Electricity: measuring and testing – Fault detecting in electric circuits and of electric components – Of individual circuit component or element
Reexamination Certificate
2000-05-23
2001-01-16
Metjahic, Safet (Department: 2858)
Electricity: measuring and testing
Fault detecting in electric circuits and of electric components
Of individual circuit component or element
C324S1540PB
Reexamination Certificate
active
06175246
ABSTRACT:
FIELD OF THE INVENTION
The present invention is related generally to integrated circuit testers, and more specifically to testing AC functional characteristics, particularly propagation delay times, of high speed integrated circuit devices.
BACKGROUND OF THE INVENTION
A circuit tester applies digital signal patterns called test vectors to the electrical connections (pads or pins) of a device under test (“DUT”). A schematic diagram of an available tester is provided in FIG.
1
. Tester
1
comprises a central processing unit (“CPU”)
10
connected to external elements such as terminal
11
, printer
12
, and port
13
for establishing a link with other apparatus such as a communication link with a host computer. CPU
10
is also connected to bus
14
for transmitting and receiving data to and from various tester elements.
The DUT is connected to the tester through a contactor socket (not shown) and a load board
15
. The input/output connections of the socket normally correspond to the pins of the device to be tested, each pin being connected to a specific card referred to as an electronic pin
16
(only one of which is shown in
FIG. 1
) which constitute in plurality a set of boards, arranged generally as a ring. Pins
16
comprise a bidirectional link with a test vector memory
17
and a specific link with a terminal of load board
15
. Pins
16
also provide access to common circuits for providing reference voltages
19
and time delays
19
′ via general bus
14
. Each electronic pin
16
comprises amplitude and time shaping circuits for adjusting the signals received from memory
17
to match the references provided by circuits
19
and
19
′. Load board
15
may comprise buffer impedances and read relays for connecting the integrated circuit pins of the DUT either to an electronic pin
16
or to central measuring unit (CMU)
18
which applies and measures currents and voltages.
As integrated circuit devices become faster, the propagation delay (TPD) through a device (e.g., from an input pin to an output pin) must be more precisely measured to ensure accurate verification of device operation. Referring to
FIG. 2
, TPD can be defined as the length of time that passes from point B in the timing diagram for signal DUT
IN
, to point C in the timing diagram for signal DUT
OUT
. Since the specifications for an especially fast device might call for a very small TPD (e.g., 5 nanoseconds), and the propagation delay of the path through the testing apparatus itself (tester delay from point A to point B in
FIG. 2
) may be comparable to the TPD of the device (e.g., 8 nanoseconds), the tester delay must be accurately measured and accounted for to properly measure TPD at a fine scale. Too much error in tester propagation delay measurement leads to unreliable TPD measurement, thereby potentially decreasing the number of verified-TPD high-speed devices yielded after testing. Also, it is critical to place points B and C at the appropriate point in a rising or falling skew line, to ensure accurate pass/fail measurements during testing.
Referring next to
FIG. 3
, it is presently known to use comparator circuit
20
to measure tester propagation delay through a tester signal path (between the pin electronics within the test head (not shown), through connections
44
to the load board
15
, contactor
40
, and DUT
42
) using a technique commonly referred to as Time Domain Reflectometry (TDR). In a TDR measurement, a step waveform signal is sent through a test channel from driver
30
in the test head to a high impedance load (e.g., the open socket) in load board
15
. The signal reflects from the open socket and returns to its source. The elapsed time is measured when the reflected waveform returns to its source, and the tester path delay is half the measured elapsed time.
While TDR can provide valuable delay information, its accuracy is limited by the specifications of comparator circuit
20
, which is illustrated in detail in FIG.
4
. In comparator circuit
20
, digital to analog converter (DAC)
24
provides a reference voltage V
OL
for comparison to driver-supplied input signal V
OUT
by comparator
26
. Driver-supplied input signal V
OUT
is provided by driver
30
in FIG.
3
. The output of comparator
26
is then forwarded to logic comparator
22
. Each time logic comparator
22
receives an activation signal from strobe
28
, logic comparator
22
provides either a logic high or logic low signal depending upon the output of comparator
26
. Strobe
28
is activated at a precisely known time, thereby accurately indicating the logic level of the signal forwarded by comparator
26
.
There are two variables that determine the accuracy of comparator circuit
20
: the step size (also known as comparator voltage resolution) of the test stimulus digital signal received by DAC
24
, and the precise timing of the strobe signal received from strobe
28
. Strobe resolution is often the limiting factor in testing error. While strobe resolution is limited to about a 0.1 nanosecond step (100 picoseconds, two percent of a 5 nanosecond delay), presently available reference voltage resolution (generated by DAC
24
and forwarded to comparator
26
in
FIG. 4
) is generally about a 2.5 millivolt step for a 0 to 3 volt signal (only about 0.08 percent error from coarse resolution). In device lots having a TPD standard deviation of 100 picoseconds or less, the error created by the coarse resolution of strobe signals is therefore significant and can lead to costly reductions in yield. Therefore, there is a need in the integrated circuit testing field for a method of increasing the effective resolution of test strobe signals.
SUMMARY OF THE INVENTION
A method and system are provided for increasing the accuracy of AC parametric testing. The invention provides for the derivation of highly accurate response characteristics through linear interpolation and extrapolation. The invention further allows for the utilization of test signals having almost any discrete time value and voltage magnitude by modifying the input signals and output testing points according to extrapolated voltage and timing values, thereby substantially reducing measurement error, increasing test efficacy, and boosting high-speed device yield.
The invention provides a method of precisely estimating tester delays (signal propagation delay time in an integrated circuit testing apparatus), wherein a plurality of tester delay measurements are taken and additional delays are estimated by linearly interpolating the measured delays.
The invention also provides a method of precisely estimating propagation delays through a device via linear extrapolation. A desired test point (desired output voltage at a given time) is established. Using a sample device, output voltage measurements are taken at a plurality of strobe times. Using these voltage measurements, a slope is established on a time vs. voltage plot for the sample device. This slope is then applied to a line through the desired test point. Where a desired test point falls between strobe times on a tester, linear extrapolation from the desired test point is used to calculate what voltages must be tested for at the two bracketing strobe times in order to guarantee the desired performance at the desired test point. One or more devices are then tested for the calculated voltages at the corresponding bracketing strobe times. If a device passes both tests, it may be assumed (to a desired level of accuracy) that the device would pass the test at the desired test point, if it were possible to make such a test.
Two test points bracketing the desired test point may be used, as previously described. This method provides a relatively high degree of accuracy, even where a DUT has a response curve with a slightly different slope than the sample device. In an embodiment requiring less accuracy only one test point is used. In another embodiment, two test points not bracketing the desired test point are used.
REFERENCES:
patent: 5123286 (1992-06-01), Baumgartner
patent: 5428626 (1995-06-01), Fr
Mack Ronald J.
Statovici Mihai G.
Cartier Lois D.
Kobert Russell M.
Metjahic Safet
Tachner Adam H.
Xilinx , Inc.
LandOfFree
Method of increasing AC testing accuracy through linear... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of increasing AC testing accuracy through linear..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of increasing AC testing accuracy through linear... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2556257