C12 will be transferred to C13. Q23 will be turned off and power will be restored to the output. If, however, the overload has not been removed, then C15 is still charged, so that "Fault Low” will be regenerated, and Q23 will be turned back on. Repeated pressing of the reset switch (SWl) will charge up C13 from C12 and nothing will happen. This prevents an already hot Fl or F2 from getting repeated hits when SWl is pressed. Thus the virtual fuse protects the internal real fuse.
  • Embodiments of the timing control subsystem 730 provide a number of innovative features.
  • One such feature is that some embodiments of the timing control subsystem 730 provide very high efficiency.
  • Embodiments use very low-power components and high- impedance circuits to minimize power consumption.
  • Another such feature is that some embodiments of the timing control subsystem 730 provide high voltage control.
  • the connection of the low-voltage, low-power control section to the high-voltage relay- power control section is accomplished via an innovative coupling using a variation of the grounded base configuration, where the bases of Q14 and Q29 are referenced to the +15-Volt power supply.
  • the emitters of those transistors are connected to the open collector outputs of U3.
  • timing control subsystem 730 provide power savings in LED illumination of the indicator LEDs (LED 4 and LED 1). The improvement in efficiency is not so much the savings of power when the LEDs are illuminated, as this is the time when there is either a fault or a undesired condition, and is not the predominant operating condition of the micro- ATS 700.
  • the innovative way of powering these LEDs eliminates the need for an additional power supply, and the attendant losses associated with such an addition, is an improvement in efficiency.
  • Both of these LEDs are powered by current through the rest of the circuitry that is utilized, regardless of the conditions of the LEDs otherwise.
  • the current is being passed through LED 4 from the current utilized to operate the timing control subsystem 730.
  • LED 4 When not needed (not illuminated) LED 4 is turned off by the shorting transistor Q22.
  • the only condition when a fault is necessarily indicated (e.g, by LED 1) is when the A relay (RY2) is active, and the Gatekeeper and Neutral relays (RY3, RY1) are shunted into the off condition, thus disconnecting power from the source to the output. In this continuous condition, the current necessary to power the A relay (RY2) is passed through LED1. This current path is already necessary to power the relay, thus it can be used to power the LED with no additional power supply circuitry. This design feature reduces power consumption, and simplifies the total design.
  • FIG. 14 shows a circuit diagram of an illustrative "A” & “B” power switching subsystem 735 for use in some embodiments of a micro-ATS 700.
  • the "A” & “B” power switching subsystem 735 controls when the selected power source to the micro- ATS 700 is switched, either from “A” power 760 to "B” power 765 or “B” power 765 to "A” power 760.
  • Embodiments also control over-current condition switching and relay sequencing, using RY1, RY2, and RY3 to control the flow of power from the "A” power 760 and "B” power 765 inputs to the output power 770 node. Both "A" power 760 and “B” power 765 can be protected by fuses Fl and F2, respectively.
  • FIG. 15 shows a circuit diagram of an illustrative disconnect switch subsystem 745a for use in some embodiments of a micro-ATS 700.
  • This switch is used to disconnect the alternate power source (via node A2 775) from the power supply subsystem 705 when not required.
  • application of the alternate power source occurs only in one condition.
  • the relays of the "A" & "B" power switching subsystem 735 are configured with the A relay (RY2) in the disconnect (i.e., energized) position, and the Gatekeeper relay (RY3) and the Neutral relay (RY1) in the disabled (i.e., non- energized) position.
  • the base of Q4 is pulled negatively by the output of the timing control subsystem 730 and the voltage amplifier via the signal GC On 825. Current through the base of Q4 to the emitter thus clamps off current through the Gatekeeper relay (RY3) and the Neutral relay (RY1).
  • RY Power On 820 is active to energize the via the emitter and base of Q4 to GC On 825, which is pulled to round.
  • the A relay (RY2) is on (i.e., energized), and the Gatekeeper relay (RY3) and the Neutral relay (RY1) are held off. It is worth noting that, in this condition, there is no source of power available to the output power 770 node. This is the state of the relays during a "fault" condition (e.g., an over- current state).
  • the indicator LED (LED 4) is on, and Q22 is not on.
  • the indicator LED (LED 1) is turned on by the current into the base of Q4.
  • the current through ZD 11 (green) thus illuminates the LED in U5 , providing current to the base of Q3.
  • Q3 then conducts and "shorts" out both conduction paths through BR6 allowing AC power to pass from C16 to A2 775.
  • A2 775 then supplies the power supply subsystem 705 with AC source power only during a fault condition.
  • Embodiments of the disconnect switch subsystem 745 provide a number of innovative features.
  • One such feature is that some embodiments of the disconnect switch subsystem 745 provide very high efficiency. As illustrated, the use of three relays in series, rated at 48 volts each, can allow the application of directly rectified AC mains voltage to the relays. This eliminates additional power conversion circuitry, thus reducing parts counts as well as increasing efficiency.
  • Another such feature is that some embodiments of the disconnect switch subsystem 745 use current limiting capacitor (CI 6) when only the A relay (RY2) is activated continuously, which allows use of the otherwise unused "normally open” contact of that relay, and further reduces parts count.
  • disconnect switch subsystem 745 provide relay sequencing when all three relays release simultaneously to prevent arching.
  • the coupling of the fly-back suppression diode D48 on RY2 to the return path rather than directly across the relay provides a slight contact timing delay between the A relay (RY2) losing power and the Gatekeeper relay (RY3) and Neutral relay (RYl). This is accomplished by the anode of D48 being connected to the common rail instead of the more traditional connection to the relay coil.
  • the A relay (RY2) has a higher impedance for the fly-back current to sink into, as the sinking current is also going through RYl and RY3.
  • FIG. 16 shows a circuit diagram of an illustrative output current detect subsystem 740a for use in some embodiments of a micro-ATS 700.
  • Embodiments of the output current detect subsystem 740a detect and measure the presence and various characteristics of the output current from the automatic transfer switch. This also circuit tends to mimic characteristics of a fuse, but only slightly below the thresholds of 15-Amp Fast Blow physical fuses that may be used in the "A" power 760 and "B" power 765 inputs to the micro-ATS 700. This allows the micro-ATS 700 to protect itself without blowing actual physical fuses, which must be replaced.
  • the neutral out of the micro-ATS 700 has a current sense transformer on it.
  • This transformer has a diode bridge, formed by D28, D29, D30, and D31, which full- wave rectifies the "Load Current Sensed Signal" (LCSS, described above with reference to the timing control subsystem 730).
  • C22 and R48 filter out the higher frequencies of the AC current and provide proper impedance loading of the current transformer.
  • Rl 1 and Thermistor RT1 provide thermal compensation so the micro-ATS 700 can remain accurate over a wide range of temperatures.
  • C14, R50, D32, D3, C15, R51, and R75 can be configured and selected to effectively emulate the time VS current opening threshold of a 14.5-Amp fast-blow fuse, with the time part of the curve advanced to open about 33% faster than the equivalent 15-Amp fast-blow fuse.
  • Embodiments of the output current detect subsystem 740 provide a number of innovative features.
  • One such feature is that some embodiments of the output current detect subsystem 740 employ a combination of capacitors and resistors that result in an analog representation of the characteristics of a fast blow fuse.
  • the principal characteristics of the timing of a fuse are as follows: it can carry significant over-current for a short period of time; and, after some time, the fuse will "blow" at or near the rated current of the fuse. This is primarily controlled by the thermal characteristics of the fuse material itself, and how much mass is being heated to the melting point by the applied current.
  • a pseudo two-pole filter is formed by these components to emulate the desired characteristics.
  • the circuit reduces the parts count to the minimum by having the two halves of the two poles interact with each other by allowing current from the first pole to charge the second pole without the reverse.
  • the second pole is discharged by a controlled discharge path through R32.
  • This single point of discharge can thus be used to bias both poles of the filter and alter the overall curve of the response time of this circuit without significantly changing the shape of the curve.
  • This is useful for adjusting the "rating of the fuse" by a simple one component change, specifically R32.
  • the programmed threshold of maximum current rating of the electronic, or virtual, fuse circuit breaker
  • RT1 and Rl 1 are thermister that has a negative temperature coefficient. As temperature goes up, the resistance goes down, in a predictable fashion. By placing these two components in series across the current sense transformer, the load impedance presented to that transformer is affected by temperature. The selection of values for the Thermistor RT1 and Rl 1 result in the overall performance of the virtual circuit breaker effectively mimicking its electro-mechanical equivalent.
  • FIG. 17 shows a circuit diagram of an illustrative piezoelectric device driver subsystem 750a for use in some embodiments of a micro-ATS 700.
  • Embodiments of the piezoelectric device driver subsystem 750a include an innovative implementation of a power driver for driving a piezoelectric buzzer (or similar device).
  • "NOT FAULT” goes low, it provides a ground path for the oscillator formed by Q25, Q27, R40, R42, R44, R39, CIO, and CI 1.
  • Q25 or Q27 When power is applied, either Q25 or Q27 will turn on. If Q25 turns on, the voltage on both sides of CIO will drop, thereby providing a low to the base of Q27 and turning Q27 off. Then CIO will charge up via R44. When the voltage at the base of Q27 reaches approximately 0.6 volts, Q27 will turn on. Both sides of CI 1 will drop. Q25 will turn off, and so on, back and forth.
  • Q21 and Q26 are emitter followers. As the collector of Q25 charges up when Q25 is off, Q23 emitter will follow its base driving the piezoelectric buzzer. When Q25 turns on it will sink connect from the piezo through D2, and similarly for Q26, D4, and Q27.
  • Embodiments of the piezoelectric device driver subsystem 750 such as the
  • circuits illustrated and described with reference to FIG. 17 provide a number of innovative features.
  • One such feature is that the circuit provides high efficiency with a low parts count. As illustrated, high efficiency is achieved by combining the oscillator and amplifier necessary to drive the piezo into a power oscillator, and the combination also tends to minimize parts count. In this configuration, nearly all the current expended in the circuit for both the oscillator function and the power amplifier function is applied to the piezo crystal for conversion to acoustic energy.
  • micro-ATSs 700 form novel embodiments of micro-ATSs 700 and provide a number of features.
  • One such feature is ultra-low power consumption. Use of a transformer-less power supply reduces overall loss in converting from 120 VAC mains voltage to DC voltages required for internal circuits.
  • Another such feature is that virtually no power is consumed on the non-connected side (e.g., on the "B" power 765 side when the load is connected to "A" power 760).
  • the use of optical isolation in the power supply virtually eliminates cross currents between the "A" power 760 and "B" power 765 inputs.
  • Still another such feature is that arc suppression is provided at the disconnection of the "B" power 765 side on the contacts of the affected relays by timing the break of the relay contacts at the zero crossing of the current presented to those contacts.
  • Additional features are realized by selections of certain circuit components and/or topologies.
  • One such feature is that ultra low power consumption can be achieved by utilizing a capacitor current limit on the AC input.
  • the selected capacitor (0.22uF) limits 60 Hz to about 10 milliamps into the 48-volt load. At 50 Hz, the limit is 8 milliamps, both within operating range of the connected relays.
  • Another such feature is that use of the same path of current in the "A" power loss detect subsystem 715 for illuminating the "on A" indicator, activating the optical isolator link to the
  • Another such feature comes from use of the Normally Open contact of the A disconnect Relay (RY2) as an AC Mains diversion path for and alternate power source during a fault condition. In this condition, only the A relay is activated. The alternate path described allows a second rectifier and a capacitor current limit on the AC Mains, so the current rating of the coil of the single activated relay (RY2) is within operating range of 8 - 10 milliamps.
  • Another such feature is that use of total circuit pass current on the low voltage timing control subsystem 730 to energize the indicator LED (LED 4) that indicates approaching the over current condition, and shunting the LED to the off condition when not in use, eliminates the need for an auxiliary power supply for this device.
  • SCR simile requires very little current to cause to latch into a conduction state. This latched state will continue until passing current through the pair ceases. When it ceases, the latch will disconnect and not allow current to flow until re-initiated.
  • the otherwise inaccessible second PN junction of the SCR simile is accessible, thus allowing the action at this point to also affect Q15 at a certain point in the sequence to be beneficial.
  • the Q15 is the timer reset for delaying the return to "A" power 760 from an "A" loss state. This must be reset every time the SCR simile goes into conduction.
  • the avalanche activation of the SCR simile at that time can be utilized to force the reset of the timing capacitor C8 via Ql 8.
  • An optically coupled synchronization pulse and level from the "B" power 765 source is used for two functions: to synchronize transfer from the "B" power 765 to the "A” power 760 at the zero-crossing of the "B” power 765; and to provide a source of signal that indicates the loss of the "B” power 765 and thus force the connection to the "A” power 760, regardless of the state of other circuits that normally would influence transfers to the "B” power 765 side.
  • the optically coupled sync signal holds off the signal from the optical coupler that indicates that "A" power 760 is ready to be transferred to until the sync pulse appears (via Q19, Q20, and the Ul opto-coupler transistor).
  • the release of the hold on the signal from Q33 allows the circuit to initiate the integrate function for the comparator, U3 to utilize to complete the transfer.
  • the release of the hold function then latches the sync pulses out so no additional pulses can be present via bias on the base of Q19.
  • This arrangement of Q21, Q26, Q25, and Q27 forms a transistor pair oscillator.
  • Q25 and Q27 form the basis of an astable oscillator with R39, R40, R42, R44, CIO, and CI 1 forming the components that determine oscillation.
  • the innovative addition of D2, D4, Q21, and Q26 turn the oscillator into a power oscillator capable of driving piezo components bilaterally (e.g., one side of the Piezo is at +15, while the other is at Common; then it switches, and the opposite is true).
  • This same power oscillator can be used for other applications such as driving a miniaturized switching power supply, or for signal source in a small tester. Numerous applications exist for a miniature, low power, very low cost power oscillator.
  • ATSs can be deployed at various points in the power distribution topology to provide for automatic failover from a primary power source to a backup power source typically at one of three points in the power distribution topology: the panelboard on the wall, where the branch circuits originate; the end of the branch circuit in the rack where the power is fed to plugstrips; or between the plugstrip and the EDP equipment being powered.
  • the choice of where to place auto-switching functionality in a power distribution topology can involve consideration of a number of issues.
  • Power distribution topologies used in data centers can be considered rooted tree graphs (i.e.,
  • FIGS. 18A - 18D show illustrative power distribution topologies.
  • a power distribution topology 1800a is shown having an ATS 1845a disposed in the root nodes 1830 of the topology.
  • the power distribution topology 1800a includes a core infrastructure 1810, root nodes 1830, distribution nodes 1850, and leaf nodes 1870.
  • the root nodes 1830 is considered to be "downstream" of the core infrastructure 1810.
  • Each portion of the power distribution topology 1800a includes power distribution components.
  • the utility grid 1805 feeds a site transformer 1815 in the core infrastructure 1810 (e.g., a step down transformer).
  • the core infrastructure 1810 also includes a local generator 1820.
  • the site transformer 1815 and the local generator 1820 act as two independent sources of power for the power distribution topology 1800a. In other embodiments, other independent sources, such as independent utility grids, and/or more than two independent sources may be used.
  • main switch gear 1825 can allow the entire core infrastructure 1810 to be switched to alternate power (e.g., from the site transformer 1815 to the local generator 1820).
  • UPSs uninterruptable power supplies
  • Each UPS 1835 can be considered as a root node 1830 in the power distribution topology 1800a.
  • ATS 1845a Power paths coming from the main switch gear 1825 and the UPS(s) 1835 are routed to an ATS 1845a disposed as a root node 1830.
  • the ATS 1845a is configured to automatically switch between core power coming from the core infrastructure 1810 and power coming from the UPS(s) 1835 (e.g., in case of a full or partial power failure in the core infrastructure 1810. In this way, the ATS 1845a provides reliable, constant power to downstream components.
  • power is delivered from the root nodes 1830 to the distribution nodes 1850.
  • the output power from the ATS 1845a is delivered to one or more panelboards 1855.
  • multiple power distribution panelboards 1855 are used, since they come only so large in power capacity and number of circuit breaker stations. Also, it may be more efficient to locate panelboards 1855 so as to minimize the average power whip length, and as many as is practical may be used to accomplish this purpose.
  • the distribution nodes 1850 are considered as originating at the main panelboards 1855 and ending at equipment racks 1875 (e.g., leaf nodes 1870). In the illustrative embodiment, power is distributed at the equipment racks 1875 via plugstrips 1880.
  • the plugstrips may also have circuit breakers in them. It is worth noting that some of skill in the art confusingly refer to both panelboards 1855 and plugstrips 1880 as power distribution units (PDUs). Accordingly, for the sake of clarity, the terms panelboards and plugstrips are used herein.
  • PDUs power distribution units
  • the equipment racks 1875 include EDP equipment 1885. Many types of EDP equipment 1885 are possible, and may be rack-mounted in the equipment racks 1875. Each piece of EDP equipment 1885 can be plugged into one or both plugstrips 1880. For example, EDP equipment 1885 having multiple internal power supplies may be plugged into multiple plugstrips 1880
  • the ATS 1845a is deployed as a root node 1830 of the power distribution topology 1800a, upstream of the distribution nodes 1850 and leaf nodes 1870.
  • This upstream placement may provide ATS functionality with only a single (or relatively few) ATSs.
  • power distribution efficiency the amount of power that is "lost" by the insertion of ATSs into the power distribution system.
  • No ATS is 100% efficient (i.e., they all have a loss factor).
  • CBEMA Computer and Business Equipment Manufacturers Association
  • Mechanical relay-based ATSs use one or more mechanical relays to switch between their input power sources.
  • relays have two primary loss factors, the contact area of the relay, and any power the relay may require to maintain it in the "ON" state (i.e., in which it is conducting current).
  • the shape and material of the contacts is carefully chosen and engineered to minimize resistance across the contacts, yet minimize or prevent arcing across the contacts when they are switching. Also, since some arcing may occur in some circumstances, the contacts must be designed to minimize the possibility of the arc "welding" the contacts shut, which can be highly undesirable.
  • the contacts are mounted (e.g., on an armature) so that they can be moved to accomplish their switching function.
  • the contact mass, range of motion, mechanical leverage, and force used to move the armature are all relay design issues.
  • the range of motion is dictated by the gap needed between the contacts to minimize arcing at the maximum design current level. As the maximum design current is increased, the gap also tends to increase.
  • the mass of the contact is accelerated by the force applied to the armature, which has a practical limit.
  • the relay transfer time can exceed a threshold time limit (e.g., the CBEMA recommend maximum of approximately 20 milliseconds of power outage for continued operation of modern switched power supplies).
  • a threshold time limit e.g., the CBEMA recommend maximum of approximately 20 milliseconds of power outage for continued operation of modern switched power supplies.
  • Well designed relay-based ATSs typically manifest a loss factor of about 0.5% or less. They also have power supplies to power their internal logic that typically use in the range of 12 - 20 watts in operation.
  • the switches use solid state semiconductors to accomplish switching between their input power sources and their output load. They can typically switch faster than comparable relay-based switches, because they use semiconductor-based switching rather than mechanical relays. However, the semiconductors also have a loss factor, and the efficiency of this type of switch is often less than that of a relay based switch (e.g., around 1%). Also, they are often less reliable, unless they are built with redundant internal failover capability, which can make them appreciably more expensive. As with their relay-based counterparts, solid state -based ATSs typically have power supplies to power their internal logic that use in the range of 12 - 20 watts [should this be 20?
  • Automatic transfer switching can be done in the core infrastructure 1810 to insure continuity of connection to a valid power source, such as utility power grid feeds or generators.
  • a valid power source such as utility power grid feeds or generators.
  • the transfer time of relay-based switches that can handle the power capacities required in the core infrastructure 1810 is typically too slow to avoid shutdown by connected EDP equipment 1885 for the reasons described earlier (e.g., according to the CBEMA guidelines). Accordingly, ATSs of this type tend to be placed upstream of UPS units, where brief power outages that these switches create on transfer can be covered by the UPS units.
  • the single ATS 1845 a will switch every branch circuit in a given panelboard 1855 to a secondary power source when the primary power source fails.
  • this type of upstream placement of the ATS 1845a in the power distribution topology 1800a can mitigate certain issues (e.g., power loss, transfer time, etc.) that can be exacerbated when many ATSs must be used throughout a power distribution topology 1800a.
  • failure of the single ATS 1845a upstream of the panelboard 1855 can cause many downstream EDP devices to be deprived of power.
  • a typical panelboard has a capacity of 225KVA, and 84 or 96 circuit breaker stations.
  • FIG. 18B shows another illustrative a power distribution topology 1800b having an ATS 1845b disposed further downstream, in the distribution nodes 1850 of the topology.
  • the power distribution topology 1800b is illustrated as having the same core infrastructure 1810 and similar root nodes 1830, distribution nodes 1850, and leaf nodes 1870 as those shown in FIG. 18A.
  • the ATS 1845b is implemented in the distribution nodes 1850.
  • the alternate power paths feed distribution nodes 1850.
  • the distribution nodes 1850 include one or more main panelboards 1855a and one or more zone panelboards 1855b. Some or all of the zone panelboards 1855b include an integrated ATS 1845b (or multiple integrated ATSs 1845b).
  • the alternate power sources e.g., from the main switch gear 1825 and the UPS(s) 1835
  • the zone panelboards 1855b can then supply power to equipment racks 1875 in their respective zones.
  • ATSs 1845b there are still relatively few ATSs, which can still reduce issues associated with ATS inefficiencies, etc.
  • placing the ATSs 1845b further downstream allows the ATSs 1845b to handle power transfer only for a subset of the equipment racks 1875 in a large data center (i.e., a particular zone). Accordingly, failure of an ATS 1845b will cause a more limited number of downstream EDP devices to be deprived of power. This can have a less catastrophic impact to the data center than from a similar failure with a farther upstream ATS.
  • FIG. 18C shows yet another illustrative a power distribution topology 1800c having an ATS 1845c disposed even further downstream, in the leaf nodes 1870 of the topology.
  • the power distribution topology 1800c is illustrated as having the same core infrastructure 1810 and similar root nodes 1830, distribution nodes 1850, and leaf nodes 1870 as those shown in FIGS. 18A and 18B.
  • the ATS 1845c is implemented at or near the equipment racks 1875 in the leaf nodes 1870.
  • the main switch gear 1825 and the UPS(s) 1835 feed distribution nodes 1850 (e.g., main and/or zone panelboards 1855).
  • the distribution nodes 1850 in turn feed the leaf nodes 1870.
  • panelboards 1855 supply power to equipment racks 1875.
  • an ATS 1845c is disposed at or near each equipment rack 1875 or set of equipment racks 1875.
  • the ATS 1845 c is configured to fit within a rack space, as discussed more fully below.
  • the ATSs 1845c are place on top of equipment racks 1875, next to equipment racks 1875, or in any other useful location.
  • the primary power path is distributed from the panelboard 1855 to a first plugstrip 1880a of the equipment rack 1875.
  • a second plugstrip 1880b of the equipment rack 1875 is feds by the ATS 1845c, which is configured to switch between the primary and secondary power sources, as needed.
  • EDP equipment 1885 having only a single plug can be plugged into the second plugstrip 1880b and powered by the primary power source if of sufficient quality, or otherwise by the secondary power source.
  • other EDP equipment 1885 e.g., having multiple internal power supplies, two plugs, etc. is plugged into both plugstrips 1880.
  • Rack space in a data center can be very expensive.
  • the data center infrastructure of generators, UPS units, power distribution, raised floor, computer room cooling, raised floors, etc. is often a very large capital investment and a large ongoing operational expense.
  • One standard rack unit (“1U") of rack space in a standard forty-two-unit (“42U”) equipment cabinet is equivalent to 2.5% of the space available in that rack.
  • rack-mounted ATSs in large numbers in equipment racks uses a lot of rack space, which represents a loss of space that can be used for EDP equipment and would therefore be very undesirable. Accordingly, ATSs do not tend to be deployed in that configuration.
  • FIG. 18D shows still another illustrative a power distribution topology 1800d having ATSs 1840d disposed still further downstream at the EDP equipment 1885, in the end leaf nodes 1870 of the topology.
  • the power distribution topology 1800d is illustrated as having the same core infrastructure 1810 and similar root nodes 1830, distribution nodes 1850, and leaf nodes 1870 as those shown in FIGS. 18A - 18C.
  • the main switch gear 1825 and the UPS(s) 1835 feed distribution nodes 1850 (e.g., main and/or zone panelboards 1855), which feed the leaf nodes 1870.
  • panelboards 1855 supply power to equipment racks 1875.
  • an ATS 1845d is disposed in, at, or near each piece of EDP equipment 1885 (or at least those pieces of EDP equipment 1885 for which the ATS functionality is desired).
  • the ATS 1845d is configured so that a number of ATSs 1845d can be combined to fit within a rack space and to run in parallel for a number of connected pieces of EDP equipment 1885, as discussed more fully below.
  • the ATSs 1845d may be the micro ATSs 700 described above, and may be deployed in parallel with integrated control logic to construct a large capacity, fast, efficient, and relatively low cost ATS for use in providing redundant power distribution for EDP equipment 1885.
  • the ATSs 1845d are integrated into power cords, into pieces of EDP equipment 1885, or disposed in any other useful way.
  • a single ATS failure would only impact the EDP equipment 1885 it serves (i.e., likely to be only one EDP device).
  • this configuration could involve deployment of many more ATSs than in any of the other configurations illustrated in FIGS. 18A - 18C.
  • Each ATS can add inefficiencies in power distribution (e.g., power loss), space usage (e.g., by taking up valuable rack space), etc.
  • embodiments include a micro-ATS 700 (e.g., acting as any of ATS 1845b - ATS 1845d) that can be deployed in configurations where size and efficiency are of concern.
  • the Zonit ⁇ 8TM is very small (e.g., 4.25-inches x 1.6- inches x 1-inch, or less than 10 cubic inches) and very efficient (e.g., less than 0.2 volts at maximum load loss).
  • Certain implementations use no rack space, as they are self- mounted on the back of each piece of EDP equipment 1885, incorporated in the structure of the equipment rack 1875 outside the volume of the equipment rack 1875 used to mount EDP equipment 1885, incorporated in rack-mounted plugstrips 1880, or incorporated in an in-rack or near-rack panelboards 1855 (i.e., any of which being possible due to the small form-factor of the micro-ATS).
  • the micro-ATS 700 is small enough to be integrated directly into the EDP equipment 1885 itself.
  • the small form factor of the micro-ATS 700 can enable usage of 24-inch outside-to- outside width EDP equipment racks 1875.
  • These types of equipment racks 1875 can provide certain advantages. For example, the cabinets fit exactly on two-foot by two- foot raised floor tiles, which makes putting in perforated floor tiles to direct air flows easy, since the racks align on the floor tile grid. Further, the equipment racks 1875 can save precious data center floor space. NEMA equipment racks 1875 are not typically standardized for overall rack width, and the narrower the rack is, the more racks can be fit in a given row length.
  • a 24-inch equipment rack 1875 will save three inches over the very common 27-inch width equipment racks 1875, thereby allowing for one extra equipment rack 1875 for each eight equipment racks 1875 in a row (i.e., nine 24-inch equipment racks 1875 can be fit into the floor space of eight 27-inch equipment racks 1875).
  • Narrower equipment racks 1875 are becoming more practical with modern EDP equipment 1885, for example, since almost all models now utilize front to back airflow cooling (i.e., as opposed to side-to-side cooling, which used to be common, but has now almost completely disappeared).
  • a 24-inch equipment rack 1875 has appreciably less space on the sides for ancillary equipment like vertical plugstrips 1880, ATSs, etc., such that those components must have as small a form- factor as is practical to fit into the equipment racks 1875.
  • the micro-ATS 700 allows efficient, cost-effective and rack space saving per device or near per device (ratios of one micro-ATS 700 per one piece of (or a low integer number of) EDP equipment 1885) and allows highly parallel and highly efficient auto-switched power distribution methods to be utilized. It should be pointed out that the ratio of micro-ATS 700 units to EDP equipment 1885 can be selected to optimize several interrelated design constraints, reliability, cost and ease of moving the EDP equipment 1885 in the data center. The one-to-one ratio maximizes per-device power reliability and ease of moving the device while keeping it powered up.
  • this can be performed using a device-level ATS, like the micro-ATS 700, by doing a "hot walk,” in which the device is moved by first unplugging one ATS power cord, moving the plug to a new location, unplugging the second ATS power cord, etc.
  • Long extension cords make "hot walks” easier. Ethernet cables can be unplugged and reinserted without taking a modern operating system down and TCP/IP connections will recover when this is done.
  • cost can be reduced by using ratios other than one- to-one for micro-ATS 700 units to pieces of EDP equipment 1885.
  • a limiting factor can be micro-ATS 700 power capacity and what raised level of risk the data center manager is willing to take, since the more devices connected to any ATS the greater the impact if it fails to function properly.
  • FIG. 19 an illustrative traditional power distribution topology is shown, according to some prior art embodiments.
  • the topology includes a "double conversion" technique using UPS units (e.g., 1835), which in some recent implementations include flywheel UPS devices.
  • UPS units e.g., 1835
  • flywheel UPS devices Even some of the best double conversion UPS units used in data centers have power efficiencies that vary as their load changes.
  • FIG. 20 illustrates an efficiency versus load graph for a typical double- conversion UPS unit.
  • standard UPS units may typically average 85 - 90% efficiency
  • flywheel UPS units may average around 94% efficiency at typical load levels. This level of efficiency may be unacceptable in many instances, for example, when power costs are stable and relatively low, and when climate impacts of carbon- based fuels are appreciated.
  • power has quickly changed from an inexpensive commodity to an expensive buy that has substantial economic and environmental costs and key implications for national economies and national security.
  • a traditional UPS powered data center more typically has efficiencies in the 88-92% range, for example, because data center managers tend to run UPS units at less than 100% capacity to account for any needed equipment adds, moves, or changes.
  • EDP equipment 1885 can have various costs.
  • One cost arises from the purchase of the additional power supply(s).
  • the supplies are often specific to each generation of equipment, and therefore must be replaced in each new generation of equipment, which can be as short as three years in some organizations for typical servers.
  • servers are currently most cost-effective when bought in the "pizza box" form factor.
  • servers deployed in large data centers are typically all
  • Commododity Intel X86 architecture compatible central processing units (CPU's). These servers are used to power most of the large server farms running large websites, cloud computing running VMWare or other virtualized solutions, high performance computing (HPC) environments, etc. Commodity servers have great pressure to be cost competitive, especially as regards their initial purchase price. This in turn can influence the manufacturers' product managers to choose the lowest cost power supply solutions, potentially at the expense of yielding the best power efficiency. Another cost of additional power supplies is that each power supply has an associated loss factor. For example, power supplies are not 100% efficient and costs can be reduced by designing the supplies to run most efficiently at a given load range (e.g., typically +/- 20% of the optimum expected load). Many power supplies have an efficiency curve that resembles that shown in FIG.
  • a product manager for a server manufacturer may desire to sell a server in two configurations, with one or two power supplies, and may prefer to stock only a single power supply model to avoid additional costs due to stocking, selling, and servicing two models of power supplies.
  • This can trade capital expense (the server manufacturer can sell the server at a lower initial price point) for operational expense.
  • the DC output bus will typically be a common shared passive bus in the class of commodity server that is most often used in large scale deployments.
  • Adding power source switching to this class of server to gain back efficiency may be too expensive for the market being served and can add another potential point of failure for which additional costs may be incurred to add redundancy for greater reliability.
  • EDP power supplies are almost all auto-ranging (i.e., they accept 110-240V input) and all switched (i.e., they draw on the AC input power for just a short period of time, convert that energy to DC, then repeat).
  • Power supplies of this type can be more resistant to power quality problems, because they only need to "drink" one gulp at a time, not continuously. If the input AC power voltage range is controlled within a known range, they will typically function very reliably. While the power supplies may require sufficient energy in each "gulp" and that the input power is within the limits of their voltage range tolerance, they may not require perfect input AC waveforms to work well. This can make it possible to use a data center power distribution system that is much more efficient than a fully UPS-supplied power system at a very reasonable capital expense.
  • Embodiments address efficient power distribution (e.g., in data centers) using highly parallel automatic transfer switching.
  • UPS unit(s) e.g., due to conversion losses.
  • One technique for avoiding these losses is to use filtered utility line power, though this can bring a set of issues that need to be solved before the methodology can be practical.
  • FIG. 21 shows an illustrative power distribution topology 2100, according to various embodiments.
  • the power distribution topology 2100 can be considered as having a core infrastructure, which can include a site transformer 1815 (e.g., a step down transformer fed by the utility grid 1805), a local generator 1820, and a main switch gear 1825.
  • the site transformer 1815 and the local generator 1820 act as two independent sources of power for the power distribution topology 2100.
  • main switch gear 1825 can allow the entire core infrastructure to be switched to alternate power (e.g., from the site transformer 1815 to the local generator 1820).
  • Power is provided from the core infrastructure to the root nodes of the power distribution topology 2100.
  • core power is delivered from the main switch gear 1825 to one or more UPSs 1835 and a Transient Voltage Surge Suppression (TVSS) unit 2110.
  • the TVSS acts as a very efficient (e.g., often above 99.9%) and mature technology for filtering input power to the distribution nodes of the topology.
  • UPSs 1835 may typically be between 85% and 94%> efficient (or less).
  • the root nodes of the power distribution topology 2100 effectively deliver an efficient, primary (“A”) power source 760 and a less efficient, secondary (“B”) power source 765.
  • A primary
  • B secondary
  • Each of these power sources is then routed through the distribution nodes of the power distribution topology 2100.
  • one or more panelboards 1855 e.g., main and/or zone panelboards 1855 are used to distribute "A" power 760 and "B" power 765 to the equipment racks 1875 at the leaf nodes of the power distribution topology 2100.
  • the panelboards 1855 deliver power to receptacles 2125 (e.g., standard 120-volt NEMA receptacles).
  • Each equipment rack 1875 includes at least a first plugstrip 1880a having a plug 2120 configured to plug into a receptacle 2125 supplying "A" power 760, and a second plugstrip 1880b having a plug 2120 configured to plug into a receptacle 2125 supplying "B" power 765. Accordingly, both "A" power 760 and "B" power 765 are delivered by the power distribution topology 2100 all the way to the equipment racks 1875 in which the EDP equipment 1885 is installed.
  • Each piece of EDP equipment 1885 for which ATS functionality is desired can be supplied with redundant power (i.e., both "A” power 760 and “B” power 765) through the plugstrips 1880 by exploiting micro-ATS 700 functionality.
  • micro- ATSs 700 are electrically coupled with (e.g., plugged into) "A" power 760 and "B” power 765 (e.g., via the first and second plugstrips 1880, respectively), and are electrically coupled with one or more pieces (e.g., typically one or a small number) of EDP equipment 1885.
  • the micro ATSs 700 are configured to deliver "A" power 760 to the EDP equipment 1885 with it is of sufficient quality and to switch over to "B" power 765 when a partial or complete power failure is detected on "A" power 760.
  • FIGS. 22A and 22B show illustrative parallel micro-ATS modules 2200, according to various embodiments. According to some implementations, the parallel micro-ATS modules 2200 are configured to fit within a single rack space ("1U").
  • a rack-mountable enclosure 2205 contains two or more micro-ATSs 700 (e.g., twelve) configured to operate in a parallel fashion.
  • the parallel micro-ATS modules 2200 connect to "A" power 760 and "B" power 765 (e.g., via an "A" power cord 2210a and a “B” power cord 2210b, respectively).
  • the amperage of the "A” and “B” power sources can be chosen to match the number of auto-switched output receptacles (described below) and their anticipated average and/or maximum power draw.
  • the parallel micro-ATS modules 2200 takes input power from "A" power 760 and "B” power 765 and distributes to its component micro-ATSs 700.
  • the "A” and “B” power sources may be single phase, split-phase or three-phase, though they may typically be substantially identical.
  • Each component micro-ATSs 700 feeds an output receptacle (ATS receptacle 2225 of FIG. 22A) or a hard-wired power cord (output cord 2235 of FIG. 22B).
  • the ATS receptacle 2225 or output cord 2235 is accessible via the face (e.g., rear face) of the enclosure 2205 for connection with EDP equipment 1885 in the equipment rack 1875.
  • Certain embodiments also include one or more ATS indicators 2230 that can indicate, for example, whether a particular micro-ATS 700 is functioning properly, which power source is currently being supplied, etc.
  • the parallel micro-ATS modules 2200 include circuit breakers 2215, optionally with visual power status indicators 2220, to allow disconnecting the parallel micro-ATS modules 2200 electrically from the branch circuits that feed it (i.e., from "A” power 760 and/or from "B” power 765).
  • some embodiments also include a control module 2250 to provide control functionality for improved parallel operation of the component micro-ATSs 700 in the parallel micro-ATS modules 2200.
  • the control module 2250 helps parallelize operation of the multiple component micro-ATSs 700.
  • the control module 2250 offloads certain functionality of the component micro-ATSs 700.
  • the parallel micro-ATS module 2200 is designed to switch all its micro-ATSs 700 upon detection of the same condition (e.g., a particular threshold reduction in power quality).
  • certain implementations move functionality of the detection components (e.g., one or more of the "A" power voltage range detect subsystem 710, "A" power loss detect subsystem 715, output current detect subsystem 740, etc.) to the control module 2250.
  • the control module 2250 detects a power failure in the "A" power 760 source, for example, it may force all the component micro-ATSs 700 in the parallel micro-ATS module 2200 to switch their outputs to "B" power 765.
  • Embodiments of the parallel micro-ATS modules 2200 are configured so that the enclosure 2205 can be mounted within the equipment rack 1875, on top of the equipment rack 1875, or the side of the equipment rack 1875.
  • the size of the enclosure 2205 can be minimized due to the very small form factor of component micro-ATSs 700.
  • Some implementations of the enclosure 2205 are configured to have dimensions within one NEMA standard rack unit (e.g., 1.75-inches in height).
  • embodiments of the micro-ATSs 700 have dimensions of 4.25 inches deep by 1.6 inches high by one inch wide, so that twelve or more micro-ATSs 700 can easily fit within a parallel micro-ATS modules 2200, along with any cabling, control circuitry, buses, cooling, etc.
  • FIG. 23 shows an illustrative power distribution topology that includes a rack-mounted parallel micro-ATS module 2200, according to various embodiments. Though not shown, it is assumed that a core infrastructure is used to provide at least two
  • the power sources can be delivered through one or more root nodes to one or more distribution nodes. As illustrated, root nodes of the power distribution topology 2300 effectively deliver a primary ("A") power source 760 and a secondary (“B”) power source 765 through one or more panelboards 1855.
  • A primary
  • B secondary
  • the panelboards 1855 deliver power to receptacles 2125 (e.g., standard 120-volt NEMA receptacles).
  • Each equipment rack 1875 includes a parallel micro-ATS modules 2200 that can connect with the "A" power 760 and the "B" power 765 sources using input power cords 2210 (e.g., via respective plugs 2120 configured to plug into respective receptacles 2125). Accordingly, both "A" power 760 and "B" power 765 are delivered by the power distribution topology 2300 all the way to the equipment racks 1875 in which the EDP equipment 1885 is installed via the parallel micro-ATS modules 2200.
  • Each piece of EDP equipment 1885 for which ATS functionality is desired can be supplied with redundant power (i.e., both "A" power 760 and "B" power 765) by being connected to a component micro-ATS 700 of the parallel micro-ATS module 2200.
  • the connection to the parallel micro-ATS module 2200 may be implemented using a receptacle (e.g., ATS receptacle 2225 of FIG. 22A) or an output power cord (e.g., output cord 2235 of FIG. 22B).
  • the illustrated embodiment shows cords 2305 connecting the parallel micro-ATS module 2200 with each piece of EDP equipment 1885.
  • hydra power cords are included.
  • the cords 2305 may be combined into a single hydra cord to further allow the
  • Hydra cords can be optimized to increase power efficiency delivery, cord routing, eliminate cord tangle, and incorporate locking power cord functionality.
  • the hydra power cords are connected to the parallel auto-switch module via standard receptacles, locking or non-locking or directly attached via hard- wire. In environments where the contents of the equipment rack are pre-designed, the hydra cords can be used as a wiring harness for the "programmed deployment.”
  • the number of heads on the hydra cord can be varied to match the desired average power output to each connected end-user device.
  • the length and gauge of the hydra power cord (both the main feed section and the separate feeds to each "hydra head") can be optimized to minimize electrical transmission losses and power cord tangle by optimizing the cord lengths for each hydra cord to supply power to a particular set of equipment positions in the equipment rack 1875.
  • a set of appropriately sized hydra cables can be used to feed each equipment location in the rack at whatever interval is desired, such as one NEMA standard equipment mounting space.
  • locking power cord technologies can be used to improve the security of power delivery.
  • standard NEMA L5-15 locking receptacles for 120V service or NEMA L6-15 receptacles for 200V+ service can be used.
  • the "hydra cord head" on the output cords can be equipped with IEC locking receptacles (CI 3 and CI 9) using various technologies.
  • the enclosure 2205 takes up rack space, it can also eliminate the need for in rack plugstrips 1880, which are typically mounted vertically in the equipment rack 1875. Accordingly, data center floor space can be optimized by reducing the width of each equipment rack 1875. For example, equipment racks 1875 are often around 27" wide to allow adequate space to mount a variety of vertical plugstrips 1880, which do not have industry standardized dimensions.
  • the NEMA standard equipment width that is most commonly used is 19 inches. Therefore the total width and depth of the rack determine its floor area usage. By eliminating the need to run anything but power cords and network cords down the sides of the rack, it is possible to specify narrower racks, down to a width of approximately 21 inches.
  • a typical rack-mountable ATS can cause the entire set of EDP equipment 1885 in the rack to fail if the ATS fails (e.g., as in embodiments of the configuration illustrated in FIG. 18C).
  • the highly parallel implementation described herein can minimize the domain of failure to only the one (or small subset) of end-user devices that are powered by each individual micro-ATS 700 in the parallel micro-ATS module 2200. This can appreciably improve reliability, servicing, etc.
  • Power distribution topologies like the ones illustrated in FIG. 21 and 23, provide a number of features.
  • One such feature involves input voltage range control.
  • Modern power supplies can tolerate a wide range of power quality flaws, but they typically cannot survive lengthy input power over- voltage conditions.
  • the TVSS unit 2110 can filter transient surges and spikes, but it does not compensate for long periods of input power over-voltage (i.e., these are passed through to the root nodes).
  • embodiments address out-of-range voltages (e.g., modern power supplies are not typically damaged by under-voltage conditions, but will still shutdown) by switching to the conditioned UPS power if the utility line power voltage goes out of range.
  • Some embodiments implement over-voltage protection at the utility step-down transformer (e.g., site transformer 2110).
  • Auto-ranging transformers of this type are available and can often be ordered from utility companies. Configurations have a set of taps on their output coil and automatically switch between them as needed to control their output voltage to a specified range.
  • Step-down transformers of this type of this type are not usually deployed by utility companies (e.g., because of cost), but they can often be specified and retrofitted for power distribution topologies if requested.
  • ATSs can be placed in a number of locations throughout the topology, for example, including at panelboards, at the end of a branch circuit, at the device level, etc.
  • the power distribution topologies shown in FIGS. 21 and 23 illustrate implementing the switching by placing micro-ATSs 700 at the device level. It should be noted that a semiconductor-based ATS could be used upstream of the UPS, but this can be relatively very expensive, and a failure of the ATS could result in potentially catastrophic effects, as all of the powered EDP units could have their power supplies damaged or destroyed if the ATS unit fails to switch.
  • Another feature of power distribution topologies using micro-ATSs 700 is the availability of auto-switching of all single power supply (or cord) EDP devices. If utility line power fails, it is desirable for all single power supply EDP devices to be switched to a reliable alternate power source, such as power supplied via the UPS. EDP equipment 1885 may require that the switching is accomplished within a predetermined maximum time (e.g., the CBEMA 20 millisecond guideline). Notably, while plugging all devices directly into the UPS would provide highly reliable power, it would also appreciably reduce power distribution efficiency over any implementation that only uses the UPS during the times when utility power is down. This can have significant impacts in environments like large server farms, where the cost constraints are such that single power supply configurations for the massive number of servers are greatly preferred for cost and efficiency reasons, and services will not be much or at all interrupted by the loss of a single or a few servers.
  • EDP equipment 1885 implemented with multiple power supplies typically share the load among their available power supplies. It is possible to build an EDP device that switches the load between power supplies, so that only one or more are the active supplies and the others are idle, but as described earlier, this is rarely done for both cost and reliability reasons.
  • Embodiments ensure that multi-power-supply EDP equipment 1885 draw on only filtered utility line power if it is available (and of sufficient quality) and switch to the UPS only if it is not. To this end, embodiments auto-switch each secondary power supply unit between the utility line and UPS power. Otherwise, the UPS unit will bear a portion of the data center load, which can lower the overall efficiency of the power distribution.
  • Still another feature of power distribution topologies using micro-ATSs 700 is avoidance of harmonic reinforcing power load surges. If utility line power fails, all EDP devices must draw on the UPS unit until the generator starts and stabilizes. Modern generators used in data centers have very sophisticated electronics controlling their engine “throttle". The control logic of the generator is designed to produce maximum stability and optimum efficiency. However, it takes a certain amount of time to respond to a changed electrical load and then stabilize at that new load. If the load put on the generator changes too fast in a repeating oscillation pattern, it is possible to destabilize the generator, by defeating its control logic and forcing it to try to match the oscillations of the power demands. This can either damage the generator or force it to shutdown to protect itself. In either case the data center can potentially go off-line, which can be a very undesirable result. There are several potential scenarios that can potentially cause this problem.
  • Utility line power is outside the control of the data center operator. It can be affected by weather, equipment faults, human error and other conditions. It can fail intermittently which poses a potential hazard to the core data center power infrastructure. If utility power goes on and off intermittently, and the timing of the on-off cycles is within a certain range, auto-switching between the utility line source and the generator (even filtered by the UPS units) can result in harmonic reinforcing power load surges being imposed on the generator.
  • Another such scenario occurs when there is load/voltage oscillation.
  • a load is switched onto the local generator 1820, especially a large load, its output voltage can momentarily sag.
  • the local generator 1820 may then compensate by increasing throttle volume and subsequent engine torque, which increases output current and voltage.
  • the local generator 1820 auto-starts.
  • the local generator 1820 may be switched into the system by the main switch gear 1825 (e.g., the local generator 1820 is switched into the system to feed the utility line power side of the system in preference to feeding through the UPS in order to maintain redundant feeds to the racks w/ EDP equipment).
  • the local generator 1820 sags under the large load suddenly placed on it and may respond by increasing its throttle setting.
  • the local generator 1820 overshoots the high voltage cutoff value of the micro-ATS 700 units, causing them to switch back to UPS power, thereby removing the load from the local generator 1820.
  • the local generator 1820 then throttles back and its output voltage returns to normal levels.
  • the micro-ATS 700 units switch back to the generator, causing it to sag again.
  • the sag and return of the local generator 1820 throttle can repeat, causing a harmonic reinforcing power load surge to build up and destabilize the local generator 1820.
  • implementation of a power distribution topology using micro-ATSs 700 involves a number of features, particularly when the implementation is to include safe, reliable, and economical use of filtered utility line power.
  • These features include input line power voltage range control, auto-switching of single power cord EDP devices, auto-switching of dual (or N+l) power supply EDP devices, prevention of harmonic reinforcing load surges.
  • These features can be realized by auto-switching at the device or near-device level in the power distribution topology.
  • to realize these features are examples of input line power voltage range control, auto-switching of single power cord EDP devices, auto-switching of dual (or N+l) power supply EDP devices, prevention of harmonic reinforcing load surges.
  • micro-ATSs 700 are used having a number of characteristics. It will be appreciated that the micro-ATS 700 embodiments described above manifest these characteristics either by virtue of their design as a single micro-ATS 700 embodiment or when combined into a highly parallel ATS embodiment (e.g., as part of a parallel micro- ATS module 2200).
  • Embodiments of the micro-ATS 700 prefer and select the primary power source (e.g., "A" power 760) when it is available and of sufficient quality. For example, maximum efficiency may be realized by ensuring that, if utility line power is available and of sufficient quality, it is being used to power all loads.
  • the primary power source e.g., "A" power 760
  • Embodiments of the micro-ATS 700 also protect against out-of-range voltage conditions on the primary power source and switch to the secondary power source if the primary power source is out of range (and switch back to the primary source when it returns to the acceptable range and is stable). It is also desirable that the micro-ATS 700 switch to the secondary power source as a precaution when other primary power source issues are detected. Some embodiments may not manifest this characteristic, as they may exploit the fact that modern power supplies are relatively immune to power quality issues other than input voltage range.
  • embodiments of the micro-ATS 700 transfer within the CBEMA 20- millisecond limit in both directions (i.e., from primary to secondary power and from secondary to primary power). In fact, some embodiments switch from "A" power 760 to "B" power 765 in 14 - 16 milliseconds. While circuit embodiments described above can be configured to achieve even faster switching times, switching times are selected also to maximize rejection of false conditions that could initiate a transfer. In some embodiments, transfer times between "B" power 765 and "A" power 760 are approximately 5 milliseconds once initiated (e.g., after a delay, as described below).
  • this transfer timing can be achieved because most "B” power 765 to "A” power 760 transfers occur after "A" power 760 has returned to an operational condition, so that the micro-ATS 700 can select the time to make the transfer when both power sources are up and running.
  • Embodiments of the micro-ATS 700 also incorporate a delay factor in secondary to primary power switching (except if the secondary power source fails).
  • the delay factor chosen is sufficient to allow modern generators to stabilize their throttle settings and not oscillate.
  • the time is selected to be outside of the normal response time characteristics of most typical generators to prevent harmonic reinforcing load surges by allowing the generator time to adapt to the load change and to stabilize its output.
  • embodiments of the micro-ATSs 700 are configured to use minimal or no valuable rack space that could be used for EDP equipment 1885.
  • embodiments mount in a "zero-U" fashion or are otherwise integrated into or near the rack without using rack space.
  • Certain embodiments are integrated into EDP equipment 1885 directly.
  • Other embodiments are integrated into plugstrips 1880 or in-rack or near- rack power distribution units such as the Zonit Power Distribution Unit (ZPDU). These embodiments may trade a minimal amount of rack space usage against access at the rack to the circuit breakers controlling power to the plugstrips 1880 in the racks.
  • the ATS function is integrated into every sub-branch output of the local power distribution unit, so that each one is auto-switched. This can be a worthwhile trade-off to some data center managers.
  • micro-ATS 700 when deployed as parallel micro- ATS modules 2200 also "spread" the load on the source being transferred.
  • each micro-ATS 700 has a small degree of variability in its timing of transfers from "B" power 765 to "A" power 760 with respect to other micro-ATS 700 (e.g., as an artifact of the manufacturing process). The variability is not large in real time but is significant in electrical event time.
  • the variance When running the micro-ATSs 700 as components of a parallel micro-ATS modules 2200, the variance "spreads" the load being transferred as seen by the power source, for example, a generator or UPS unit.
  • the load appears to the power source as a large number of micro-ATS 700 transfers over a time window. This can be beneficial to generators and UPS units, since it distributes a large number of smaller loads over a period of time, thereby reducing instantaneous inrush.
  • Embodiments of the micro-ATS 700 are also highly efficient, reliable, and inexpensive. When deploying micro-ATS 700 units at the device level, there will be a large number of them, and they must therefore be efficient enough to offset the cost of their purchase. Embodiments use less than 100 milliwatts when on the primary power source in normal operational mode. In a related sense, they must be highly reliable and otherwise inexpensive to purchase.
  • features of the micro-ATS 700 allow deployment of device-level ATS functionality that is reliable, efficient, cost-effective, and of minimal impact to rack space usage.
  • a population of highly reliable ATS units at the device level can produce much higher per-device power reliability levels than an ATS that switches a branch circuit or an entire panelboard can.
  • the chance that all micro-ATSs 700 will fail at the same time and affect all their connected, auto-switched EDP devices is infinitesimal as compared to the chance that a single ATS deployed closer to the root of the power distribution topology will fail.
  • micro-ATS 700 units in parallel micro- ATS modules 2200 also allows for a relatively fast transfer time because each component ATS has smaller relay contacts faster transfer times. It will be further appreciated that the circuit embodiments described above allow the micro-ATS 700 to be more efficient than many traditional ATS unit designs of the same power handling capacity by a factor of 10 or more. It is worth noting that use of traditional ATS designs in a highly parallel configuration may not be practical for at least this reason, as the net result would likely be to consume more power rather than less power, regardless of the capital expense of the switching units used.
  • micro-ATS 700 embodiments are highly cost-effective, due, for example, to relatively low manufacturing costs, relatively long service life, etc.
  • a related concern may be preserving rack space wherever possible.
  • space in a data center equipment rack or cabinet can be very expensive, and embodiments of the micro-ATS 700 can be deployed to consume little or no rack space (e.g., by being integrated into the rack structure outside of the volume in the rack where EDP equipment is mounted).
  • rack space e.g., by being integrated into the rack structure outside of the volume in the rack where EDP equipment is mounted.
  • Each server may use around 3 - 6 watts of 120-volt power, such that a 15-amp ATS can handle 2 - 4 servers.
  • the ATS units are 1U rack-mounted devices, using one ATS for every 3 servers would still consume 25% of the rack space. This may typically be too inefficient a use of expensive rack space to be practical.
  • use of a parallel micro-ATS module 2200 would consume much less rack space (e.g., even a rackmounted implementation typically uses a single rack space).
  • micro-ATSs 700 can also increase efficiency for traditional power distribution via UPS load shifting.
  • loads when double-conversion UPS units are used in a data center as the power sources (e.g., as illustrated in FIG. 19). This is usually due to the nature of the end use equipment having dual power supplies that distribute the load more or less equally to both the "A" and "B" power supply inputs. Also, as discussed above, this can reduce UPS efficiency, since they are typically loaded at below 50% to provide fully redundant power.
  • micro-ATSs 700 Using large numbers of micro-ATSs 700, it is possible to raise the efficiency of such a power distribution system. All of the electrical load for EDP equipment 1885 in the data center can be "load shifted" via micro-ATS 700 units onto one of the two UPS units, increasing the efficiency of that UPS unit (e.g., as shown in the UPS efficiency curve shown in FIG. 20).
  • the other UPS unit is at idle and will only be used if the primary unit fails.
  • the UPS units must be designed to handle this type of load being immediately placed on them, but almost all modern UPS units can do this. The result tends to be an increase of approximately 3 - 5% in the efficiency of the data center.
  • Implementations can be deployed incrementally for each piece of EDP equipment 1885, which can reduce service impacts.
  • every power supply (or corded) EDP device will be connected via a micro-ATS 700 to the "A" and "B" UPS units.
  • Every dual or N+l power supply EDP device will have one power supply connected to the "A" UPS via a normal power cord, and the second or all other N+l power supplies will be connected to the "A" and “B” UPS units via micro-ATS 700 units. Accordingly, when the "A" UPS unit is available, it takes all the load; and when it is not, the "B" UPS carries the load.
  • the micro-ATS 700 units can be deployed at a per-device ratio, or at a ratio of one micro-ATS 700 to a low integer number of EDP devices (e.g., respecting micro-ATS 700 power capacity limits).
  • UPS units While only one pair of UPS units is discussed here the methodology scales to larger data centers that have many UPS units deployed in pairs for redundancy. Similarly, the illustrative embodiments discussed herein show limited configuration options for the sake of clarity. It will be appreciated that embodiments can be adapted to any number of equipment racks, EDP equipment, ATSs, panelboards, plugstrips, etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Power Engineering (AREA)
  • Business, Economics & Management (AREA)
  • Emergency Management (AREA)
  • Power Sources (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

La présente invention concerne des systèmes et des procédés de distribution de puissance redondante et fiable. Certains modes de réalisation comprennent des micro-commutateurs de transfert automatique (micro-ATS), incluant divers composants et diverses techniques pour faciliter une fonctionnalité de commutation automatique fiable dans un petit encombrement (par exemple, inférieure à 10 pouces cubiques, avec au moins une dimension inférieure à une hauteur de bâti aux normes NEMA). D'autres modes de réalisation comprennent des systèmes et des techniques pour intégrer un certain nombre de micro-ATS dans un module à commutation automatique parallèle destiné à une distribution de puissance redondante vers un certain nombre de dispositifs. Des mises en œuvre du module à commutation automatique parallèle sont configurées pour être montées soit à l'intérieur, soit au-dessus, soit sur le côté de bâtis d'équipement normalisés. D'autres modes de réalisation encore proposent des topologies de distribution de puissance qui exploitent la fonctionnalité des micro-ATS et/ou des modules de micro-ATS parallèles.
PCT/US2011/047496 2010-08-11 2011-08-11 Distribution de puissance redondante parallèle WO2012021759A2 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CA2807996A CA2807996C (fr) 2010-08-11 2011-08-11 Distribution de puissance redondante parallele

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
US37275610P 2010-08-11 2010-08-11
US37275210P 2010-08-11 2010-08-11
US61/372,756 2010-08-11
US61/372,752 2010-08-11

Publications (2)

Publication Number Publication Date
WO2012021759A2 true WO2012021759A2 (fr) 2012-02-16
WO2012021759A3 WO2012021759A3 (fr) 2012-05-24

Family

ID=45568203

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2011/047496 WO2012021759A2 (fr) 2010-08-11 2011-08-11 Distribution de puissance redondante parallèle

Country Status (2)

Country Link
CA (1) CA2807996C (fr)
WO (1) WO2012021759A2 (fr)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2015148686A1 (fr) 2014-03-25 2015-10-01 Zonit Structured Solutions, Llc Commutateur de transfert automatique pour barres blindées de puissance
US9430318B2 (en) 2014-01-27 2016-08-30 Microsoft Technology Licensing, Llc Health monitoring and recovery for infrastructure devices
US9467006B2 (en) 2013-09-23 2016-10-11 Trippe Manufacturing Company Automatic transfer switch for three-phase applications
CN106415514A (zh) * 2014-04-03 2017-02-15 菲尼克斯电气公司 一种电接口模块
CN106464004A (zh) * 2014-05-15 2017-02-22 亚马逊科技公司 灵活电力支持冗余母线槽系统
CN106627668A (zh) * 2016-12-26 2017-05-10 合肥工大高科信息科技股份有限公司 基于二乘二取二架构的列车监控服务器系统及控制方法
CN108767846A (zh) * 2018-04-27 2018-11-06 天津市百利电气有限公司 双电源转换装置
US10615578B2 (en) 2016-07-29 2020-04-07 Karle Innovations Ltd. Electrical equipment with additional compartment and wiring to account for temperature limitations of connected conductors

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040076148A1 (en) * 2002-10-21 2004-04-22 Thomas Ferry Method and apparatus for automatic transfer switch
US20070052294A1 (en) * 2002-11-01 2007-03-08 Rudy Kraus Apparatus for providing high quality power
US20080317021A1 (en) * 2007-06-21 2008-12-25 American Power Conversion Corporation Method and system for determining physical location of equipment
US20100141038A1 (en) * 2007-03-14 2010-06-10 Zonit Structured Solutions, Llc Automatic transfer switch module

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20040076148A1 (en) * 2002-10-21 2004-04-22 Thomas Ferry Method and apparatus for automatic transfer switch
US20070052294A1 (en) * 2002-11-01 2007-03-08 Rudy Kraus Apparatus for providing high quality power
US20100141038A1 (en) * 2007-03-14 2010-06-10 Zonit Structured Solutions, Llc Automatic transfer switch module
US20080317021A1 (en) * 2007-06-21 2008-12-25 American Power Conversion Corporation Method and system for determining physical location of equipment

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9467006B2 (en) 2013-09-23 2016-10-11 Trippe Manufacturing Company Automatic transfer switch for three-phase applications
US9430318B2 (en) 2014-01-27 2016-08-30 Microsoft Technology Licensing, Llc Health monitoring and recovery for infrastructure devices
WO2015148686A1 (fr) 2014-03-25 2015-10-01 Zonit Structured Solutions, Llc Commutateur de transfert automatique pour barres blindées de puissance
CN106662904A (zh) * 2014-03-25 2017-05-10 佐尼特结构解决方案有限责任公司 用于功率母线槽的自动转换开关
EP3123274A4 (fr) * 2014-03-25 2017-09-13 Zonit Structured Solutions, LLC Commutateur de transfert automatique pour barres blindées de puissance
AU2015236051B2 (en) * 2014-03-25 2020-10-08 Zonit Structured Solutions, Llc Automatic transfer switch for power busways
CN106415514A (zh) * 2014-04-03 2017-02-15 菲尼克斯电气公司 一种电接口模块
CN106464004A (zh) * 2014-05-15 2017-02-22 亚马逊科技公司 灵活电力支持冗余母线槽系统
CN106464004B (zh) * 2014-05-15 2019-08-06 亚马逊科技公司 灵活电力支持冗余母线槽系统
US10615578B2 (en) 2016-07-29 2020-04-07 Karle Innovations Ltd. Electrical equipment with additional compartment and wiring to account for temperature limitations of connected conductors
CN106627668A (zh) * 2016-12-26 2017-05-10 合肥工大高科信息科技股份有限公司 基于二乘二取二架构的列车监控服务器系统及控制方法
CN108767846A (zh) * 2018-04-27 2018-11-06 天津市百利电气有限公司 双电源转换装置

Also Published As

Publication number Publication date
CA2807996C (fr) 2017-08-08
WO2012021759A3 (fr) 2012-05-24
CA2807996A1 (fr) 2012-02-16

Similar Documents

Publication Publication Date Title
US10474220B2 (en) Parallel redundant power distribution
US20210344194A1 (en) Power distribution using hydra cable systems
US9996128B2 (en) Parallel redundant power distribution
EP2962381B1 (fr) Distribution de courant redondante parallèle
EP3123274B1 (fr) Commutateur de transfert automatique pour barres blindées de puissance
CA2807996C (fr) Distribution de puissance redondante parallele
US10928878B2 (en) Reserve power system for data center
US20170244274A1 (en) Transfer switch with arc suppression
US9081568B1 (en) Electrical power system with automatic transfer switch failure protection
WO2014146776A1 (fr) Ensemble appareillage de commutation, système de distribution de puissance à courant continu (cc), carte de distribution de puissance cc
US20240155801A1 (en) Power distribution using hydra cable systems
TW201225480A (en) Power supply of data center
US11543870B2 (en) Power system and operation method thereof
Duguay System design concepts for a compact power system featuring 250 W rectifiers
CN103683475A (zh) 一种大功率模块化ups电源
KR20120038760A (ko) 스마트 다중전원 절체장치

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 11817072

Country of ref document: EP

Kind code of ref document: A2

ENP Entry into the national phase in:

Ref document number: 2807996

Country of ref document: CA

NENP Non-entry into the national phase in:

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 29-05-2013)

122 Ep: pct application non-entry in european phase

Ref document number: 11817072

Country of ref document: EP

Kind code of ref document: A2