Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/22—Detection or location of defective computer hardware by testing during standby operation or during idle time, e.g. start-up testing
  • G06F11/26—Functional testing
• • G—PHYSICS
  • G11—INFORMATION STORAGE
  • G11C—STATIC STORES
  • G11C29/00—Checking stores for correct operation ; Subsequent repair; Testing stores during standby or offline operation
  • G11C29/04—Detection or location of defective memory elements, e.g. cell constructio details, timing of test signals
  • G11C29/08—Functional testing, e.g. testing during refresh, power-on self testing [POST] or distributed testing
  • G11C29/12—Built-in arrangements for testing, e.g. built-in self testing [BIST] or interconnection details
  • G11C29/18—Address generation devices; Devices for accessing memories, e.g. details of addressing circuits
  • G11C29/30—Accessing single arrays
  • G11C29/32—Serial access; Scan testing
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
  • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
  • G01R31/28—Testing of electronic circuits, e.g. by signal tracer
  • G01R31/317—Testing of digital circuits
  • G01R31/3181—Functional testing
  • G01R31/3185—Reconfiguring for testing, e.g. LSSD, partitioning
  • G01R31/318533—Reconfiguring for testing, e.g. LSSD, partitioning using scanning techniques, e.g. LSSD, Boundary Scan, JTAG
  • G01R31/318583—Design for test
• • G—PHYSICS
  • G11—INFORMATION STORAGE
  • G11C—STATIC STORES
  • G11C29/00—Checking stores for correct operation ; Subsequent repair; Testing stores during standby or offline operation
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
  • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
  • G01R31/28—Testing of electronic circuits, e.g. by signal tracer
  • G01R31/317—Testing of digital circuits
  • G01R31/3181—Functional testing
  • G01R31/3185—Reconfiguring for testing, e.g. LSSD, partitioning
  • G01R31/318533—Reconfiguring for testing, e.g. LSSD, partitioning using scanning techniques, e.g. LSSD, Boundary Scan, JTAG
  • G01R31/318541—Scan latches or cell details
• • G—PHYSICS
  • G11—INFORMATION STORAGE
  • G11C—STATIC STORES
  • G11C29/00—Checking stores for correct operation ; Subsequent repair; Testing stores during standby or offline operation
  • G11C29/04—Detection or location of defective memory elements, e.g. cell constructio details, timing of test signals
  • G11C29/08—Functional testing, e.g. testing during refresh, power-on self testing [POST] or distributed testing
  • G11C29/12—Built-in arrangements for testing, e.g. built-in self testing [BIST] or interconnection details
  • G11C29/18—Address generation devices; Devices for accessing memories, e.g. details of addressing circuits
  • G11C29/30—Accessing single arrays
  • G11C2029/3202—Scan chain

Definitions

• Integrated circuit chips must be tested for manufacturing flaws. Manufacturing flaws are often modeled using fault models.
• the Stuck- At fault model is the most basic fault model.
• a Stuck- At fault occurs when a particular connection in the circuit remains at ("stuck at") a low level (known as Stuck-At 0 or SAO) or remains at (“stuck at") a high level (known as Stuck-At 1 or SAl).
• a SAO fault on a connection is detected when the circuit is controlled to place a 1 (high) value on that connection, but a 0 (low) value is observed on that connection.
• a SAl fault on a connection is detected when the circuit is controlled to place a 0 value on that connection, but a 1 value is observed on that connection. Circuits whose internal connections are easily controlled and observed are more testable than circuits whose internal connections are more difficult to control and observe.
• Scan testing is a widely used technique to detect manufacturing defects in integrated circuits.
• a scanned circuit some or all of the flip-flops are replaced by scan flip- flops.
• a scanned circuit is using "full scan” if all the flip-flops are scan flip-flops; otherwise the circuit is using "partial scan”.
• Scan provides a method of increasing the controllability and observability of a circuit.
• Mux-scan is the most common method of implementing scan test.
• a mux-scan flip-flop has two modes: a normal operation mode and a scan shift mode.
• normal operation mode the scan flip-flops implement the user's desired (non-test mode) behavior.
• scan shift mode the scan flip-flops are interconnected into one or more shift registers (scan chains).
• Figure 1 shows a prior-art mux-scan flip-flop with asynchronous reset and asynchronous preset pins.
• SE scan enable
• SE scan enable
• SE scan enable
• SE scan shift mode
• CLK clock shift
• a 0 value on the RN asynchronous pin loads the flip-flop with a 0 value.
• a 0 value on the SN asynchronous pin loads the flip-flop with a 1 value.
• the SN asynchronous preset and RN asynchronous reset pins both must remain at a 1 value during scan shift mode. Otherwise, incorrect values will be loaded into the scan chains.
• Flip-flops may be designed to have reset priority (i.e., when SN and RN both have a 0 value, the flip-flop is asynchronously loaded with a 0 value).
• Flip-flops may also be designed to have preset priority (i.e., when SN and RN both have a 0 value, the flip-flop is asynchronously loaded with a 1 value).
• scan-in operation By placing the scan flip-flops in scan shift mode and applying clocks, values can be loaded into the scan flip-flops (scan-in operation) and extracted from the scan flip-flops (scan-out operation).
• a typical scan test sequence is: scan-in, normal operation (with one or more clocks applied to the circuit), and scan-out. Circuit defects are detected when the scan-out values do not match the expected values for a good circuit.
• One scan test sequence can detect multiple faults.
• DFT rules design- for-test rules
• Gate arrays and logic arrays both use pre-designed logic that can be configured by routing to implement a user-design circuit. Manufacturing steps prior to customization routing can be done before the user-designed circuit has been designed.
• the pre-designed logic in a gate array has usually been implemented using custom designed logic. Custom designed logic usually requires more time to design and to validate than logic implemented in standard cell.
• the Modular Array also known as structured array and more recently as a logic array, uses standard cells to implement the pre-designed logic. This technique is described in U.S. Patent Application Serial No. 10/447,465, filed May 28, 2003, and published December 2, 2004 as Publication No. US 2004/0243966, Modular Array Defined by Standard Cell Logic, which is owned by the owner of the instant application, and to the extent permitted by the type of the instant application, is herein incorporated by reference for all purposes. Constructing logic array building blocks from standard cells allow logic arrays to be implemented more quickly than if custom designed logic was used to implement the building blocks.
• the available macros of both gate-arrays and logic arrays are implemented as a base array portion of pre-designed lower layers and a customization portion of at least one upper layer. These macros are necessarily pre-selected and pre -placed. (This is in contrast with conventional standard cell design, which permits the designer of the application circuit to make largely unrestricted selection and placement of instances of cells copied from a standard cell library.)
• the available macros are allocated to the corresponding functions of the application (user) circuit netlist.
• the macros are implemented using custom circuit design and generally are relatively primitive functions.
• the macros are implemented using a pre-existing standard cell library (that frequently is specified by the user rather than the logic array tool vendor) and generally include both primitive and relatively higher-order functions.
• mux-scan can certainly be used to test user-designed circuits.
• the designer must follow all the associated DFT rules. Providing a test method that insures effortless automatic compliance with DFT rules and is compatible with logic arrays, would make it even easier and faster for a designer to have a user-designed circuit manufactured.
• Partition test is a concept described in U.S patents 6,223,313 and 6,611,932, which are owned by the owner of the instant application, and to the extent permitted by the type of the instant application, are herein incorporated by reference for all purposes.
• the invention may be implemented in numerous ways, including as a process, an article of manufacture, an apparatus, a system, a composition of matter, and a computer readable medium such as a computer readable storage medium or a computer network wherein program instructions are sent over optical or electronic communication links.
• these implementations, or any other form that the invention may take, may be referred to as techniques.
• the order of the operations of disclosed processes may be altered within the scope of the invention.
• the Detailed Description provides an exposition of one or more embodiments of the invention that enable improvements in performance, efficiency, and utility of use in the field identified above.
• the Detailed Description includes an Introduction to facilitate the more rapid understanding of the remainder of the Detailed Description.
• Logic blocks for IC designs (including gate-array, standard cell, or logic array designs) provide Design- for-Test-enabled flip-flops (DFT-enabled FFs) that inherently insure compliance with DFT rules associated with scan shifting.
• Test scan-chains are configured by daisy-chaining instances of the logic block in a manner transparent (invisible, hidden) to user-designed application circuits, which can be designed without any user-inserted test structures or other regard for DFT considerations.
• User asynchronous set and reset inputs and all Stuck- At faults on all user pins on these DFT-enabled FFs are observable via capture and scan-out.
• a first type of these DFT-enabled FFs features addressable control to partition test the application circuit.
• a second type of these DFT-enabled FFs features integral capture buffering that eliminates the need for partition test, simplifying control logic and reducing the number of test vectors needed.
• Providing the DFT-enabled FF cells in a given design flow enables transparent effortless automatic compliance with DFT rules when porting a pre-existing application design (such as from an FPGA prototype) that was developed with little or no regard for testability.
• a gate level netlist designed without observation of DFT rules is transformed in a single-pass to a gate level netlist that complies with all DFT rules, by using DFT-enabled FF cells for each application specified FF and interconnecting the DFT-enabled FF cells to form scan-chains.
• the DFT-enabled FFs cells are used in gate- array, standard-cell, and logic array design flows for the design of integrated circuits (chips).
• standard-cell flows at least one type of DFT-enabled FF is added to the standard cell library.
• the pre-placed pre-selected set of logic macros is expanded to include at least one type of DFT-enabled FF.
• the DFT- enabled FF macros are implemented using components from an existing standard cell library. In other embodiments, DFT-enabled FF macros are implemented as custom circuit designs.
• Fig. 1 shows a prior-art mux-scan flip-flop with asynchronous reset and asynchronous preset pins.
• Fig. 2 shows a first DFT-enabled FF embodiment.
• Fig. 3 shows high-level waveforms associated with this DFT-enabled FF's normal operation mode, scan shift mode, and capture mode.
• Fig. 4 illustrates setting the DFT-enabled FF's CLK pin to a 0 value when a rising edge at the core flip-flop's CLK pin (MCLK signal) is not desired.
• Fig. 5 shows high-level waveforms for capture mode.
• Fig. 6 shows a variation of the DFT-enabled FF.
• Fig. 7 shows high-level waveforms for capture mode - specifically when both R and S have a 1 value, resulting in the core flip-flop with a final capture value of 1 (preset priority).
• Fig. 8 illustrates adding a NOR gate to the DFT-enabled FF of Fig. 2 as an alternative method of a generating a "FREEZE LOW-equivalent" signal.
• Fig. 9 illustrates adding a NOR gate to the DFT-enabled FF of Fig. 6 as an alternative method of a generating a "FREEZE LOW-equivalent" signal.
• Fig. 10 shows the truth table associated with this implementation - when either or both CC or CR is set to a 1 value, FREEZE LOW is at a 0 value.
• Fig. 11 shows how adding two inverters and 2 NORs to the DFT-enabled FF (previously shown in Figure 2) allows the DFT-enabled FF to behave as a DFT-enabled latch.
• Fig. 12 shows how adding one inverter to the DFT-enabled FF (of Figure 2) allows the DFT-enabled FF to behave as a DFT-enabled buffer test point.
• Figure 13 shows a DFT-enabled latch embodiment that behaves like a flip-flop during scan shift, but behaves like a latch during normal operation and capture mode.
• Figure 14 is the first of two examples of unstable circuits.
• the flip- flop oscillates between 0 to 1 and 1 to 0.
• Figure 15 shows another example of a non-deterministic circuit - in this case, a ring oscillator.
• Figure 16 shows how two DFT-enabled buffers break the loop, without affecting the user circuit's functionality - the user circuit still oscillates.
• Figure 17 shows how one DFT-enabled buffer breaks a loop, without affecting the user circuit's functionality - the user circuit still oscillates.
• Figure 18 shows a simplified diagram of a DFT-enabled FF - with the following test pins SDI, SCLK, FREEZE LOW and the output pin Q is shown.
• Figure 19 shows how the simplified DFT-enabled FFs (of Figure 18) can be placed in a two-dimension array and selected using row and column addressing.
• Fig. 20 illustrates a DFT-Enabled FF that eliminates the need for partitioning circuits, simplifies loop breaking, and reduces the vector count required to test circuits.
• Fig. 21 illustrates the timing of the DFT-Enabled FF of Fig. 20. DETAILED DESCRIPTION
• word labels including but not limited to: first, last, certain, various, further, other, particular, select, some, and notable
• word labels may be applied to separate sets of embodiments; as used herein such labels are expressly not meant to convey quality, or any form of preference or prejudice, but merely to conveniently distinguish among the separate sets.
• the order of some operations of disclosed processes is alterable within the scope of the invention.
• multiple embodiments serve to describe variations in process, method, and/or program instruction features
• other embodiments are contemplated that in accordance with a predetermined or a dynamically determined criterion perform static and/or dynamic selection of one of a plurality of modes of operation corresponding respectively to a plurality of the multiple embodiments.
• a method for designing an integrated circuit comprising: identifying the fabrication process of the integrated circuit and a process-qualified standard cell library; receiving a gate-level netlist specifying at least a portion of the integrated circuit including a plurality of sequential storage functions; programmatically mapping the gate-level netlist onto a target implementation architecture of at least the portion of the integrated circuit including implementing each sequential storage function of the netlist via a respective DFT-enabled storage circuit having integral design- for-testability functionality; and wherein the target implementation architecture for at least the portion of the integrated circuit specified by the netlist uses the standard cell library to implement function blocks of a logic array, the logic array having a base array portion of pre-designed lower layers and a customization portion of at least one upper layer, the function blocks being allocatable to the functions of the netlist.
• each sequential storage function has normal-operation pins including a clock input and at least one asynchronous control input.
• EC6 The method of EC5, wherein the DFT-enabled storage circuit provides observability of all stuck-at faults on the normal-operation pins.
• DFT-enabled FF embodiments are taught herein that insure inherent compliance with the majority of the DFT rules associated with mux-scan. Further embodiments combine the DFT-enabled FF with partition test techniques to achieve testing methods that can test user- designed circuits without requiring any user- inserted test structures - in other words, the test methods are transparent to the user.
• the DFT-enabled FF includes logic that ensures correct scan shift mode, and also provides a novel method of observing (capturing) data.
• Figure 2 shows an implementation of the DFT-enabled FF.
• the user pins are D, CLK, R, S, and Q. From a user's functional viewpoint (e.g., when the DFT-enabled FF is not in test mode), when R has a 1 value, the core flip-flop is asynchronously loaded with a 0 value. From a user's functional viewpoint (e.g., when the DFT-enabled FF is not in test mode), when S has a 1 value, the core flip-flop is asynchronously loaded with a 1 value.
• the core flip-flop has active low asynchronous RN and SN pins, while the DFT-enabled FF implements the user-controlled asynchronous reset R and preset S pins as active high signals, because in current common semiconductor technologies, a 2-input NAND gate usually has a smaller area than a 2-input OR gate of equivalent drive.
• This diagram shows a DFT-enabled FF whose core flip-flop has reset priority, which means the DFT-enabled FF should also implement reset priority.
• the test pins are SDI (scan data in), SCLK (scan clock), SE (scan enable), and FREEZE LO W (freeze - active low).
• SDI scan data in
• SCLK scan clock
• SE scan enable
• FREEZE LO W freeze - active low.
• the Q of one DFT-enabled FF is connected to the SDI of the next DFT-enabled FF in the scan chain.
• Some DFT-enabled FF implementations may include a separate SDO (scan data out) pin. In that case, the SDO of one DFT-enabled FF is connected to the SDI of the next DFT-enabled FF in the scan chain.
• This DFT-enabled FF has 3 modes of operation: normal operation mode, scan shift mode, and capture mode.
• a typical scan test sequence is: scan- in, capture mode, and scan- out. Circuit defects are detected when the scan-out values do not match the expected values for a good circuit.
• Figure 3 shows high-level waveforms associated with this DFT-enabled FF's normal operation mode, scan shift mode, and capture mode. (In a scan chain of N-bit depth, there are N SCLK cycles of scan-in and N SCLK cycles of scan-out.)
• the DFT-enabled FF's SE pin is at a 0 value, and FREEZE LOW is at a 1 value, the DFT-enabled FF is in normal operation mode.
• the DFT-enabled FF When the DFT-enabled FF's SE pin is at a 1 value, and FREEZE LOW is at a 0 value, the DFT-enabled FF is in scan shift mode.
• scan shift mode the asynchronous reset RN and asynchronous preset SN pins on the core flip-flop are forced to a 1 value, independent of the values on R and S.
• the user pins, D and CLK, are also bypassed - SCLK is used to clock SDI data into the core flip-flop.
• the DFT-enabled FF ensures that the core flip-flop is not affected by normal operation (user circuit) values. In other words, the core flip-flop is "frozen" with respect to user circuit inputs.
• the asynchronous reset RN and asynchronous preset SN pins on the core flip-flop are forced to a 1 value, independent of the values on R and S - which means the designer does not need to add DFT-related logic to block asynchronous preset and reset signals.
• SCLK (not the user pin CLK) clocks the scan chains - which means the designer does not need to add DFT-related logic which is required when a user circuit has some flip-flops clocked on the rising edge and some flip- flops clocked on the falling edge, multiple clock domains, internally generated clocks (such as clock dividers), and/or gated clocks.
• the DFT-enabled FF is in capture mode when the DFT-enabled FF exits scan shift mode and enters capture mode in order to observe values on the DFT-enabled FF's user pins.
• exiting scan shift mode is known as "unfreeze” or “unfreezing”
• freeze is known as "freeze” or “freezing”.
• Capture mode can observe user circuit values presented on all user pins (D, CLK, S, and R).
• the state of CLK is inferable through manipulation of the SCLK and FREEZE LOW.
• the DFT-enabled FF's SCLK pin is set to a 0 value and the DFT-enabled FF's CLK pin is a 1 value - a rising edge is automatically generated at the core flip-flop's CLK pin when FREEZE LO W transitions from a 0 value to a 1 value. Since the SE signal is at a 0 value, the core flip-flop clocks in the value presented at its D input.
• Figure 3's capture mode waveforms show the rising edge presented to the core flip-flop's CLK pin (MCLK signal). However, if when the DFT-enabled FF's SCLK pin is set to a 0 value, the DFT-enabled FF's CLK pin is a 0 value, then the core flip-flop's CLK pin (MCLK signal) will remain a constant 0 value (regardless of the FREEZE LOW pulse) as shown in Figure 4.
• the user circuit's values on the R and S pins control the asynchronous reset and preset behavior of the core flip-flop, which enables these functional paths inside the DFT-enabled FF to be verified.
• Figure 6 shows a variation of the DFT-enabled FF.
• the user pins are D, CLK, R, S, and Q. From a user's functional viewpoint (e.g., when the DFT-enabled FF is not in test mode), when R has a 1 value, the core flip-flop is asynchronously loaded with a 0 value. From a user's functional viewpoint (e.g., when the DFT-enabled FF is not in test mode), when S has a 1 value, the core flip-flop is asynchronously loaded with a 1 value.
• This figure shows a DFT- enabled FF whose core flip-flop has preset priority, which means the DFT-enabled FF should also implement preset priority.
• the buffer in the diagram adds sufficient delay to the FREEZE LOW delay signal to ensure that if both R and S have a 1 value when the DFT- enabled FF enters capture mode, the core flip-flop will correctly be loaded with a 1 value.
• Figure 7 shows high-level waveforms for capture mode - specifically when both R and S have a 1 value, resulting in the core flip-flop with a final capture value of 1 (preset priority).
• FIG. 8 and 9 shows an alternative method of a generating a "FREEZE LO W- equivalent" signal.
• Figure 10 shows the truth table associated with this implementation - when either or both CC or CR is set to a 1 value, FREEZE LOW is at a 0 value.
• the test pins are SDI (scan data in), SCLK (scan clock), SE (scan enable), and CC (cut column) and CR (cut row).
• Using two signals to place the DFT-enabled FF into freeze mode provides a convenient way of addressing (selecting) a sub-set of DFT-enabled FFs, when the DFT-enabled FFs are arrayed in a regular 2- dimensional array.
• Adding a NOR gate to DFT-enabled simulated latches, DFT-enabled buffers, and DFT-enabled latches also allow them to be mixed with DFT-enabled FFs for arraying in a regular 2-dimensional array.
• the DFT-enabled FF can also implement latch behavior in a user circuit.
• Figure 11 shows how adding two inverters and 2 NORs to the DFT-enabled FF (previously shown in Figure 2) allows the DFT-enabled FF to behave as a DFT-enabled latch.
• the user pins are Latch D, Latch E, and Q.
• the test pins remain SDI, SCLK, SE, and FREEZE LO W. In both normal operation mode and capture mode, when Latch E has a 1 value, the core flip-flop's value depends on the Latch D value. When Latch E has a 0 value, the core flip-flop holds its current value.
• the DFT-enabled FF can also implement a buffer test point in a user circuit.
• Figure 12 shows how adding one inverter to the DFT-enabled FF (previously shown in Figure 2) allows the DFT-enabled FF to behave as a DFT-enabled buffer test point.
• the user pins are TA and TZ.
• the test pins remain SDI, SCLK, SE, and FREEZE LOW.
• the core flip-flop's value (and TZ pin) is 1.
• the core flip-flop's value (and TZ pin) is 0.
• DFT-enabled buffer test points are useful increasing the testability of a user circuit's without modifying the user circuit's functionality.
• the DFT-enabled FF that implements latch behavior (shown in Figure 11) will be referred to as a DFT-enabled simulated latch, while the DFT-enabled FF that implements a buffer test point (shown in Figure 12) will be referred to as a DFT-enabled buffer.
• Figure 13 shows a DFT-enabled latch embodiment that behaves like a flip-flop during scan shift, but behaves like a latch during normal operation and capture mode.
• the user pins are D, E, and Q.
• the test pins remain SDI, SCLK, SE, and FREEZE LOW.
• DFT-enabled FFs and DFT-enabled latches can be placed within the same scan chain. If the user pin E is tied high, then the DFT-enabled latch can be used as a DFT-enabled buffer.
• Embodiments are envisioned that consolidate some of the DFT-enabled FF' s logic into integrated cells.
• a mux-scan flip-flop or mux flip-flop can be used instead of a discrete mux and flip-flop.
• Other envisioned embodiments consolidate all of the DFT-enabled FF's logic into one integrated cell.
• the DFT-enabled FFs may be implemented with only an asynchronous reset, or with only an asynchronous preset, or without any asynchronous reset or preset.
• the appropriate NAND gate(s) and buffer would not be required and can be removed as an optimization.
• the muxes shown in Figures 11 and 12 can also be removed as an optimization.
• Embodiments are next described where the DFT-enabled FF, DFT-enabled simulated latch, DFT-enabled buffer, or DFT-enabled latch circuits are combined with addressable control and partition test.
• This testing method can test user circuits without requiring complex DFT rules or any user-inserted test structures.
• This testing method can also test "unstable" circuits; two examples of unstable circuits are shown in Figures 14 and 15. These circuits are usually considered non-deterministic and untestable from a normal scan test perspective.
• the flip-flop oscillates between 0 to 1 and 1 to 0.
• the flip-flop's user pins are shown because this scenario applies to flip-flops in general. If the flip-flop output has a 0 value, then the asynchronous preset S pin is asserted and the flip-flop is loaded with a 1 value. If the flip-flop output has a 1 value, then the asynchronous reset R pin is asserted and the flip-flop is loaded with a 0 value. A known value can be scanned into the flip-flop. However, the capture (observed) value for this flip-flop is not deterministic because of speed variations caused by manufacturing, temperature and voltage.
• Figure 15 shows another example of a non-deterministic circuit - in this case, a ring oscillator.
• the value of OZ switches from 0 to 1 and 1 to 0.
• the value of OZ at a specific time X after power is applied to the circuit is non- deterministic because the initial conditions at power up cannot be controlled sufficiently. If OZ is driving a flip-flop's user clock pin, then that flip-flop's value at time X also cannot be determined, which means the capture (observed) value for this flip-flop is also non-deterministic.
• loop breaking a concept known as "loop breaking" is introduced.
• component types within a circuit assigned into three different categories: O break, 1 break, and 2_break. For counting purposes, O breaks are ignored, while two 1 breaks is equivalent to one 2_break.
• DFT-enabled simulated latches Figure 11
• DFT-enabled buffers Figure 12
• DFT-enabled latches Figure 13
• DFT-enabled FFs are l break cells, when encountered through their user asynchronous preset or reset pins.
• DFT-enabled FFs are 2_break cells, when encountered through their user clock or D pin.
• the entire user circuit is traversed using graph traversal techniques (that are generally known in the art), while keeping track of the number of "breaks" encountered.
• a loop is detected, if a cycle exists (i.e., a node is visited more than once) and the number of breaks is less than two. If the number of breaks is zero, then insert two DFT-enabled buffers to break the loop.
• Figure 16 shows how two DFT-enabled buffers break the loop, without affecting the user circuit's functionality - the user circuit still oscillates. If the number of breaks is one, then insert one DFT-enabled buffer to break the loop.
• Figure 17 shows how one DFT-enabled buffer breaks a loop, without affecting the user circuit's functionality - the user circuit still oscillates. To simplify Figures 16 and 17 only the user pins from each DFT- enabled buffer is shown. Figure 16 also shows only the user pins of the DFT-enabled FF. By inserting DFT-enabled buffers, unstable circuits now have deterministic capture values and are also more testable.
• DFT-enabled FF when the DFT-enabled FF unfreezes, it becomes sensitive to the values on the user pins.
• the DFT-enabled simulated latch, DFT-enabled buffer, and DFT- enabled latch also become sensitive to values on the user pins when they unfreeze.
• all concepts and capabilities attributed to the DFT-enabled FF also apply to the DFT-enabled simulated latch, DFT-enabled buffer, and DFT-enabled latch.
• DFT-enabled FFs are divided into partitions using the following approach:
• Two DFT-enabled FFs are considered "adjacent" if the output from one will influence the state of the other. For example, two DFT-enabled FFs are adjacent if (1) there are connection lines directly from the Q output of one DFT-enabled FF to any user pin on the second DFT-enabled FF or (2) there is connection as described in (1) but with additional interposition of combinational logic only.
• Partitions are determined by graph coloring techniques that are generally known in the art.
• the DFT-enabled FFs within the tested partition can enter capture mode, while the DFT-enabled FFs in all other partitions must remain frozen.
• capture values no longer depend on the order that each DFT-enabled FF unfreezes. In other words, partition test guarantees deterministic capture values.
• Addressable control is a method of that enables DFT-enabled FFs (of the type described herein) to be selected individually, in groups, or in entirety to operate in normal operation mode, scan shift mode, and capture mode.
• the address selection can be pre-designed or can be hard-wired.
• One approach of a pre-designed addressing scheme utilizes row and column addressing, where each DFT-enabled FF is assigned a unique row and column address. Row and column addressing would be efficient in a logic array or gate array. Row and column addressing enables testing of any user-designed circuit that can implemented in the logic array (or gate array); prior knowledge of the user-designed circuits is not required.
• Figure 18 shows a simplified diagram of a DFT-enabled FF - with the following test pins SDI, SCLK, FREEZE LOW and the output pin Q is shown.
• the test pin SE and the user pins are not shown.
• Figure 19 shows how the simplified DFT-enabled FFs (of Figure 18) can be placed in a two-dimension array and selected using row and column addressing.
• there are M+l columns (numbered right to left: 0 to M) and there are N+l rows (numbered bottom to top: 0 to N).
• the DFT-enabled FFs in a column are connected into scan chains.
• Each column has its own column select line (numbered from right to left: CC O to CC M), while each row has its own row select line (numbered from bottom to top: CR O to CR N).
• a row is frozen by setting that row's select line to a 1 value
• a column is frozen by setting that column's select line to a 1 value.
• a DFT-enabled FF is frozen when either or both of its row select line or column select lines is at a 1 value.
• Figures 8 and 9 show two examples of DFT-enabled FFs that are compatible with this row and column addressing method. In this case, each DFT-enabled FF' s CC test pin would be connected to the appropriate column select line, and each DFT-enabled FF' s CR test pin would be connected to the appropriate row select line.
• DFT-enabled FFs By using the row and column select lines, DFT-enabled FFs can be addressed (frozen or unfrozen) individually, in groups, or in entirety. When a partition is being tested, only the DFT-enabled FFs in that partition can be selected to enter capture mode.
• DFT-enabled FFs are not arranged in rows and columns - for example, in a standard cell ASIC - hard-wired addressing can be used to select individual, groups or all of the DFT-enabled FFs in the user-designed circuit.
• Hard-wired addressing requires prior knowledge of the user-designed circuit.
• the DFT-enabled FFs in the user-designed circuit are divided into partitions. Each partition requires a separate and distinct FREEZE LOW signal. All the DFT-enabled FFs in a given partition share the same FREEZE LOW signal.
• every partition's FREEZE LOW has a 0 value.
• capture mode only one partition's FREEZE LOW switches between 0 to 1 to 0 values.
• Fig. 20 illustrates a DFT-Enabled FF 2000 that eliminates the need for partitioning circuits, simplifies loop breaking, and reduces the vector count required to test circuits.
• the Nor gate 935 that is used to generate FREEZE LOW from CC and CR in the DFT- Enabled FF 900 of Fig. 9 is eliminated, and instead a single freeze low signal is brought into the cell 2000.
• the output of flip flop 2025 is run through 2- 1 mux 2040 that is controlled by a new signal called freeze output.
• the other input of the mux 2040 is connected to the output of latch 2030 which has it's D input connected to the output of the new mux 2040 and it's enable input connected to the freeze output signal.
• Fig. 21 illustrates the timing of the DFT-Enabled FF 2000 of Fig. 20.
• the freeze output line is held high, and during the Capture mode cycle freeze output is low and encloses the freeze low pulse.
• Loop breaking also gets simpler when using the DFT-Enabled FF 2000 of Fig. 20, because now a flip flop that feeds back to it's own set or reset pin is not a problem and does not need a loop breaking element added. Pure combinational loops must be still broken but they only require a single loop breaking element under all circumstances.
• interconnect and function-unit bit-widths, clock speeds, and the type of technology used are variable according to various embodiments in each component block.
• the names given to interconnect and logic are merely exemplary, and should not be construed as limiting the concepts described.
• the order and arrangement of flowchart and flow diagram process, action, and function elements are variable according to various embodiments.
• value ranges specified, maximum and minimum values used, or other particular specifications are merely those of the described embodiments, are expected to track improvements and changes in implementation technology, and should not be construed as limitations.

Landscapes

• Engineering & Computer Science (AREA)
• General Engineering & Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Theoretical Computer Science (AREA)
• Computer Hardware Design (AREA)
• Quality & Reliability (AREA)
• Tests Of Electronic Circuits (AREA)
• Semiconductor Integrated Circuits (AREA)

Priority Applications (3)

Applications Claiming Priority (4)

Publications (1)

Family

ID=38832108

Family Applications (1)

Country Status (5)

Cited By (2)

Families Citing this family (22)

Citations (2)

Family Cites Families (19)

• 2007
  • 2007-06-09 US US12/303,938 patent/US8122413B2/en active Active
  • 2007-06-09 JP JP2009514561A patent/JP5131997B2/ja active Active
  • 2007-06-09 WO PCT/US2007/070821 patent/WO2007146849A1/en not_active Ceased
  • 2007-06-09 KR KR1020097000531A patent/KR101006822B1/ko active Active
  • 2007-06-09 EP EP07798355A patent/EP2030114B1/en active Active

Patent Citations (2)

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events