Methods And Apparatus For Testing Electronic Circuits
Methods and apparatus are provided for testing to determine the existence of defects and faults in circuits, devices, and systems such as digital integrated circuits, SRAM memory, mixed signal circuits, and the like. In particular, methods and apparatus are provided for detecting faults in circuits, devices, and systems using input control signals to generate controlled-duration, controlled pulse-width, transient power supply currents in a device under test, where said transient power supply currents are of controllable bandwidth and can be used as observables to determine faulty or defective operation. Additionally, methods and apparatus are provided to permit high bandwidth sensing of transient supply currents as need to preserve the narrow widths of these current pulses. These methods may include autozero techniques to remove supply current leakage current and DC offsets associated with practical current sensing currents. The sensed transient supply currents can be compared to single or multiple thresholds to assess normal or faulty or defective operation of the device under test.
Latest University of North Carolina at Charlotte Patents:
- Photodetector focal plane array systems and methods based on microcomponents with arbitrary shapes
- Method and apparatus for generating high current, fast rise time step-functions
- Fiber based finishing tools
- Battery energy storage system controller systems and methods
- SELECTIVE LASER VAPORIZATION OF MATERIALS
The present application is a divisional of U.S. patent application Ser. No. 11/593,655, filed Nov. 7, 2006, which is a divisional of U.S. patent application Ser. No. 10/900,915, filed Jul. 28, 2004, now U.S. Pat. No. 7,148,717, which is a divisional of U.S. patent application Ser. No. 10/237,670, filed Sep. 10, 2002, now U.S. Pat. No. 6,833,724, which claims priority to U.S. Provisional Application No. 60/318,599, filed Sep. 10, 2001, the entire contents of all of which are incorporated herein by reference.FIELD OF THE INVENTION
The present invention relates to methods and apparatus for testing of faults and defects within electronic devices, such as semiconductor memory circuits.BACKGROUND OF THE INVENTION
Electronic devices such as memory chips can contain undesired faults and defects that arise due to imperfections in manufacturing processes. These faults and defects include open circuits, short circuits, and out-of-tolerance components. Testing for these faults during manufacture is impractical and costly in light of the new process technologies currently being used to manufacture electronic circuits and devices. Therefore, self-test circuits have been incorporated within modern circuits and devices in order to improve reliability, to eliminate defective and faulty circuits and devices, and to reduce cost.
One example of an incorporated self-test circuit measures the quiescent DC power supply current. This method is referred to as the IDDQ method. In the IDDQ method, power supply currents outside a predetermined range indicate a faulty circuit or device. Examples of this method can be found in U.S. Pat. Nos. 6,342,790, 6,301,168, and 6,144,214. In general, an apparatus incorporating quiescent DC power supply current for self-test monitors the current from a power supply using a current sensor. Therefore, the apparatus effectively monitors the supply current delivered to the Device Under Test (DUT) and transmits a final output signal that is in proportion to the supply current delivered to the DUT. The final output signal is analyzed to determine whether or not any faults exist in the DUT.
The IDDQ method, however, requires quiescent defect currents of appreciable magnitude relative to the quiescent currents in fault-free circuits and devices. But the difference between faulty currents and fault-free currents may be negligible in many components. Moreover, quiescent leakage currents associated with emerging technologies are large enough to render these IDDQ methods ineffective in detecting faulty circuits.
These emerging technologies include deep sub-micron integrated circuit technology. Sub-micron circuits have increased leakage currents and associated large DC supply currents that greatly reduce the effectiveness of the IDDQ methods. Additional challenges to testing of sub-micron circuits include significant increases in circuit size, rapid increases in clock frequencies, the dominance of the interconnect delay, the transmission-line behavior of interconnects, the reduced level of the power supply voltage, the increase in leakage currents, the increase in power consumption, and the increased sensitivity of circuit arrangements to process defects. Therefore, fault monitoring circuitry was developed for measurement of the transient power supply current. Methods employing such circuitry are referred to as IDDT methods wherein transient power supply currents outside a predetermined range indicate a faulty circuit or device.
U.S. Pat. No. 6,414,511 discloses an example of an IDDT method. In general, an apparatus incorporating transient power supply current for self-test monitors the current from the power supply using a current sensor. Therefore, the apparatus effectively monitors the transient supply current delivered to the Device Under Test (DUT) and transmits a final output signal that is in proportion to the transient supply current of the DUT. The final output signal is analyzed to determine whether or not any faults exist in the DUT.
Prior implementations of the IDDT methods, however, do not adequately address the need for large bandwidths to handle the extremely short duration underlying transients in deep sub-micron technology. These prior methods have relied on dubious and inadvertent integration of the high-speed transients (i.e., lowpass filtering of transients) to allow slower circuits to process the transients. In such methods, the extremely high-speed transients native to the DUT are the transients being observed and measured for the purpose of detecting faulty circuits. The associated high-speed transients have led to over-simplified and questionable solutions, which have not found application in industry. Moreover, the bandwidth of such transients in digital circuits are by their very nature faster than the highest clock speeds of the circuit, since the duration of the transient is commonly shorter than the logic rise and fall times.
Therefore, a need exists to find alternative methods and apparatus that effectively and practically utilize transient power supply currents in testing devices and circuits for faults and defects. Suitable methods and apparatus should be capable of testing with little disruption of the circuits under test, with little added hardware, and with minimal effect on manufacturing costs. In addition, suitable methods and apparatus should be capable of the necessary testing at lower speeds and bandwidths to be advantageous for implementations in circuits, devices, and systems.SUMMARY OF THE INVENTION
The present invention is directed to apparatus and methods for detecting faults in circuits, devices, and systems. Suitable circuits, devices and systems include integrated circuits for memory cells, combinational logic, mixed-signal circuits, analog circuits, and the receiver and transmitter portions of cellular phone handsets. In memory chips, the transient supply current apparatus and methods can be used to detect faulty memory cells and memory chips. Similarly, transient supply current apparatus and methods can be used to detect faulty devices and sub-circuits within combinational logic, mixed-signal circuits, and analog circuits.
In one embodiment of the present invention, an apparatus for controlled pulse-width transient power supply testing is used. The apparatus includes a current measuring device, a Device Under Test (DUT) such as a Static Random Access Memory (SRAM) cell or a bank of memory cells, and input control signals including a test-pulse generator and control signals to set signal levels within the DUT. Suitable current measurement devices are capable of measuring the transient current delivered to the DUT. The test-pulse generator includes at least one output that is electrically coupled to and in communication with the DUT. In operation, the test-pulse generator controls the DUT to generate a transient power supply current within a sub-circuit of the DUT. This transient power supply current is of a controlled pulse-width and duration. The generated transient current is monitored and measured by the current measurement device. The measured transient current is then analyzed for indications of defects or faults in the DUT.
In another embodiment of the present invention, digital portions of the DUT are analyzed. In this embodiment, the DUT is exposed to a plurality of input control signals. One or more signals from this plurality of input control signals are selected and set to cause signal levels internal to the digital portions of DUT to have values that are intermediate to the logic levels. For example, the logic levels are about 0 volts and about 5 volts and the intermediate voltage level within the digital portion is about 2.5 volts. These intermediate values in conjunction with the other input control signals help induce DUT power supply current of a controlled pulse-width and duration that is then monitored for deviations from values for fault-free devices, such deviations indicating a faulty circuit.
In another embodiment of the present invention, the apparatus includes a first resistor which is coupled into the supply current lead of the DUT, and a high pass filter containing a capacitor and a second resistor is electrically coupled to the DUT supply line between the DUT and the first resistor. The second resistor is then electrically coupled to a cascade of a plurality of wideband voltage amplifiers. The output of the amplifiers is electrically coupled to a comparator circuit. In operation, the first resistor converts the supply current transients to voltage transients. The current transients can be native to the DUT or deliberately introduced. These voltage transients are then AC coupled through the capacitor and second resistor to reject the DC voltage associated with the high leakage current while preserving the shape and magnitude of the voltage pulses. These voltage pulses, i.e. the IDDT supply current pulses, are amplified by the plurality of differential, wideband amplifiers. The comparator assesses the output from the amplifiers to determine normal or abnormal operation of the DUT.
The apparatus and methods of the present invention provide the unexpected results and advantages of direct observability into the switching profile of the DUT. In addition, the test time is significantly reduced since it is not necessary for a circuit to reach steady-state or static supply current operation following test stimuli. The apparatus and method of the present invention eliminate measurement problems associated with quiescent current and are particularly well suited for emerging sub-micron technologies. Since faults can be assessed directly form the transient supply current, the need for separate read operations is eliminated. The apparatus and methods of the present invention detect a wide range of faults and defects including resistive opens and shorts, capacitive opens, pattern sensitive faults, tunneling, and crosstalk. The method and apparatus of the present invention substantially reduce test costs and improve fault coverage. The need to implement costly test methods such as separate test pins or supply partitioning to overcome limitations of large leakage currents is eliminated.
The method of the present invention can be used to test various types of electronic circuits and devices and is particularly well suited for sub-micron circuits and devices. Referring initially to
The high bandwidth of transient power supply currents associated with the DUT 300 are illustrated in
The present invention is directed to apparatus and methods for determining defects and faults in electronic devices having at least one input that is coupled between a power supply and a ground to receive a supply current therefrom. In one embodiment of the present invention, as illustrated in
In operation, the input control signal bus 525 delivers one or more of the plurality of input control signals to the DUT 506, causing a first plurality of devices, components, or circuits within DUT 506 to conduct current. The voltage and current levels of the control signals are adjusted to affect varying levels of currents in the DUT 506 and to selectively choose to activate different devices or components within the DUT 506 as desired to generate transient supply current pulses capable of being observed at the second output of the current sensor. Either signal currents or signal voltages can be used. In addition, the input signals can be adjusted to control the pulse-width of the current conducted within the DUT. Therefore, the devices, components, or circuits within the DUT will conduct the current for the duration of the defined controlling pulse-width. The sum of the conducted currents from this first plurality of devices is the current delivered to the DUT through the supply current line 524. The current sensor 504 monitors this supplied current and outputs a proportional final output signal through the output signal lead 526. The final output signal is monitored and analyzed for defects or faults in the devices, components, or circuits selected to conduct current. In an alternative embodiment, one or more controlled pulse-width signals on signal bus 525 can be used. In addition, quiescent or DC components appearing in final output signal 526 can be removed or accounted for when analyzing for faults.
The DUT also includes at least one digital portion. In an embodiment for testing digital portions of the DUT 506, selected signals from the plurality of signals on the input control signal bus 525 are typically set to force voltages within those digital portions to be intermediate levels. The digital portion is responsive to a first voltage corresponding to a logic low state and a second voltage corresponding to a logic high state. The input control signals can be selected to expose the digital portion to a third voltage between the first voltage and the second voltage. For example, the intermediate voltage level can be about 2.5 volts for a logic low voltage level of about 0 volts and a logic high level of about 5 volts).
In another embodiment of the present invention, as illustrated in
In one embodiment, the final output signal is proportional to the power supply current multiplied by a selected negative constant. Quiescent or DC components appearing in the final output signal may be removed or appropriately processed or accounted for during analysis of the output signal. For purposes of illustration, the output signal 626 is assumed to not load or drain current from the remainder of the circuit, and for nominal purposes the resistance of sensing resistor 604 is 1 ohm. If the SRAM 606 has a schematic and pin diagram corresponding to those shown in apparatus 300 of
The voltage and current levels delivered to the SRAM through the first and second control signal leads 625, 627 can be adjusted to affect current levels in the SRAM 606 and to select different devices in the SRAM 606 to activate. In an alternative embodiment, signals delivered to the SRAM through the first signal lead 625 can be held at a constant 5 volts (logic high), and the input signal delivered to the bit and bitbar pins through the second signal lead 627 can be set to pulses with predetermined amplitude and predetermined DC bias level, so as to effect correspondingly similar transient controlled-duration power supply current pulses.
The electrical performance of this arrangement 600 is illustrated in
Next, one or more logic control signals are applied as desired to one or more inputs of the DUT 709 to selectively activate circuits of interest within the DUT to cause a controlled-duration transient pulse of power supply current through selected components. Both signal voltages and signal currents can be used. The voltage and current levels of the control signals can be adjusted to affect the levels of currents in the DUT to selectively choose to activate different devices, components or circuits therein. The second signal output is monitored and analyzed, and if the second output signal of the current sensor is within normal circuit variation, the circuit is determined to not be defective 710. The nominal values of current delivered to the DUT will have some variation in manufacture, necessitating a range of current values for non-defective or fault-free circuits. If, however, the second output signal is unacceptably outside normal circuit variation, the circuit is determined to be defective 711.
In another embodiment of the present invention, the supply current transients are converted to voltage transients through a resistance, either parasitic or deliberately introduced, connected between the voltage supply and interior circuitry or components of the DUT. These small voltage transients or pulses are then AC coupled to reject the DC voltage associated with high leakage current while preserving the shape and magnitude of the voltage pulses. The voltage pulses, representative of the IDDT supply current pulses, are amplified by a cascade of a plurality of differential, wideband amplifiers. In one embodiment, four cascade amplifiers are used, each having a voltage gain of nearly four and a 3-dB bandwidth of 1.6 GHz. The composite voltage gain and bandwidth of the four amplifiers is approximately 233 and 550 MHz respectively, implying a very high gain-bandwidth product of 128 GHz. The high gain, combined with the rejection of DC leakage current through AC coupling, permits the sensor to detect small, e.g. 180 μA peak, transient supply currents in the presence of tens of milliamps of supply current leakage. This high bandwidth permits the sensor to assess supply current pulses having widths of approximately 1 ns, while permitting periodic test speeds of 200 MHz. This method is suitable for use in the testing of general analog, mixed-signal, and digital circuits in deeper submicron CMOS processes and emerging integrated circuit processes.
The high pass network is electrically coupled to a cascade of four, wideband voltage amplifiers 717, 718, 719, 720 to amplify the voltage pulse. Suitable amplifiers have a nominal gain of 4 and a 3 dB bandwidth of 1.5 GHz for a composite gain of approximately 233 at a 3 dB bandwidth of 550 MHz. In one embodiment, a comparator 721 is coupled to the output of the cascade amplifiers to assess the sensor output voltage and to determine normal or abnormal operation of the DUT. In another embodiment, multiple comparators are electrically coupled to the amplifier output. In yet another embodiment, an analog-to-digital (A/D) converter is electrically coupled to the amplifier output to provide for more comprehensive test evaluation. The current sensor illustrated is suitable for use with any sub-micron type DUT for example either a 0.35 μm or a 0.18 μm CMOS process. The high voltage gain combined with wide bandwidth permits processing of small IDDT current pulses with little bandwidth loss or corresponding increase in pulse width. In addition, the wide bandwidth performance permits IDDT testing at a repetition or clock rate of about 200 MHz. The apparatus also includes a first diode-connected NMOS device 722 electrically coupled between the power supply 714 and the first amplifier 717. The NMOS device provides a DC input common-mode level for the first amplifier that is appropriately below the power supply voltage. This input common mode level is VDD−VGS1, where VDD is the supply voltage and VGS1 is the operating gate-source voltage of the first NMOS device 722. A second diode-connected NMOS device 723 is coupled between the power supply 714 and each amplifier to establish internal operating bias current for the four amplifier stages.
The amplifiers are used to amplify the transient voltage pulse to a level permitting an assessment of the IDDT level. As is shown in
where gm726 (gm727) is the transconductance of MOS device 726 (727), gds728 (gds729) is the output conductance of MOS device 728 (729), and gm730 (gm731) and gmb730 (gmb731) are the transconductance and body-effect transconductance of MOS device 730 (731). The first term of (1) is the voltage gain of the differential pair (726 and 727) driving resistive loads consisting of 728 and 729 that operate in the ohmic, linear, or triode region. The second term is the voltage gain of the source followers where some gain loss below unity occurs due to the presence of body-effect transconductance. The selection of amplifier stage gain and bandwidth depends upon the particular test application and includes factors such as the gain required and bandwidth required to preserve the narrow transient voltages. Additionally, the number of amplifier stages depends upon the test application with higher sensitivity applications (e.g., SRAM testing) requiring more stages compared to lower sensitivity applications (e.g., combination logic testing) requiring less stages.
In one embodiment, the apparatus and methods of the present invention includes autozeroing circuitry and steps to remove the DC errors associated with DC supply current leakage or the DC offset errors associated with practical implementations of amplifiers or a cascading of amplifiers used to amplify the signal associated with the transient supply current. In one embodiment, the autozeroing circuitry includes capacitive autozero circuits where the DC error associated with the DC supply current leakage or the DC offset errors associated with the amplifiers or cascading of amplifiers is stored on a capacitor or capacitors during an autozero phase. In a subsequent signal phase, the desired transient supply signal is coupled through the capacitor or capacitors while removing the undesired dc error associated with DC supply current leakage and/or DC offset errors. The autozero and signal phases can be controlled by the digital system clocking or otherwise controlling the DUT or DUT's. In another embodiment, the autozero circuitry includes continuous-time or gated negative feedback circuits where any amplifier output or outputs of cascaded amplifiers are compared to a reference voltage or reference voltages, including ground, in a circuit or circuits that then apply the required signal or signals to cancel the DC error. This negative feedback correction of the DC error can be applied at each stage of a cascaded amplifier or at selected stages. Gated negative feedback autozeroing can be enabled during an autozero phase and disabled during a signal phase to prevent any undesired cancellation of the desired transient supply signal. These phases can be controlled using the digital system clocking within the DUT or by otherwise controlling the DUT or DUT's. Continuous-time negative feedback autozeroing requires no additional control and can run continuously.
In one example of autozeroing, a CMOS amplifier stage has a DC input offset voltage associated with the mismatches in the MOS devices. This DC input offset will be approximately 2 to approximately 20 mV (3 sigma), a level that is well above the transient input voltage pulse. Therefore, some form of a MOS input autozero circuitry is preferred for the amplifiers. Suitable autozero circuitry will store the input offset voltage across a capacitor during the autozero mode. In the signal mode, this capacitor would be placed in series with the voltage pulse to be measured and the amplifier input, thereby passing the narrow voltage pulse while subtracting out the input offset voltage. The two modes of operation, autozero and signal, can be controlled by the clocking system associated with testing of digital circuits. Here, the autozero mode can be invoked while digital test stimuli are being initialized. During actual IDDT measurement, the MOS autozero circuit can be placed in the signal mode.
By way of example, for an SRAM size of 2L words and a cell group size of 2M words, where the word size can be 8-512 bits wide, the sensor selection is accomplished by decoding the L-M high-order bits of the address lines. This decoding can be performed by the address decoder 735. A Pass/Fail logic block 736 is provided to process the outputs of the sensors 733 into a Pass/Fail indication. SCS and MCS represent Single Cell Switching and Multiple Cell Switching conditions respectively, representing different fault conditions.
In another embodiment of the method of the present invention, the difference in the peak values of two transient supply voltages, IDDT, are compared to stored values for this difference associated with fault-free circuits. Deviations from the stored difference values indicate a fault or defect in the tested circuits. As is illustrated in
The peaks are represented by bars extending upwards between a lower current level 751 and an upper current level 752. There is no transient current response less than the lower current level 751 or greater than the upper current level 752 for the given vectors. Therefore, complementary pairs of current peaks have a specific difference in value. The upper and lower current levels define and bound this difference between a lower value and an upper value. As illustrated, the difference is bound between 0 and 112 μA. In order to test a device, the device is exposed to one or more of the complementary pairs of vectors associated with a pair of baseline transient current peaks, for example 753,754, and the generated complementary values of IDDT are measured. The differences between the generated transient current pairs are calculated. If these calculated differences are within the upper and lower bounds of differences as shown for the fault-free device, the tested device passes. If the calculated differences are outside the upper and lower bounds, then the tested device fails, indicating a defect or fault.
This method provides the unexpected results and advantages that the difference between transient current peaks cancels the quiescent part of the current response, and hence reduces the effect of leakage currents. In addition, a defective device circuit can draw IDDT spikes with peak magnitudes that are within the good circuit limits or slightly outside the limits; however, the difference of such peak magnitudes can amplify the abnormal behavior of IDDT and produce a value outside the good circuit limits as defined by the difference in IDDT spikes. For example as illustrated in
In an alternative embodiment of the method of the present invention utilizing the upper bound 752 and the lower bound 751 of the peak transient currents illustrated in
Many modifications and other embodiments of the invention will come to mind to one skilled in the art to which this invention pertains having the benefit of the teachings presented in the foregoing descriptions and the associated drawings. Therefore, it is to be understood that the invention is not to be limited to the specific embodiments disclosed and that modifications and other embodiments are intended to be included within the scope of the appended Claims. The embodiments set forth herein are provided to fully convey the scope of the invention to one of skill in the art. Although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.
1. An apparatus for determining defects and faults in electronic devices having at least one input that is coupled between a power supply and a ground to receive a supply current therefrom, the apparatus comprising:
- a current sensor electrically coupled between the power supply and a device under test, the device under test comprising an SRAM device, and the current sensor comprising: at least one input electrically coupled to the power supply; a first output electrically coupled to the device under test and capable of supplying a supply current voltage to the device; and a second output capable of delivering a signal proportional to the supply current delivered to the device under test.
2. An apparatus for determining defects and faults in electronic devices having at least one input that is coupled between a power supply and a ground to receive a supply current therefrom, the apparatus comprising:
- a resistor electrically coupled between the power supply and a device under, the resistor comprising: at least one input electrically coupled to the power supply; a first output electrically coupled to the device under test and capable of supplying a supply current voltage to the device under test; and a second output capable of delivering a signal proportional to the supply current delivered to the device under test.
3. A method for determining faults or defects in an electronic device that is coupled between a power supply and a ground to receive supply current therefrom while the device is in operation, the method comprising:
- exposing a fault-free device to a plurality of pairs of complementary control signal vectors;
- generating a plurality of complementary pairs of baseline transient supply currents associated with the signal vectors;
- exposing the device under test to one or more of the complementary pairs of vectors;
- measuring the transient supply currents generated;
- comparing the generated transient supply currents to the baselines currents; and
- determining if the generated transient supply currents are within a normal circuit variation of the baseline transient currents.
4. The method of claim 3, further comprising:
- calculating a baseline difference for each complementary pair of baseline transient currents;
- calculating a test difference for each complementary pair of generated transient currents; and
- comparing the baseline differences to the corresponding test differences.
Filed: Dec 22, 2009
Publication Date: Apr 22, 2010
Applicant: University of North Carolina at Charlotte (Charlotte, NC)
Inventors: David M. Binkley (Charlotte, NC), Rafic Zein Makki (Charlotte, NC), Thomas Paul Weldon (Charlotte, NC), Ali Chehab (Beirut)
Application Number: 12/644,804
International Classification: G01R 31/02 (20060101);