Socket calibration method and apparatus

Details Socket calibration method and apparatus Socket calibration method and apparatus

2000-02-28

2002-12-10

Le, N. (Department: 2858)

Electricity: measuring and testing

Testing and calibrating electric meters

Reexamination Certificate

active

06492797

ABSTRACT:

FIELD OF THE INVENTION
This invention relates to calibrating timing on a tester which tests integrated circuits.
BACKGROUND
This invention relates to calibrating automatic test systems for testing integrated circuits. Automated test equipment (ATE) is used to simulate the operating conditions that an integrated circuit will experience when used in an application. An integrated circuit undergoing testing is also known as a device under test (DUT).
The ATE is controlled by a computer which executes a set of instructions (the test program). The ATE must present the correct voltages, currents, timings and functional states to the DUT and monitors the response from the device for each test. The ATE then compares the result of each test to pre-defined limits and a pass/fail decision is made.
FIG. 1
depicts a typical environment in which integrated circuits are tested. A tester (ATE)
101
which contains the test program, memory, pattern generators and other circuitry, is electrically coupled to the test head
103
, which contains the circuitry to supply input signals to the DUT
109
and to receive output signals from the DUT
109
. A device handler
108
is often used to automate handling of DUTs to be tested, increasing throughput.
Timing calibration refers to correcting for time delays in routing a signal from the tester to a DUT (or conversely, from a DUT to the tester) due to constraints present in the testing equipment. Thus, it is essentially a software correction for a hardware-based error. These time delays are due to, among other things, the several layers of interconnect between the tester and the DUT, as shown in FIG.
1
. For example, if a test engineer desired to supply a signal to a DUT at a time value of 4 nanoseconds (ns) after the start of the test program, then he/she would need to take into account the electrical path lengths between the tester
101
and test head
103
, between the test head
103
and the loadboard
105
, between the loadboard
105
and the fixture
107
, and finally, between the fixture
107
and the DUT
109
as effecting the required travel time for the signal. Thus, the instruction from the tester to supply a signal to the DUT must be earlier in time than 4 ns after the start of the test program. Obviously, as the number of bits of data transferred per second on an electrical path or wire (transfer rate) increases, it becomes more difficult to determine if the DUT meets its performance specifications, due to tighter timing tolerances. In another test equipment configuration, the loadboard
105
, fixture
107
and DUT
109
mate with the device (DUT) handler
108
(as shown in dashed outline) to facilitate more rapid testing of a large number of devices
One method of timing calibration currently practiced in the ATE industry uses an intended DUT. The intended DUT is inserted into a fixture. A pulse is programmed on a driver circuit of the tester and a comparator is used to measure when the resulting pulse occurs at a selected reference terminal of the DUT to obtain the timing offset that occurs while comparing with the common reference terminal. Then a pulse is programmed on the reference terminal of the DUT and a comparator circuit on the tester is used to measure when the resulting pulse occurs at the tester to obtain the timing offset that occurs while driving with the reference terminal. The timing offsets are then adjusted to match previously measured timing data obtained from another tester on this DUT. However, this method of tester timing calibration has two significant disadvantages. First, the method requires having measured timing data from another tester (and this tester itself must be calibrated to some specific standard, not a device). Secondly, the method incorrectly assumes that device performance characteristics are constant over time, and are unaffected by minor changes in operating environment (e.g., temperature, power supply voltage, load impedance).
FIG. 2A
shows how robots are currently used in the ATE industry to make contact with the electrical connections
104
mating the test head
103
to the loadboard
105
(known as “pogo pins”) to measure the associated path length. However, as a result of practicing this method, tester timing accuracy can only be specified to the level of the loadboard/test head interface
102
—the electrical path length through the loadboard
105
to the fixture
107
holding the DUT
109
remains unaccounted for. Alternately, a robot
111
may be used which can make contacts at the fixture level (see alternate position
115
of robot arm
113
, represented by a dashed line), but these robots are often slow, because the robot arm has to pick up and move between each fixture (when multiple parts are tested simultaneously in multiple fixtures). Time is also lost when using the robot because the test head must be dismounted from the handler, the robot attached, measurements made, and then the handler must be remounted before being able to test parts. Moreover, the robot has many precision mechanical parts, so it is subject to frequent breakdowns without expensive maintenance. Faster robots are also very expensive.
FIG. 2B
shows the use of time domain reflectometry (TDR) as a method currently used in the ATE industry to measure the electrical path length
112
from the pogo pins (connecting the test head
103
to the loadboard
105
) up to the fixture
107
. This could be done with an oscilloscope
114
, but the tester itself could also be used. However, this method assumes that there is very accurate measurement of the electrical path length up to the pogo pins. Any error here is compounded by the measurement of the electrical path length from the pogo pins to the fixture. Moreover, the accuracy of TDR measurements is on the order of +/−50 picoseconds (ps) using the tester. The accuracy of TDR measurements is improved (on the order of +/−15 ps) using an oscilloscope. However, even this accuracy is insufficient as it represents too much of the total error budget in testing high performance integrated circuits. Therefore, there remains a need for more accurate timer testing calibration for high performance circuits.
SUMMARY
The present invention is directed to a method and apparatus for calibrating tester timing accuracy, and the method and apparatus are particularly well-suited for testing integrated circuits that have a small number of high performance (high-speed) terminals, such as the 64/72-Mbit Direct Rambus DRAMs which operate at 800 MHz transfer rates, but have only 30 high speed signal terminals. The tester itself conventionally tests integrated circuits to determine if they meet performance specifications. According to the present invention, for calibration, the tester measures its own properties using reference blocks that are fabricated in relevant respects to be as similar as possible to the integrated circuits that are to be tested. For example, the reference blocks have the same relevant external dimensions as the integrated circuits to be tested, the same electrical connections, and the signal terminals in the same locations.
To practice the present method, a set of reference blocks is fabricated. The number of reference blocks required is equal to the number of signal terminals being subject to timing calibration on the DUT. A signal trace electrically connects a different signal terminal to a common reference terminal on each reference block in the set. To perform the timing calibration, the reference blocks may be mounted on a single fixture one at a time, or using multi-site fixtures, multiple reference blocks may be used in parallel. The fixture provides electrical connection of the reference block to the loadboard, and ultimately, the tester.
The tester then programs a pulse on the signal terminal of the reference block, and measures the amount of time that elapses until the resulting pulse occurs on the reference terminal of the reference block. Polarity is then reversed, and the tester programs a pulse on the reference terminal of the reference block

Affiliated with

Also associated with

Rating

Socket calibration method and apparatus does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Socket calibration method and apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Socket calibration method and apparatus will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2993633