GB2479183A

GB2479183A - Rebreather apparatus having a sol-gel oxygen sensor

Info

Publication number: GB2479183A
Application number: GB1005498A
Authority: GB
Inventors: Alexander Roger Deas; David Coyne
Original assignee: Individual
Current assignee: Individual
Priority date: 2010-04-01
Filing date: 2010-04-01
Publication date: 2011-10-05
Also published as: GB201005498D0

Abstract

A technique and methods for a low oxygen alarm system for a rebreather wherein a sol-gel derived film oxygen sensor is used as an accurate oxygen level sensor within a breathing loop. An example embodiment is described using a fluorescence decay measurement channel containing a ruthenium sot-gel oxygen sensor with a variable bandwidth low pass filter producing a matched filter receiver for the light emission to maintain the signal to noise ratio over a wide range of lifetime values ensuring accurate measurement. The measurement channel includes amplitude control, offset control and bandwidth control. A controller adapts the measurement channel bandwidth, adapts the measurement channel timing to minimise measurement time and power consumption. The controller further minimises power consumption by adapting the measurement channel timing and the bias current of the light excitation source.

Description

INTELLECTUAL

. .... PROPERTY OFFICE Application No. GB 1005498.9 RTM Date:26 July 2010 The following terms are registered trademarks and should be read as such wherever they occur in this document: Teledyne OceanOptics Intellectual Property Office is an operating name of the Patent Office www.ipo.gov.uk

TITLE OF INVENTION

A sol-gel sensor low oxygen alarm

BACKGROUND OF THE INVENTION

Technical Field

The present invention relates to the measurement of the oxygen and provision of a low oxygen alarm system. Specifically, the present invention comprises a technique and methods for measurement of the oxygen levels in a breathing gas within confined spaces or in a rebreather using a sal-gel oxygen sensor. Problems of calibration and drift of sol-gel sensors are addressed.

Further, the present invention comprises methods and techniques for the measurement of the fluorescence decay lifetime of a sol-gel gas sensor deployed in an oxygen monitoring system.

1 5 Backcjround of the Invention Closed circuit rebreathers are employed in many applications such as the military, fire fighting, commercial diving and sports diving. At the heart of each rebreather is a gas measurement system for monitoring the amount of oxygen, as well as other gases, within the breathing loop and allowing adjustment of the oxygen either manually or automatically to preserve the users breathing. Within the gas measurement system are a number of sensors including one or more oxygen sensors. Further, associated with each sensor or each sensor type is a measurement channel that interfaces to the sensor or sensors to determine the constituent components of the gas within the breathing loop.

Commonly used oxygen sensors for rebreather products are based on galvanic micro-fuel cells. Oxygen sensors manufactured by Analytical Industries and Teledyne are often used in rebreather products. These sensors contain an anode, cathode and an electrolyte generally aqueous potassium hydroxide. A membrane isolates the gas from the electrolyte. Oxygen passes through the membrane and causes a chemical reaction in the electrolyte and anode/cathode whereby electrons are generated. With an external load these electrons flow from the cathode to the anode and form a current and hence a voltage across the external load. The amount of oxygen per unit volume of the gas is directly related to the output voltage. Accordingly the amount of oxygen in the gas and the pressure of the gas both affect the output voltage in a linear manner. Further, the output voltage is also a function of temperature. Calibration is required at a known oxygen level, temperature and pressure to obtain subsequent accurate measurements.

A measurement channel for a galvanic oxygen sensor may include one or more analogue amplifiers, signal processing elements such as bandwidth limiting filters, an analogue to digital converter, a microcontroller and a digital display.

Typically the maximum output voltage from a galvanic oxygen sensor is no more than a few tens of millivolts and although direct measurement of such a voltage by an analogue to digital converter is possible it is not unusual to include some form of amplification of the signal before conversion to digital form by an analogue to digital converter.

1 5 Some issues exist with galvanic oxygen sensors for example: -The output voltage from a galvanic oxygen sensor is linear but it is still dependant temperature. Temperature sensors that match the temperature response of the sensor are required to obtain accurate measurements.

-Calibration of the oxygen sensor is required using a known oxygen concentration. As the galvanic sensor is linear and develops zero volts output in the absence of oxygen only a single point calibration scheme is necessary. Typically air is used as the partial pressure of air at sea level is almost constant all over the world.

-In a rebreather application it is necessary that the sensor respond to changes in oxygen partial pressure in a few seconds. Faster response means a thinner membrane.

-Mechanical robustness. Sensors with a higher output voltage and faster response time use thinner membranes to allow the diffusion of oxygen into the electrolyte. Shock can damage the membrane resulting in leakage of the electrolyte which is caustic and can cause subsequent damage to other components in the vicinity.

-The material in the electrodes of a galvanic sensor is consumed by the presence of oxygen. Typically, sensors must be replaced every 12 months or more frequently if used in oxygen-rich atmospheres.

-The use of a single type of sensor is contrary to good safety design practice where diverse technologies should be used to sense critical parameters, such as the partial pressure of oxygen.

Even with these issues the galvanic oxygen sensor is the sole oxygen sensor in rebreathers today One emerging type of oxygen sensor is based on sol-gel derived film technology which has the diversity benefits of being formed from a totally orthogonal technology and uses different measurement techniques and therefore a different measurement channel to galvanic oxygen sensors. It would be beneficial to use such a sensor in a safety critical system such as a rebreather as the orthogonal technology would improve diversity.

Over the last 10 to 15 years much work has been published in the field of oxygen sensors based on the properties of sol-gel materials. It has been known for many decades that certain materials can fluoresce when illuminated under light of a certain wavelength. Some materials when doped or injected with dyes or other materials may emit light under illumination and respond to the oxygen concentration through a relationship between the intensity of the emitted light or the fluorescence decay lifetime time constant.

A commonly used dye material in sol-gel oxygen sensors is a fluorescence ruthenium complex dye that is trapped in a micro-porous support matrix through which smaller analyte species e.g. oxygen may diffuse and interact. Under suitable illumination fluorophores may be excited into a higher energy state. Some of these fluorophores decay back to the base state with the emission of heat while others decay back to the base state with the emission of a photon. The emitted photon is always at a longer wavelength than the wavelength used to generate excited states in the material and the amount by which the wavelength change is known as the Stokes shift. Some of the published work on sal-gel sensors includes: -Detection of oxygen in gas [Enhanced Fluorescence Sensing Using Sol-Gel Materials, B. D. MacCraith and C. McDonagh, Journal of Fluorescence, Vol. 12, Nos. 3/4, December 2002] -Detection of dissolved oxygen in water [Phase Fluorometric Dissolved Oxygen Sensor, C. McDonagh et al, Sensors and Actuators B, Vol. 74, 2001, p124-130] -Detection of carbon dioxide in gas [Lifetime-based Optical Sensor for High-level P002 Detection Employing Fluorescence Resonance Energy Transfer, C von Bültzingslöwen et al, Anaytica Chimca Acta, 480, 2003, p275-283] -Detection of temperature [Temperature-corrected Pressure-sensitive Paint Measurements Using a Single Camera and a Dual-lifetime Approach, J. Hradil et al, Measurement Science and Technology, Vol. 13, 2002, p1552-1557] -Detection of pH [Novel Sol-Gel Derived Films for Luminescence-based Oxygen and pH Sensing, D. Wencel et al, Materials Science-Poland, Vol. 25, No.3,2007, p767-779] The sol-gel based oxygen sensors are not without issues. Some of the issues are: -Stability and life of the sensor. The large-scale manufacturing of sal-gel sensors is relatively recent and lifetimes are now being reported of 12 months or more. This is comparable to the lifetime of a galvanic oxygen sensor.

-Photo-bleaching may occur if the sol-gel material is exposed to too high a level of illumination. It is beneficial to be able to operate a sol-gel sensor at a low illumination level.

-The emission from the sol-gel sensor is not a linear function. Typically a sol-gel oxygen sensor is most sensitive at low oxygen levels and less sensitive at high oxygen concentrations. This, plus saturation of the fluorescence intensity at high oxygen levels precludes the use of a sal-gel sensor in an application required to operate over a wide range of oxygen partial pressures.

-Calibration of a sol-gel oxygen sensor may require more than two known gas concentrations to provide a wide measurement range. Operation at low oxygen levels and a small range may be possible with single-point calibration.

It can be seen that some of the issues present in a galvanic oxygen sensor are also present in sol-gel based oxygen sensors.

An example of a company producing commercially available oxygen sensors is OceanOptics Inc. OceanOptics Inc produces a range of oxygen sensors based on two manufacturing methods: -sol-gel material deposited on the end of fibre optic -sol-gel material deposited on adhesive patches or glass Sensors based on the first manufacturing method are typically employed in the scientific and medical markets while sensors based on the second manufacturing method may be additionally employed in low-cost food packaging gas analysis markets for example. In addition to sensors measurement systems are also available, typically based on the use of a spectrometer. Due to cost, size and power consumption the use of a spectrometer in a rebreather product is not preferable.

1 5 Much work has concentrated on developing materials that fluoresce with an intensity related to the absence of oxygen: it is similar to photo-fluorescence of plankton when disturbed in water. These physics make sol-gels unsuitable for oxygen measurement in rebreathers because when the FF02 is more than one atmosphere, the sensor saturates. Rebreathers typically operate with FPO2s of 1.3. This means that currently available sol-gel sensors cannot be used to control oxygen levels in rebreathers.

However, although currently available sol-gel sensors may not be suitable for measurement of oxygen over a wide range of oxygen partial pressures, a sol-gel sensor may be employed at low oxygen levels where it has increased sensitivity. Thus it is beneficial to use a sol-gel sensor in an application where low levels of oxygen are being monitored. Low oxygen levels is the biggest single safety risk in a rebreather system.

In a breathing apparatus lack of oxygen, known as hypoxia, is one of the greatest problems a user can face. It occurs when oxygen levels are critically low, whereupon blackouts can occur without any warning signs. A sensor with high sensitivity to low oxygen levels is appropriate to use in a system for the detection of hypoxia, and to provide diversity to the oxygen measurement system. It is beneficial to use a sol-gel sensor in an oxygen hypoxia detection measurement system.

As the present invention is concerned with the use of sol-gel sensors it is appropriate to review some of the most commonly used techniques for monitoring the oxygen concentration when a sol-gel sensor is deployed in an oxygen measurement system.

Many references have discussed the physics and equations governing the fluorescence emission which is a well understood process. The intensity of the emitted radiation is a function of the presence of oxygen. Additionally the rate of decay of the intensity is a function of the presence of oxygen. Due to issues with stability of the illumination source it is not usual to use intensity measurements to determine the oxygen concentration. It is more common to measure the rate of decay of the intensity. The emission intensity from a sol-gel sensor typically decays with a mono-exponential decay time constant.

A typical sol-gel oxygen sensor measurement system comprises a light source, a sensor with a sol-gel derived film, a light detection system and a measurement channel. Depending on the techniques used in the measurement channel there may be additional components involved such as optical filters, optical waveguides and lenses. In a low-cost system the removal of as many mechanical and optical components as possible is preferable.

One method of measuring the fluorescence decay lifetime is to make a direct measurement of the decay lifetime (Wickersheim, K. A. and Sun, M. H. (1987). Fiberoptic thermometry and its applications, Journal of Microwave Power 22 (2): 85-94). A short pulse is used to illuminate the sol-gel sensor. When the illumination is turned off a first measurement is made at a fixed time after the turning off of the illumination during the period where fluorescence light emission is taking place. The time at which the measurement is taken is noted. The next step is to calculate a level based on the measured intensity level, often the original intensity divided by "e". The time at which the emitted light intensity passes through this calculated level is then determined using a comparator. The fluorescence decay lifetime is then the difference in the two recorded times. At very low intensity levels this process must be repeated over many illumination cycles due to the probabilistic nature of the emission process. This method is also susceptible to DC offsets and requires that other measurements are taken to eliminate such offsets. The other drawback to this method is that it is mainly applicable to sol-gel sensors that have relatively long decay time constants. As the decay lifetime decreases it becomes necessary to use higher speed analogue to digital converters, digital to analogue converters and high speed processing. One advantage of this technique is that it can be implemented with minimal optical filters reducing cost.

Another method of measuring the fluorescence decay lifetime is to sample the decay curve and apply a curve fitting procedure to the measured intensity levels to determine the decay lifetime (described in Optoelectronic Fibre-Based Fluorescence Detection Sensor Systems, Vladislav A. Kuzminskiy, PhD Thesis, Stony Brook University, May 2007 and attributed to Sun). For short emission decay time constants this method requires a significant amount of high-speed analogue circuitry and much processing of the data and is potentially has a slow response. A variation on this last method is to make one measurement per decay cycle varying the measurement point over many cycles. This method reduces, but does not eliminate, the high-speed processing hardware required at the expense of longer processing time.

To overcome some of the problems with the methods already mentioned integration of the emission signal is often employed in the measurement channel.

Such a system is proposed in US patent 5,039,219 which employs two equal periods where the waveform is integrated allowing removal of DC offsets. This method is not simple to implement when the decay lifetime is short.

Another method of measuring the fluorescence decay lifetime is single photon counting systems. This method is very applicable to instances where the fluorescence sample is susceptible to photo-bleaching i.e. degradation of the sample under strong illumination. This technique invariably uses a light receiver based on a photo multiplier tube (PMT) which may even require cooling to low temperatures to improve the signal to noise ratio. This method also requires counting and averaging of measured lifetimes over many cycles. Clearly this technique is power-hungry, bulky, expensive and not suitable for portable applications such as rebreathers.

Another method of measuring the fluorescence decay lifetime is to apply light with sinusoidal modulation to the fluorescence sample, see for example US patent application 20030099574, US patent 6157037 and US patent 4845368. The number of fluorophores excited by the light is also modulated and emission of light will occur at the same frequency as the light source but with a delay dependant on the fluorescence decay lifetime. Typically it is necessary to find an optimum operating frequency for the illumination source. Increasing the frequency increases the phase shift between the illumination and emitted light but also reduces the signal to noise ratio. Coping with a wide variation in lifetime requires the modulation source to be capable of variable frequency. To obtain an accurate measurement lengthy illumination periods may be required increasing power consumption. Typically this is the techniques used with optical fibre measurement systems which, in themselves, have a low efficiency in the light generation and collection process resulting in higher power consumption. This particular system is also employed with an optical system that employs optical filters which adds to the cost of the complete system.

Each foregoing prior art measurement method has one or more of the following drawbacks: DC offset control; high power consumption; high cost; large physical size or accuracy to the implementation of an oxygen sensor measurement system for a rebreather application.

Thus, it has been shown that it would be highly advantageous in a rebreather to have a means of detecting the onset of hypoxia, wherein it would be advantageous to have an oxygen sensor that is sensitive to low levels of oxygen wherein it would be advantageous to have an oxygen sensor and measurement system that has a low power consumption, small physical size, simple calibration and fast response to changes in oxygen levels wherein it would be advantageous to use a sol-gel derived film oxygen sensor wherein it has would be advantageous to have a sol-gel derived film sensor where low illumination levels were used.

Further, it would be highly advantageous to have a sol-gel derived film oxygen sensor technologically different to the commonly used oxygen sensors to aid in diversity.

Obiect of the present invention.

It is a primary objective of the present invention to provide a rebreather low oxygen level or hypoxia monitor.

It is a primary objective of the present invention to provide a rebreather with a low oxygen level or hypoxia monitor with alarm.

It is a primary objective of the present invention to provide a rebreather with a low oxygen level monitor or hypoxia using a sol-gel oxygen sensor.

It is further object of the present invention to provide a sol-gel oxygen sensor measurement system.

It is further object of the present invention to minimise the power consumption of the sol-gel oxygen sensor measurement system.

It is a further objective of the present invention to minimise the intensity of the excitation illumination to reduce the possibility of photo-bleaching.

It is a primary objective of the present invention to provide a lifetime decay measurement method for sol-gel sensors.

A particular form of the invention is suitable for use in low oxygen alarm measurement systems in safety critical breathing apparatus.

BRIEF SUMMARY OF THE INVENTION

The present invention relates to a system and method that uses a sol-gel sensor to detect low oxygen or hypoxia conditions in a rebreather and provide an alarm state.

The preferred embodiment uses a sal-gel sensor and measurement channel to detect the low oxygen levels within a gas mixture wherein a light excitation source illuminates the sol-gel sensor; a light collection detector picks up the light emitted from the sal-gel sensor; a first amplifier converts the current from the light collection detector to a voltage; a variable gain amplifier in conjunction with an averaging peak detector is used to increased the emission signal further maintaining the peak amplitude of the emission signal constant; a DC offset removal circuit is used to set the common mode voltage of the emission signal; a low pass filter with a variable bandwidth in conjunction with a controller is used to set the bandwidth of the low pass filter to a value relative to the measured emission decay lifetime of the emission signal dependant on the relative position of one pole or, preferably, more than one pole; a first comparator is used to detect when the intensity of the emission signal falls below a first threshold reference level; a second comparator is used to detect when the emission signal falls below a second threshold level; a controller is used to filter the comparator outputs and measure the emission lifetime decay time constant and thereby the oxygen concentration; a controller further used to set an alarm signal when the oxygen concentration falls below a certain level.

BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

For a better understanding of the present invention and the advantages thereof and to show how the same may be carried into effect, reference will now be made, by way of example, without loss of generality to the accompanying drawings in which: Figure 1 shows a block diagram of a rebreather gas measurement and monitoring system with multiple oxygen sensors.

Figure 2 shows a block diagram of a rebreather gas monitoring system in the present invention employing a sol-gel oxygen sensor for measurement of low oxygen levels and alarm annunciation where the measured oxygen level falls below a threshold.

Figure 3 shows a timing diagram of the basic timing of the present invention.

Figure 4 shows a timing diagram with the definitions of the intensity levels, voltage levels and time measurement signals as an aid to the understanding to the present invention.

Figure 5 shows a circuit diagram of a transimpedance amplifier as used in the present invention.

Figure 6 shows a circuit diagram of the first embodiment of the variable gain amplifier and the peak detector in the present invention.

Figure 7 shows a circuit diagram of the second embodiment of the variable gain amplifier and peak detector in the present invention.

Figure 8 shows a circuit diagram of the DC restore circuit in the present invention.

Figure 9 shows waveforms from modelling and simulation of a fluorescence emission event of different lifetimes passed through a fixed bandwidth filter.

Figure 10 shows a circuit diagram of the variable bandwidth low pass filter in the present invention.

1 0 Figure 11 shows a circuit diagram of the implementation of the offset cancelling comparators in the present invention.

Figure 1 2 shows a circuit diagram of the reference level generators in the present invention.

Figure 13 shows a circuit diagram of the controller producing timing signals and determining the time between comparator input signals.

Figure 14 shows a flow diagram of the controller low pass filter calibration routine in the present invention.

Figure 15 shows a flow diagram of the controller measurement routine in the present invention.

DETAILED DESCRIPTION OF THE INVENTION

Figure 1 shows a circuit diagram of a rebreather gas monitoring and measurement system. Channel 200 in Figure 1 includes two types of oxygen sensor, oxygen sensor 400 and oxygen sensor 300, each sensor having a separate measurement channels, oxygen sensor 400 being technologically different to oxygen sensor 300 as are the separate measurement channels.

Oxygen sensor 400 is an oxygen sensor based on the aforementioned galvanic micro-fuel cells with a measurement channel 401 comprising an analogue to digital converter. The output of measurement channel 401 is passed to controller 500 for further processing. Oxygen sensor 400 provides the main oxygen sensing function for the maintenance of the rebreather breathing loop.

Oxygen sensor 300 is based on a different technology to oxygen sensor 400 and is a sol-gel oxygen sensor. Oxygen sensor 300 connects to measurement channel 301 where the signal from oxygen sensor 300 is processed and conditioned before passing to controller 500 for further processing.

To complete channel 200 additional sensors are employed. Third sensor 410 measures ambient temperature while fourth sensor 420 measures ambient pressure, said sensors required to complete the calibration of oxygen sensor 300 and oxygen sensor 400. Sensor 410 and sensor 420 connect to measurement channel 401 where their outputs are digitised and passed to controller 500.

A plurality of channels such as channel 200 depicted in Figure 1 are used in a complete system to provide redundancy with a bi-directional communications link2lO.

Figure 2 shows a diagram of the present invention, a sal-gel oxygen sensor and measurement channel with low oxygen level detection and alarm annunciation comprising: oxygen sensor 400; measurement channel 401 for oxygen sensor 400; sensor 410; sensor 420; oxygen sensor 300 based on a sol-gel derived film oxygen sensor; measurement channel 301 for oxygen sensor 300; and controller 500.

Figure 3 shows the timing signals associated with the control of second measurement channel 301 while Figure 4 shows an ideal emission waveform and the voltage and time definitions respectively Oxygen sensor 300 comprises excitation light source 302 depicted by means of example as a light emitting diode; excitation light source drive circuit 308; sal-gel derived film material 304; one or more optical components 303 to guide light from excitation light source 302 to sol-gel derived film material 304; light collection device 306 and optical component 305 to guide emitted light from sol-gel sensor derived film material 304 to light collection device 306 depicted by means of example as a photo-diode.

Excitation light source 302 is turned on and off with signal 510 from controller 500. Additionally controller 500 sets the bias current of excitation light source 302 with signal 511, said excitation light intensity preferably related to the bias current in a linear and monotonic manner.

In the preferred embodiment of light collection device 306, the intensity of the emission from the sal-gel derived film material 304 produces a current of a magnitude that is related to the intensity of the incident excitation light. Emission current 307 from light collection device 306 is converted to a voltage by amplifier 330. Output of amplifier 330, signal 331, is processed by variable gain amplifier (VGA) 340 wherein the gain of VGA 340 is controlled in such a manner to maintain the peak amplitude of the emitted signal at a constant level for optimal accuracy.

Output of VGA 340, signal 341, is passed through DC restore (DCR) circuit 350.

DCR 350 sets the common mode voltage of its output, signal 351, to the reference level VCM 324. Signal 351 passes through variable bandwidth low pass filter (LPF) 360 producing signal 361. The bandwidth of LPF 360 is controlled by controller 500 to match the fluorescence decay time constant of signal 351 in effect creating a matched filter receiver.

The output of LPF 360, signal 361, is used to set the gain of the measurement channel. Signal 361 is peak detected and averaged in peak detector (PKD) 370 producing signal 371. Signal 371 from PKD 370 is used to set the gain of VGA 340 either directly or indirectly by controller 500 monitoring the output of PKD 370 and setting the gain through signal 509. PKD 370 preferably averages signal 361 with a time constant longer than the measurement cycle period to provide some immunity to noisy signals. In the case of a low-level signal in the presence of large noise it is preferable to increase PKD 370 time constant to avoid sudden changes in gain and widely varying amplitude changes between cycles. By the aforementioned techniques signal 361 is controlled in amplitude, common mode voltage and bandwidth.

Signal 361 is input to comparator 380 and comparator 390. Comparator 380 and comparator 390 are used to detect when signal 361 passes through reference level V0 322 and reference level V1 323 respectively, comparator 380 produces signal 381 while comparator 390 produces signal 391.

When signal SWAP 502 is low a rising edge in signal 381 indicates that signal 361 has passed through reference level V0 322 corresponding to intensity threshold I. Alternatively, when signal SWAP 502 is high a falling edge in signal 381 indicates that signal 361 has passed through reference level V0 322 corresponding to intensity threshold I. Similarly, when SWAP 502 is low a rising edge in signal 391 indicates that signal 361 has passed through reference level V1 323 corresponding to intensity threshold I. Alternatively, when SWAP 502 is high a falling edge in signal 391 indicates that signal 361 has passed through reference level V1 323 corresponding to intensity threshold I. The time difference between rising edge events or falling edge events in signals 381 and 391, dependant on the polarity of SWAP 502, is a measure of the lifetime of the signal 361 and by association the emission signal 307. This time difference is measured by controller 500 and output as a digital code on signal 501. Further, in controller 500, the measured lifetime is compared to a reference value representative of a low oxygen level threshold. When the oxygen level in the gas mixture being measured, as represented by the measurement of the fluorescence decay lifetime, falls below a pre-defined threshold controller 500 annunciates the event by an alarm mechanism.

To avoid overload of one or more components in the measurement channel the measurement channel is disabled or blanked during the period when the excitation light source illuminates the sol-gel sample. In some systems the potential for overload of the measurement channel is avoided through using one or more optical filters to block the excitation light shorter wavelength light.

Additionally the light source is often orientated to avoid light passing directly to the light collection device. High quality optical filters can add to the cost and it is beneficial to remove these and simplify the optic system as much as possible.

In the present invention Figure 2 shows the preferred method of disabling the measurement channel by routing all current from light collection device 306 to ground when excitation light source 302 is turned on. Switch 310 in conjunction with signal 506 from controller 500 connects signal 307 to ground during the period that excitation light source 302 is turned on. To someone skilled in the art other techniques are well understood that will place the measurement channel into a well-defined state and avoid overload. For example, it is clear that there are several places where the measurement channel could be disabled depending on various factors such as the light collection device current when strongly illuminated; the measurement channel gain and the measurement channel susceptibility to overload. In one example amplifier 330 may not overload so a switch in series with the output signal of amplifier 330 and the input signal of VGA 340 may be appropriate.

A means of injecting signals into the measurement channel is also provided to aid in calibration of the measurement channel or verification of the measurement channel. Switch 311 in conjunction with control signal 507 allows a signal to be injected to the measurement channel at the input of amplifier 330. In the preferred method controller 500 acts as a signal source providing input signal 508. The present invention is now disclosed in more detail.

Amplifier 330 converts light collection device 307 current to a voltage for subsequent amplification, filtering and common mode conditioning. A suitable scheme for amplifier 330 is a transimpedance amplifier comprising operational amplifier 332 and resistor 333 is shown in Figure 5. This implementation, and others, would be obvious to someone skilled in the art.

At the input of amplifier 330 are a number of switches. Switch 310 controlled by signal 506 from controller 500 is used to connect light collection device 306 to ground during the period where excitation light source 302 is illuminating sol-gel derived film oxygen sensor 304. This ensures that the measurement channel is not overloaded by strong illumination with subsequent recovery issues. Switch 311 with control signal 507 from controller 500 allows controller 500 to inject a known signal into the measurement channel in a calibration procedure. Controller 500 generates signal 508, preferably from a digital to analogue converter with know characteristics in the time and current or voltage domains.

In the preferred method of variable gain control, as shown in Figure 6, VGA 340 may be controlled with signal 371 from PKD 370. As the average value of the peak amplitude of signal 361 increases above reference level Vp 321 an error signal is generated that is fed to VGA 340 where the gain is reduced to bring the average value of the peak amplitude of signal 361 back to reference level V 321.

As the average value of the peak amplitude of signal 361 decreases below reference level V 321 an error signal is generated that is fed to VGA 340 where the gain is increased to bring the average value of the peak amplitude of signal 361 back to reference level Vp 321.

In more detail, signal 361 is peak detected by diode 375, capacitor 376 and resistor 377. Resistor 377 forms a discharge path for capacitor 376 and allowing the peak detector to recover from a noisy signal. Reference voltage V 321 is passed through diode 373 and resistor 374 to compensate for the voltage loss in diode 375. Transconductance amplifier 372 generates an output current dependant on the difference between the signals at its input pins. This current is integrated in capacitor 378 when switch 379 is closed by control signal 505. When switch 379 is open the voltage on capacitor is held at the voltage on the capacitor the instant that the switch was opened generating the PKD output voltage, signal 371. PKD output, signal 371, is connected to the control input of VGA 340 and servos the measurement channel gain to match the average peak amplitude of signal 361 to the reference level Vp 321. This and other techniques for implementing both peak detector 370 and VGA 340 are well understood by those

in the field.

In a second method VGA 340 may be controlled indirectly by signal 371 of PKD 370 by controller 500. In this second method as shown in Figure 7 signal 371 may be converted to a digital data word by an analogue to digital converter in controller 500. Decisions are made by controller 500 to increase or decrease the gain of VGA 340 according to a pre-defined algorithm which may be made based on a deterministic or heuristic algorithm and may include filtering and optimal control, for example. Output signal 509 from controller 500 is passed to VGA 340 directly as either a digital word, or indirectly as an analogue signal after passing through a digital to analogue converter.

In the simplest form DCR 350 could be AC coupling of the signal from the output of VGA 340. However, this method results in the average signal being set to the common mode voltage on the right-side of the AC coupling capacitor rather than the flat region of signal 361 where no emission is occurring. It is preferred to set the common mode of signal 361 to the flat region where no fluorescence emission is occurring as this common mode voltage will be set relative to all other references within the measurement system.

The preferred method of setting the common mode voltage is to use a DC restoration circuit as shown in Figure 8. In this technique a know region where the signal 341 is expected to be at a constant DC level is sampled or averaged (in the case of a noisy signal) and compared to a common mode reference level generating an error signal that is used to servo the common mode voltage of the complete signal. In the preferred embodiment of the invention DCR 350 comprises of a feedback loop with transconductance amplifier 352, capacitor 354 with switch 353 and summing amplifier 355. Transconductance amplifier 352 has a first input connected to DCR output signal 351 and a second input connected to reference voltage VCM 324 producing an output current. Over a period in time defined by signal RESTORE 504, the output current from transconductance amplifier 352 is averaged by capacitor 354, switch 353 being closed. At other times switch 353 is open and the current from transconductance amplifier 352 does not pass to capacitor 354, the voltage across capacitor 354 being held relatively constant. The voltage on capacitor 354 is subtracted from input signal 341 in summing amplifier 355 setting the common mode voltage of the output signal 351. To someone practised in the art it is clear that alternative methods of setting the output common mode could be used. For example one or more common mode feedback stages could be employed in VGA 340, placed at locations in VGA 340 to avoid build-up of offset where the gain is largest.

The fluorescence lifetime is defined as the time for a population of excited fluorophores to decay to l/e of the initial population. The time for a fluorophore to emit an individual photon is governed by a statistical process. Due to this statistical process the decay curve will not be smooth. At low intensity levels and low initial population count the decay curve is far from smooth indeed.

Figure 9 shows an example of the modelling of the variation in intensity during the emission process hereafter called probabilistic noise. A theoretical emission process is modelled for a fixed fluorophore population with a 1 ps lifetime and producing the top-left trace. This signal is passed through a low pass filter of bandwidth 500KHz producing the bottom-left trace. The removal of noise is clearly evident.

In the same model the lifetime of the emission process is increased to lOps with the same fixed fluorophore population producing the top-right trace. It can be seen that the waveform is identical in amplitude but scaled in time relative to the signal with a lifetime of lps. The lOps lifetime signal is also passed through a low pass filter where the bandwidth of the low pass filter is unchanged relative to the first simulation i.e. 500KHz. These two simulation output curves are representative of what would be obtained in a fixed bandwidth measurement channel. The output of the low pass filter shown in the bottom-right trace indicates that a significant proportion of the probabilistic noise passes through the low pass filter.

In a fixed bandwidth measurement channel the bandwidth of LPF 360 must be set to reject noise from the shortest fluorescence lifetime otherwise the signal will be distorted and the measured lifetime incorrect. In using a fixed bandwidth filter where the bandwidth is set relative to the shortest expected lifetime the noise associated with a longer lifetime emission process will be lower in frequency and pass through the filter with less attenuation degrading the measurement system dynamic range.

It is the purpose of the present invention to preserve dynamic range by varying the bandwidth of the filter in an inverse relationship to the lifetime of the signal.

Preferably LPF 360 should consist of a plurality of poles for the strongest probabilistic noise and random noise attenuation. It is recognised that placement of the one or more poles at a frequency close to the reciprocal of the fluorescence decay lifetime will increase the measured lifetime value. This error is deterministic depending on the pole placement in LPF 360 and can be corrected through application of an appropriate scaling factor for a given filter pole placement.

In the present invention LPF 360 includes a dominant pole at a first frequency and a second pole at a frequency above the first pole frequency. This aids in higher frequency noise reduction which is typical of the probabilistic process of fluorescence emission. In the preferred embodiment the second pole is placed at 20X the frequency of the first pole. Preferably the second pole should track the frequency of the dominant pole. As the first pole is increased in frequency then the second pole should be increased in frequency. As the first pole is decreased in frequency then the second pole should be decreased in frequency.

In this manner the signal to probabilistic noise ratio will be constant. This is important in the present invention as it is the intention to minimise the intensity of the excitation light source, reducing the photon count to reduce power consumption.

The action of DCR 350 also reduces the noise bandwidth of the system by providing a high-pass function through which the noise and signal must both pass.

Specifically, low frequency noise and flicker noise are attenuated by the high-pass function in DCR circuit 350. The DC restore method makes the implementation of parts or all or the measurement channel in a modern low-cost CMOS process much easier with the removal of low frequency noise sources from the signal.

When a signal passes through a low pass filter higher frequency components within the signal are attenuated. In the case of a signal as obtained from fluorescence decay with a mono-exponential decay characteristic the signal is distorted. Measurement of the decay time after passing through a low pass filter will result in a lifetime value that is longer than that of the actual fluorescence 1 0 decay. In some works much effort is put into deconvolving the filtered fluorescence decay curve to obtain an accurate estimation of the fluorescence lifetime.

However, deconvolution techniques may need to employ many calculations which may not be appropriate for a low power, fast response system. Another method of obtaining an accurate lifetime measurement result in the presence of heavy filtering is required and is now disclosed with the effect of the filter bandwidth on the accuracy of the measured lifetime.

For a fixed fluorescence lifetime it can be shown that there exists a minimum low pass filter bandwidth for a given error. For a low pass filter bandwidth of BW and sufficiently large number of measurement cycles it can be shown that: -For an error of less than 0.5% the measured lifetime must be more than 1.0/BW -For an error of less than 1% the measured lifetime must be more than 0.8/BW -For an error of less than 2% the measured lifetime must be more than 0.7/BW -For an error of less than 4% the measured lifetime must be more than 0.5/BW Other lifetime values exist for every bandwidth and accuracy setting. The above values are for a system that includes a first pole at bandwidth BW and a second pole at bandwidth 20X BW.

From the above it is clear that filtering can be applied to an emission signal and an accurate measurement obtained. It is the purpose of the present invention to match the low pass filter bandwidth to the fluorescence decay lifetime. If the lifetime was known a priori in conjunction with an accuracy setting then the low pass filter bandwidth could be set to a particular value and a measurement made over a given number of cycles. However, as will be disclosed later, the lifetime is not generally known and it is necessary to make tentative lifetime measurements and adapt LPF 360 bandwidth in an algorithmic manner.

LPF 360 preferably includes a first pole and preferably a second pole, a first pole being located at a lower frequency than the second pole. Preferably the second pole is located at a frequency of 20X that of the first pole. Further, both poles are variable and can be controlled with a one or more digital control lines 503. Further, it is preferable that the ratio of the frequency of the first pole to the frequency of the second pole remains constant at all bandwidth settings.

A preferred embodiment of LPF 360 is shown in Figure 10 and comprises a first pole formed by resistor 364 and capacitor 363 and a second pole formed by resistor 367 and capacitor 366. The first pole is isolated from the second pole by buffer 365. Buffer 362 provides a low impedance drive and isolation of capacitor 363 from the following stages.

In the preferred embodiment resistor 364 and resistor 367 are variable and controlled by one or more signals 504 from controller 500. Resistor 364 and capacitor 363 form a pole of programmable frequency as does resistor 367 and capacitor 366. A preferred method of implementing the variable resistors 364 and 367 is to have a number of resistors in series, parallel or a combination of series and parallel networks with a number of switches controlled by digital signals 504 from controller 500. Another method of implementing resistors 364 and 367 is to employ one or more voltage controlled resistors formed by, for example, field effect transistors and controlled by one or more analogue signals 504 from controller 500. To someone skilled in the art other methods may be used to implement LPF 360 without detracting from the spirit of the present invention. For example multiple capacitors switched into and out of the circuit to change the pole positions.

Figure 11 shows the preferred embodiment of comparator 380 and comparator 390 which are nominally identical. Comparator 380 (390) includes cross-point switch 383 (393) in addition to comparator cell 382 (392). Cross-point switch 383 (393) allows the inputs to comparator cell 382 (392) to be swapped allowing cancellation of the input offset error. On a first cycle the signal SWAP 502 is low and a first lifetime measurement is made with cross-point switch 383 (393) passing signals straight through to comparator cell 382 (392) inputs. On a second cycle signal SWAP 502 is high and a second lifetime measurement is made with cross-point switch 383 (393) swapping the input signals to the opposite inputs of comparator cell 382 (392). Since an averaging process is applied to all lifetime values determined from first measurement cycles and second measurement cycles any measurement error due to an input offset in either or both comparators is cancelled.

In the present invention reference level generator REF 320 is used to generate the required voltage reference levels and is shown in Figure 12.

Reference level generator 320 includes a voltage reference source 325, preferably a temperature compensated band-gap voltage source, resistors 326, 327 and 328 1 5 and further contains current source 329. Current source 329 may be derived from voltage reference source 325 but may be generated from an independent source.

In the preferred embodiment current source 329 is derived from reference voltage source 325, resistors of the same material or type as resistors 326, 327 and 328 and transistors to provide a current that is relatively independent of temperature in a method that is widely understood by those practiced in the art.

Resistors 326, 327 and 328 in conjunction with current source 329 generate reference voltage V0 322, reference voltage V1 323 and reference voltage VCM 324. Reference voltage V 321 is used in peak detector 370 to set the average peak amplitude of signal 361 to a well defined level. Reference voltage V0 322 is used in comparator 380 to detect when signal 361 passes through the intensity threshold level l. Reference voltage Vi 323 is used in comparator 390 to detect when signal 361 passes through the intensity threshold level l 323. Reference voltage VCM 324 is used in DCR 350 as the common mode reference for signal 361.

Controller 500 provides not only the function of time measurement, low oxygen level decision making, alarm annunciation but also controls the manner in which the measurement channel operates. So, controller 500 has several functions: -The production of timing signals -Lifetime measurement -Measurement channel calibration -Alarm function when a low oxygen level is detected -Power consumption minimisation Figure 13 shows the preferred embodiment of controller 500 including time base generator (TBG) 531, taking as input reference clock 520 and the measured lifetime value, producing an output clock for control signal generator (CSG) 551 related to the lifetime measurement. In some phases of the measurement algorithm the output clock of TBG 531 is not scaled by the lifetime measurement as the result is not available or may be inaccurate. During these phases a fixed frequency clock is used derived directly from reference clock 520. CSG 551 produces the main cyclic timing signals for the measurement channel, the signals SWAP 502, RESTORE 504, PEAK 505, BLANK 506 and EXCITATION 510.

TBG 531 produces an output count based on reference clock 520. During the decay measurement phase the time interval between edges in the comparator output signals 381 and 391 is measured. It is necessary to invert comparator output signals 381 and 391 dependant on the state of signal SWAP 502. Logic circuits 532, 533, 534 and 535 correct the inversion due to the swapping of comparator inputs when SWAP 502 is high generating positive edges as the result of signal 361 passing through comparator reference thresholds 322 and 323. A rising edge at the output of logic circuits 532, 533, 534 and 535 is used to capture the output from TBG 531 in registers 536, 537, 538 and 539 respectively. Further, logic circuits 532 and 534 only allow the first valid edge to be passed to registers 536 and 538 respectively while logic circuits 533 and 535 allow all edges to pass the registers 537 and 539. A first valid edge from signal 381 will latch the output of TBG 531 in register 536. Every valid edge from signal 381 will latch the output of TBG 531 in register 537. Accordingly the value stored in register 537 will be the value of TBG 531 corresponding to the last valid edge from signal 381. Similarly, register 538 records the value from TBG 531 for the first valid edge of signal 391 while register 539 records the value from TBG 531 for the last valid edge of signal 391.

Logic average circuit 540 takes as inputs the outputs of register 536 and register 537 producing the average value of the contents of register 536 and register 537 effecting noise filtering of the edges in comparator output signal 381.

Similarly logic average circuit 541 takes as inputs the outputs of register 538 and register 539 producing the average value of the contents of register 538 and register 539 effecting noise filtering of the edges in comparator output signal 391.

Logic difference circuit 542 produces an output based on the difference between the output of logic average circuit 541 and logic average circuit 540. This is a direct measure of the fluorescence decay lifetime and produces the output 501. Output 501 is valid for pairs of measurement cycles and is then averaged in logic averaging circuit 543 which is controlled by sequencer 544. Sequencer 544 is preferably controlled by an external microcontroller through bus 521 which passes parameters such as the required number of measurement cycles and accuracy.

Sequencer 544 in controller 500 then ensures correct operation of all logic circuits.

The low oxygen alarm level is stored in register 545, having been determined by the external microcontroller based on initial calibration and appropriate factory calibration data. Logic comparator 546 compares the value stored in register 545 against the averaged lifetime value from logic average circuit 543. Once the total number of measurement cycles has been reached sequencer 544 latches the output of logic comparator 546 in register 547, the state of the output of register 547 determining whether an alarm should be flagged or otherwise indicated on alarm signal 522.

Sequencer 544 further generates signals for the operation of the calibration routine, namely a digital data word which is passed to digital to analogue converter 548 producing signal 508 and signal 507 for calibration purposes.

Sequencer 544 further generates a digital data word which is passed to digital to analogue converter 549 producing a signal 511 for the purpose of setting the intensity of excitation light source 302.

Sequencer 544 further produces the control signals 503 for LPF 360 bandwidth control.

Sequencer 544 is also used in the second embodiment of the variable gain control loop with input signal being sampled by analogue to digital converter 550.

The output of the analogue to digital converter 550 is processed by sequencer 544 producing signal 509 for use with VGA 340 with digital control of the gain or in conjunction with a digital to analogue converter when VGA 340 has analogue gain control.

Timing signals EXCITATION 510, SWAP 502, BLANK 506, RESTORE 504 and PEAK 505 are generated in a cyclical manner to control the measurement of lifetime. The basic timing sequence of Controller 500 during the measurement phase is shown in Figure 3 and can be described in the following phases: -Light source illumination and blanking phase -Lifetime decay measurement phase -DC offset measurement phase In the light source illumination and blanking phase the measurement channel is first placed into the blanking state and then the light source is turned on. The light source is then turned off and the measurement channel un-blanked.

In the preferred embodiment of the present invention the period that the light source is illuminated may be variable, allowing minimum light source period in addition to varying the absolute light source output illumination through a separate adjustment mechanism.

In the lifetime decay measurement phase the peak detector is enabled. In the preferred embodiment of the present invention the lifetime decay measurement phase may be variable with a period related to the lifetime of the fluorescence emission process. As the lifetime decreases then the lifetime decay measurement period is also reduced. This allows the averaging of the measurement cycles to be completed more rapidly. Further, in the present invention, the lifetime measurement period is related to the accuracy. As the accuracy reduces the lifetime decay measurement period is also reduced. By means of an example, consider the case where the emission lifetime is 1 ps. It would be appropriate to set the lifetime decay measurement period to, say lox lifetimes or 1 Ops. Due to the statistical nature of the emission process at the end of the lifetime decay measurement period there would be a very small number of fluorophores that had not decayed, 0.0045% of the initial population, most likely this would be well below the noise floor of the measurement system. If the DC offset measurement phase was defined as 1X lifetime or 1 pS and the light source illumination phase was 1X lifetime or ips then the total measurement cycle time would be 1 2X lifetimes or 1 2ps. If then the lifetime decay measurement period was reduced to 5X lifetime or 5ps then the total measurement cycle would be 7X lifetimes or 7ps. However, the DC restore circuit would be sampling during a time when the emission process was on-going and contributing to an error in the DC offset measurement. Reducing the lifetime measurement phase to 5X lifetimes or 5ps there would be 0.67% of the initial fluorophores population remaining to decay back to the initial, non-excited state introducing an error in the measured lifetime.

Using lifetimes as a unit for the timing rather than absolute time it is possible to relate the measurement cycle to the accuracy.

As the lifetime is not accurately know during the search for the optimum bandwidth setting of LPF 360 the measurement cycle period is retained at a value consistent with the longest lifetime during the search for the optimum bandwidth setting of LPF 360. Further as the fluorescence lifetime reduces then, at a given accuracy setting and therefore a given number of lifetimes in the lifetime decay phase and the DC offset measurement phase the total measurement cycle period also reduces. Again, by means of an example, consider the above example where the fluorescence lifetime reduces from 1 ps to 0.Sps. At a high accuracy setting it may take 12X lifetimes for a complete measurement cycle which would equate to l2ps for a fluorescence decay lifetime of ips or 6ps for a fluorescence decay lifetime of 0.Sps. In this manner relating the unit time period to the fluorescence decay lifetime will allow the measurement cycle and the total measurement period to be reduced when compared to fixed unit time period timing.

The measurement cycle is repeated based on a pre-defined algorithm over a fixed or variable number of cycles alternating the signal SWAP 502 on consecutive cycles In the lifetime measurement controller 500 accepts input signals 381 and 391 and measures the time difference between rising edges in signal 381 signal 391 or falling edges in signal 381 and signal 391 dependant on the state of signal SWAP 502. As the measurement system is based on probabilistic events it is necessary to average the time measurement over many cycles dependant on the final accuracy requirement. Techniques to render accurate measurement of time are well known to those skilled in the art and may include, but are not limited to: -The use of a high frequency clock to reduce quantisation errors in time determination of each timing edge -The use of a lower speed clock making a coarse measurement and a higher frequency clock, perhaps derived from the lower speed clock in a delay locked loop, making fine measurements on the residual portion of the coarsely quantised time period -Time delay measurement using a pulse shrinking techniques in a delay locked loop -The use of noise filtering on each timing edge to allow operation in the presence of large noise In addition to the calibration of the system against a gas containing a known level of oxygen calibration of LPF 360 may be required. This may be typically implemented once per power-on cycle depending on the design and stability of LPF 360. As part of the matched filter system it is preferable to know the bandwidth of LPF 360 at each bandwidth setting or, at least at a sufficiently large number of bandwidth settings where some form of linear interpolation or non-linear interpolation can be employed to determine relevant information for any bandwidth setting not in the look-up table. The relevant information is a lifetime value associated with a bandwidth setting and accuracy requirement.

In order to perform calibration of LPF 360 a square wave may be injected into the measurement channel and the fall time be measured. From an a priori knowledge of LPF 360 topology and the ratio of the pole frequencies it is possible to assign a minimum lifetime value with a particular bandwidth and accuracy requirement. The steps to create a look-up table are shown in Figure 13: -Set LPF 360 bandwidth to a first value -Inject a square wave -Measure the falling edge time constant -Average over a number of measurement cycles if required -Modify the decay time constant by a set of constants determined from LPF 360 pole locations and accuracy settings to give the minimum lifetime associated with that particular bandwidth setting and accuracy -Store the results in a look-up table with the bandwidth setting as an entry pointer -Repeat for each bandwidth setting A look-up table would preferably include multiple entries for each bandwidth setting, the number of entries corresponding to the number of accuracy settings that are to be used. A minimum of one entry is required in the look-up table which would correspond to a single accuracy parameter for each bandwidth setting.

By means of an example, consider the case where it is required that the look-up table holds the minimum lifetime values for each bandwidth setting of LPF 360 at four accuracy settings of 0.5%, 1%, 2% and 4%. Further consider that the decay time constant of LPF 60 was measured at one bandwidth setting to be 320ns. The look-up table may then hold the minimum lifetime values of 3*320ns, 2.4*320ns, 2.0*320ns and 1.5*320ns i.e. O.96ps, O.77ps, O.67ps and O.48ps for the respective accuracy settings of 0.5%, 1%, 2% and 4%.

The scaling factors in the above example are typical of the values determined for a filter topology where there is a first pole at a first pole frequency and a second pole at 20X the first pole frequency. Other filter pole locations will require different scaling factors.

When a subsequent lifetime measurement is undertaken on a fluorescence signal the measured lifetime is compared against the accuracy limits in the look-up table as part of the decision process to determine is the bandwidth of LPF 360 is matched to the signal.

As describe in the disclosure of the present invention controller 500 is used to set the bandwidth of LPF 360. Figure 14 presents a flow chart for the preferred measurement process that is implemented in controller 500. A number of parameters are passed to the measurement routine: -Accuracy, the accuracy limit due to bandwidth limiting -Cinit, the number of cycles for a first, approximate, measurement to aid in setting LPF 360 bandwidth -Cfinal, the number of cycles for the final accurate measurement The routine can be broken down into an initial part where the bandwidth is matched to a first measurement of the fluorescence lifetime. Once LPF 360 bandwidth is matched to the lifetime then the final measurement is made over a larger number of cycles.

By means of an example a typical measurement may set the parameters Accuracy = 0.5%, Cinit = 500 and Cfinal = 2000. LPF 60 bandwidth is initially set to the maximum value with variable i = 0 (note that as this variable is increased the bandwidth is reduced). The lifetime is then measured over 500 cycles and compared against the values in the look-up table based on the above accuracy

statements and measured bandwidths.

If the measured lifetime value is lower than the accuracy value for the bandwidth setting of LPF 360 then the bandwidth of LPF 360 would need to be increased. However, this cannot occur as LPF 360 bandwidth was initialised to the highest setting. In this case an error code is generated and the measurement routine aborted.

If the measured lifetime is above value associated with the particular bandwidth setting of the LPF 360 then the lifetime is compared against the lifetime value associated with the next lower bandwidth setting. If the measured lifetime is less than the value associated with this next bandwidth setting then it is recorded that bandwidth of LPF 360 is set correctly and the measurement cycle advances to a full measurement cycle where the lifetime is averaged over Cfinal individual cycles. If however, the lifetime is still lower than the value associated with this next bandwidth setting then the bandwidth in reduced and the process repeated until the correct bandwidth setting is obtained or until the bandwidth is reduced to the minimum allowed. Should the bandwidth reach the minimum setting then an error code is generated and the routine aborted.

It is further checked within the measurement routine to ensure that at the end of the measurement period where the lifetime is averaged over Cfinal cycles the averaged lifetime result is within the acceptable accuracy limits for the particular setting of LPF 360.

Although this routine is shown in the preferred form for the present invention it is clear to those skilled in the art that methods exist to reduce the time taken to find the correct operating point of LPF 360. One possible method would be to make a decision based on one rough measurement and knowing the LPF settings and associated lifetime limit values jump straight to the optimum bandwidth setting. However, it should be noted that if the lifetime measured in the initial cycle is close to a decision boundary then it is possible to set the LPF 360 bandwidth to an incorrect setting. Another method to reduce the time taken to find the optimum bandwidth setting would be to employ a binary search.

It is preferable that a watchdog timer is deployed within controller 500 to ensure that the measurement routine does reach a stable operating point and does not hunt back and forwards between different bandwidth settings. This could occur where there were many bandwidth settings with small differences in the associated lifetime values. The solution to this particular problem would be to increase the Cinit value to reject more noise through the averaging process.

A second method of setting LPF 360 bandwidth is now disclosed. When 1 0 signal 307 to measurement channel 301 is un-blanked the rising edge of the signal 307 rises rapidly. Subsequently signal 361 is then low pass filtered by the bandwidth of measurement channel 301 and in particular LPF 360. Comparator 380 and comparator 390 will produce output signals that are related in time to the bandwidth of LPF 360. A measurement of the rise-time of signal 361 will return a value that can be used to set the bandwidth of LPF 360. For example, using the same accuracy limits, photon count and pole placements as described in the first method of setting LPF 360 bandwidth it can be stated that: -For an accuracy of better than 0.5% LPF 360 bandwidth would be set to produce a rise time less than 3.OX the fluorescence lifetime -For an accuracy of better than 1% LPF 360 bandwidth would be set to produce a rise time less than 2.4X the fluorescence lifetime -For an accuracy of better than 2% LPF 360 bandwidth would be set to produce a rise time less than 2X the fluorescence lifetime -For an accuracy of better than 4% LPF 360 bandwidth would be set to produce a rise time less than 1.5X the fluorescence lifetime Changes in the structure of controller 500 would allow measurement of the rise time and fall time of signal 361 to be made whereby sequencer 544 controlled LPF 360 bandwidth dependent on the accuracy setting and a priori knowledge of the LPF 360 characteristics and thereby rise time to emission time scaling factors.

In the preferred embodiment of the present invention controller 500 may be used to optimise the power consumption. As part of the initial power-up calibration routine the bias current of excitation light source 302 is minimised. Measurements are made over a fixed number of measurement cycles and then the bias current of excitation light source 302 is reduced until the measured result starts to deviate from the original measurement. In the measurement routine controller 500 also can reduce the period of the measurement cycle in relation to the accuracy parameter thus reducing the complete measurement time.

In the measurement routine controller 500 also can reduce the number of measurement cycles in relation to the accuracy parameter thus reducing the complete measurement time.

In each of the above cases periodic re-calibration is undertaken by controller 500 to maintain accuracy.

No measurement system is devoid of measurement errors. It is the intention to show the preferred embodiment of the present invention system is robust against errors in the components, circuits and techniques employed in the execution of the invention.

Error sources in the comparators due to input offset voltage have a pronounced effect on the measurement accuracy. In particular an offset on the comparator detecting the lower intensity level may be several times larger than the error resulting in the same offset in the higher intensity level comparator. The effect of comparator offset is an error in the lifetime determination. In the present invention the effect of input offset errors in the comparators are removed by swapping the comparator inputs on alternate cycles as shown in Figure 3. In conjunction with the swapping of comparator inputs each comparator output must be inverted in phase with the swapping of the inputs. This may be implemented in the comparator or in the controller logic. In the preferred embodiment this inversion to maintain correct signal polarity is performed in controller 500.

A large number of cycles are used to remove the statistical properties of the emitted light. Preferably an even number of cycles should be used to average the results which, with the swapping of the comparator inputs on odd and even cycles, would ensure an equal number of odd and even cycles contributing to the averaging process. However, as the total number of measurement cycles is large even using an odd number of cycles has little effect on the accuracy of the final result.

Mismatch in comparator propagation delay is generally a second order effect. Calibration will eliminate this error source. Temperature variation of the mismatch is generally so small as to be considered negligible.

It is well known that DC offsets in a measurement channel can result in erroneous results or incorrect operation. Control of DC offsets is equally important in this measurement channel. In the present invention the measurement system removes the DC component in the signal by sampling the offset every measurement cycle. Initial calibration of the measurement system will remove this error but not any drift in the DC component. The inclusion of a reference band-gap circuit with low drift as the main reference voltage source will minimise errors resulting from DC offset sources.

An advantage of reference generator REF 320 is that the voltages produced track each other in the desired proportions. An offset in reference level Vp 321 has no effect on the system other than to increase the average value of the peak voltage of signal 361. The relationship between reference level Vp 321, reference level V0 322 and reference level V1 323 is maintained by current source 329 and ratio of resistors 326, 327 and 328.

An offset in current source 329 of positive magnitude has no effect on reference level V 321. Reference level V0 322, reference level V1 323 and reference level VCM 324 will reduce in value whilst maintaining the same voltage ratios. An offset in current source 329 of negative magnitude again has no effect on the reference level Vp 321 while all other reference levels will increase in value whilst maintaining the same voltage ratios.

The absolute value of current source 329 is not an issue except in determination of the minimum supply voltage. Preferably current source 329 should be derived from the same voltage source as reference level V 321 and same component types as resistors 326, 327 and 328. Similarly, the temperature coefficient of current source 329 is not an issue as all reference levels will track each other.

Errors in the matching of resistors 326, 327 and 328 can result in a gain error in the lifetime measurement. In a monolithic integrated solution of the present invention such an error may occur due to process variations in the sheet resistance of the resistor material. Again, in a monolithic integrated solution of the present invention such an error may occur due to the difficulty in matching components that are not related through integer ratios. One solution to this problem is to select intensity threshold levels that do result in component value ratios with integer relationships. In such a scheme a scaling factor would need to be applied to the measured lifetime to correct for the different intensity threshold levels. In a monolithic integrated solution layout techniques known to those skilled in the art can minimise such errors. In any implementation a residual error would be removed by the sensor calibration process.

Resistor temperature coefficients for resistors 326, 327 and 328 cancel in the preferred embodiment when current source 329 is derived from a resistor with the same temperature coefficient.

By circuit design and calibration techniques it has been shown that the measurement channel disclosed in the present invention is robust against many sources of offsets, matching errors both static and dynamic.

It will be apparent to those skilled in the art that the methods and techniques disclosed in the present invention may be employed in other measurement systems where a time constant is to be determined.

Claims

WE CLAIM: A rebreather low oxygen alarm comprising a sol-gel oxygen sensor with a measurement channel and a controller that indicates or annunciates an alarm state when the oxygen level falls below a threshold.
2. A fluorescence decay lifetime measurement system containing a means of amplifying and conditioning the emission signal from a sol-gel oxygen sensor that emits light with a predominantly exponential fluorescence decay time constant to maintain constant peak amplitude, constant common mode level and limiting the spectral content in a deterministic manner, comprising: a sal-gel sensor excitation light detector; a first amplifier with input connected to the sol-gel sensor excitation light detector producing an output voltage related to the light intensity incident on the light collection device; a variable gain amplifier increasing or decreasing the voltage amplitude from the first amplifier output in response to a control input signal; a DC restore circuit that samples the DC content of the variable gain amplifier output, subtracting said DC content from the input signal producing an output signal with the DC level set to a known value; a variable bandwidth low pass filter with one or more control signals setting the bandwidth of said low pass filter, producing an output signal which is the low pass filtered output of the DC restore circuit; an averaging peak detector producing an output that directly controls the gain of the variable gain amplifier; a first comparator detecting when the emission signal passes through a first intensity level; a second comparator detecting when the emission signal passes through a second intensity level; a reference level generator providing a plurality of reference levels; a sol-gel sensor excitation light source with controlled light intensity; a first switch and control signal for setting the measurement channel into a quiescent point when the sol-gel sensor is directly illuminated; a second switch, control signal and signal source for injection of a calibration signal into the measurement channel and a controller for monitoring and controlling the operation and output of the measurement channel.
3. A variable gain amplifier according to claim 1 comprising one or more amplifiers connected in cascade, with one or more said amplifiers having a gain control input whereby the gain of said amplifier or amplifiers may be increased or decreased in relation to a gain control input signal from, in a first embodiment, a peak detector according to claim 3, the variable gain amplifier controlled to maintain the peak amplitude of a fluorescence decay signal constant, said gain control input being an analogue signal.
4. A variable gain amplifier according to claim 1 comprising one or more amplifiers connected in cascade, with one or more said amplifiers having a gain control input whereby the gain of said amplifier or amplifiers may be increased or decreased in relation to a gain control signal from, in a second embodiment, a controller according to claim 3, adjusting the variable gain amplifier to maintain the peak amplitude of a 1 5 fluorescence decay signal constant, said gain control signal being an analogue signal or one or more digital control signals from said controller.
5. A DC restoration circuit according to claim 1 comprising a first amplifier generating an error signal from the difference between the output of said DC restoration circuit and a common mode reference level; a first switch connecting the output of the first amplifier to a first capacitor, the switch being controlled in response to a control signal dictating the point in time to open and close in relation to the timing of the emission signal, effecting an averaging of the common mode of the emission signal for the duration that the switch is closed; a second amplifier generating an output signal from the difference between the input signal and the voltage on the first capacitor, the average common mode of the input signal.
6. A low pass filter according to claim 1 with variable bandwidth comprising: one or more input control signals that set the bandwidth in a deterministic manner; one or more poles determining the bandwidth at each control input or input signals setting, wherein the ratio of the frequency of each pole to every other pole is fixed, all poles responding in the same manner to a change in the control input or inputs.
7. A low pass filter according to claim 1 where the bandwidth of the low pass filter is set by the controller relative to the measured lifetime of the emission signal, wherein the bandwidth of the low pass filter is increased when the measured lifetime of the emission signal decreases and wherein the bandwidth of the low pass filter is reduced when the measured lifetime of the emission signal increases, the product of the low pass filter bandwidth and measured emission lifetime remaining 1 0 substantially constant at all measured emission lifetime values.
8. A peak detector according to claim 1 that measures the peak amplitude of the emission signal average the peak amplitude over a known period of time or number of measurement cycles to maintain the average peak amplitude constant, comprising: a first peak detection circuit comprising a first diode, first capacitor with a first resistor setting the decay time constant of the first peak detection circuit; a second peak detection circuit comprising a second diode and second resistor, and input connected to a first reference level, compensating the reference level for the voltage drop of the first diode in the first peak detection circuit; a transconductance amplifier producing an output current dependant on the difference between the voltage from the second peak detection circuit and the voltage from the first peak detection circuit; a switch controlled by a control signal from the controller the switch closed when the emission signal is in the emission decay phase and open at other times, the switch connecting between the output of the transconductance amplifier and a second capacitor, wherein the second capacitor averaging the current from the transconductance amplifier when the switch is closed and which holds the voltage effectively at a constant level, the level across second capacitor at the moment the switch is opened.
9. A peak detector according to claims 1 and 8 in a first embodiment of the gain control scheme wherein the output signal from the peak detector second capacitor connects directly to the control input voltage of the variable gain amplifier.
10. A peak detector according to claims 1 and 8 in a second embodiment of the gain control scheme wherein the output signal from the peak detector second capacitor connects to the controller which measures the voltage, equivalent to the amplitude of the average peak emission signal, and directly controls the gain of the variable gain amplifier to maintain the average peak voltage of the emission signal at a constant value.

1 0
11. A first comparator according to claim 1 comprising a circuit comparing the amplified, common mode controlled and bandwidth restricted emission signal from the low pass filter against a first intensity threshold level, said comparator output level changing state when the said emission signal passes through the first intensity level.
12. A first comparator according to claims 1 and 11 wherein the inputs of the comparator may be swapped depending on the state of an input control signal, the first intensity threshold level connecting to the non-inverting comparator input and the emission signal from the low pass filter connecting to the inverting comparator input when the input control signal is in the logic zero state, the comparator inputs swapped when the input control signal is in the logic one state.
13. A second comparator according to claim 1 comprising a circuit comparing the amplified, common mode controlled and bandwidth restricted emission signal from the low pass filter against a first intensity threshold level, said comparator output level changing state when the said emission signal passes through the second intensity level.
14. A second comparator according to claims 1 and 13 wherein the inputs of the comparator may be swapped depending on the state of an input control signal, the second intensity threshold level connecting to the non-inverting comparator input and the emission signal from the low pass filter connecting to the inverting comparator input when the input control signal is in the logic zero state, the comparator inputs swapped when the input control signal is in the logic one state.
15. A reference generator according to claim 1 providing a plurality of reference voltage levels that track each other and are robust against component offsets and mismatches comprising: a temperature compensated voltage source to derive a first reference level; a current source and a plurality of resistors derive a second reference level, a third reference level and a fourth reference level, the first reference level being larger than the second reference level, the second reference 1 0 level in turn larger than the third reference level, the third reference level in turn larger than the fourth reference level; the current source derived from the first reference voltage source, a current mirror and one or more resistors of the same type as the resistors used to generate the second, third and fourth reference levels.
16. A reference generator according to claims 1 and 15 comprising: a first reference level used to set the peak amplitude of the emission signal in the peak detector and variable gain amplifier; a second reference level to set the first intensity threshold in the first comparator; a third reference level to set the second intensity threshold in the second comparator; a fourth reference level to set the common mode voltage of the emission signal.
17. A switch according to claim 1 for setting the measurement channel to a well defined state during the period that the sol-gel sensor is being illuminated comprising: a first switch; a first control signal connecting first switch between the input of the transimpedance amplifier and system ground in a timely manner under the control of the first control signal.
18. A switch according to claim 1 for introduction of a test or calibration signal at the trans-impedance amplifier input comprising: a second switch; a second control signal for second switch and a signal source from the controller, said signal source being a digital or analogue signal binary waveform with fast rise-time and fast fall time relative to the time constants of the low pass filter with defined amplitude, the control signal enabling the introduction of the signal source into the measurement channel by the controller in the calibration of the measurement channel low pass filter.
19. A controller according to claim 1 in a first embodiment comprising a time bas generator; a control signal timing generator; a time measurement unit; a low oxygen detection and alarm generator; a control flow sequencer; a plurality of digital to analogue converters.
20. A controller according to claim 1 in a second embodiment comprising a time bas generator; a control signal timing generator; a time measurement unit; an alarm detection unit; a control flow sequencer; a plurality of digital to analogue converters and an analogue to digital converter.
21. A time base generator according to claim 20 that produces a variable clock by modifying the frequency of a reference clock input signal, scaling the frequency in relation to the measured and averaged fluorescence decay lifetime, the variable clock increasing when the measured and averaged fluorescence decay lifetime decreases, and decreases when the measured and averaged fluorescence decay lifetime increases in a linear manner, the scaling applied in certain parts of the measurement process, while in other parts of the measurement process no scaling is applied.
22. A control signal generator according to claims 19 and 20 that produces a set of signals for control of the measurement channel from the variable clock produced by the time base generator including: a first signal to control the light source; a second signal to control the blanking of the measurement channel during the period that the light source is illuminated; a third signal to control the peak detector; a fourth signal to control the DC restore circuit; a fifth signal controlling the swapping of the first comparator inputs and the second comparator inputs.
23. A control signal generator according to claims 19 and 20 in which control signals are generated over a measurement cycle in a cyclical manner with each measurement cycle including a plurality of time phases comprising; a light source illumination and blanking phase; a lifetime decay measurement phase and a DC offset measurement phase, wherein each time phase is based on a unit time period, the unit time period being related to the actual fluorescence decay lifetime time constant.
24. A control signal generator according to claims 19 and 20 in which control signals are generated over a measurement cycle in a cyclical manner with each measurement cycle including a plurality of time phases comprising; a light source illumination and blanking phase; a lifetime decay measurement phase and a DC offset measurement phase, wherein each time phase is based on a unit time period, the unit time period being related to the actual fluorescence decay lifetime time constant. Further, where the number of unit time periods in the decay measurement phase is varied in relation to the accuracy, a lower accuracy requirement resulting in fewer number of unit time periods while a higher accuracy results in a larger number of unit time periods, a unit time period related to the actual fluorescence decay lifetime time constant.
25. A time measurement unit according to claims 19 and 20 that: conditions the signal from the first comparator; conditions the signal from the second comparator; applies filtering to said comparator signals determining the average intensity threshold crossing point; measures the time difference between the first comparator filtered edge and second comparator filtered edge, the time measured being substantially equal to the fluorescence decay lifetime of the sol-gel oxygen sensor providing a measure of the oxygen levels in the locale of the said sal-gel oxygen sensor and further averages the measured time difference over a number of measurement cycles dependant on the required accuracy.
26. A low oxygen detection and alarm generator according to claims 19 and that comprises: a first register for holding a low oxygen alarm threshold value, the value in units of time corresponding to a fluorescence decay lifetime associated with a given oxygen level, temperature and pressure level; a digital comparator generating an output when the averaged lifetime value from the time measurement unit according to claim 21 is greater than the first register holding the low oxygen alarm threshold value; a second register that latches the output of the digital comparator at the end of the measurement period producing an alarm output signal corresponding to the situation when the measured lifetime indicates a low oxygen level.
27. A control flow sequencer according to claims 19 and 20 comprising a look up table containing a plurality of minimum lifetime values, one value for each low pass filter bandwidth setting and accuracy combination, each minimum lifetime value uniquely determined for a given pole positions of the measurement channel low pass filter.
28. A control flow sequencer according to claims 19 and 20 with control inputs and outputs controlling both the operation of the controller and measurement channel whereby the fluorescence decay lifetime can be measured and the results compared against a value selected from the look up table according to claim 29 according to a bandwidth setting and accuracy setting in a particular measurement or measurement period.
29. A controller according to claims 19 and 20 comprising: a first digital to analogue converter with a first control signal for controlling the injection of a signal used in the calibration of the measurement channel; a second digital to analogue converter and a second control signal to vary the intensity of the excitation light source and timing of said excitation light source.
30. A controller according to claim 20 comprising an analogue to digital converter in the second embodiment of the variable gain control according to claim 10.
31. A method of utilising a sol-gel oxygen sensor in a rebreather as a low oxygen level measurement and alarm annunciation system.
32. A method of measuring the fluorescence decay lifetime of a sol-gel oxygen sensor according to claim 31 using a two-point measurement method employing comparators and levels.
33. A first method of setting the low pass filter bandwidth in a sol-gel oxygen sensor system according to claim 31 incorporating a measurement channel with said low pass filter, whereby a calibration signal is injected into the measurement channel and a time constant measured at each and every low pass filter bandwidth setting for each accuracy setting, each time constant further scaled by known values generating minimum lifetime entries in a look up table.
34. A second method of setting the low pass filter bandwidth in a sol-gel oxygen sensor according to claim 31 whereby the rise time of the fluorescence decay lifetime signal is measured and the bandwidth of the measurement system adjusted to ensure the measured fluorescent 1 5 decay lifetime is larger then the measured rise time of the fluorescence decay signal, the amount by which the decay time is larger than the rise time being set by known characteristics of the measurement channel bandwidth.
35. A method of reducing the power consumption of a low oxygen level alarm system in a rebreather using a sol-gel oxygen sensor and further reducing photo-bleaching of the sol-gel oxygen sensor comprising an excitation light source whose intensity and on duration are minimised, whereby in an initial calibration, and subsequent periodic recalibrations, measurement of the fluorescence decay lifetime of the sol-gel oxygen sensor is made during a period when the oxygen level is substantially constant, either by the sol-gel sensor presented with a known oxygen level or by measurements taken over a period where the oxygen level does not change, the fluorescence decay lifetime being a measure of the oxygen level, the excitation light source intensity and on duration being reduced every measurement period while the measured lifetime remains within the measurement accuracy level.
36. A first method of minimising the measurement period of a low oxygen level alarm system in a rebreather wherein the timing of the measurement cycle is adapted to minimise the total measurement period, said measurement cycle comprising: a sol-gel illumination phase wherein the excitation light source is turned on; a decay measurement phase where the fluorescence decay lifetime is measured and a dc offset measurement phase where the DC offset of the emission signal is measured and subtracted from the signal, in a first method wherein the number of time units in the decay measurement phase time is adjusted 1 0 according to the accuracy requirement.
37. A second method of minimising the measurement period of a low oxygen level alarm system in a rebreather wherein the timing of the measurement cycle is adapted to minimise the total measurement period, said measurement cycle comprising: a sol-gel illumination phase wherein the excitation light source is turned on; a decay measurement phase of duration set by a multiple of unit time periods wherein the fluorescence decay lifetime is measured and a dc offset measurement phase where the DC offset of the emission signal is measured and subtracted from the signal, and whereby the unit period of the decay measurement phase is adjusted in relation to the measured lifetime.