Test Time Reduction Using Multiple Block Measurement Mode of the Femto Time Interval Analyzer

Introduction

Test time reduction is an important element in production testing. It can be difficult to reduce test time while trying to maintain an adequate level of test coverage. The high throughput, combined with the special arming modes of GuideTech’s Multi-channel Time Interval Analyzers (TIA), dramatically reduce the test time of timing measurements for a Device Under Test (DUT).

Scope

The purpose of this application note is to offer a technique, using the GuideTech Femto TIA that provides engineers with the ability to reduce the test times for AC measurements. The methodology for testing one device pin using the Multiple Block Measurement Mode of the Femto TIA will be covered.

Although testing multiple devices and multiple pins in parallel will also provide significant savings in test time, all of which can be performed using the GuideTech Femto TIA, will not be covered in this application note.

Definition of a Block of Measurements

A Block is the number of measurements (time tags) taken in a single execution of a start-measurement command.

For example: measuring 100 time-tags to calculate a frequency value would be considered 1 Block that contains 100 measurements. The result of this block of frequency measurements is the average of the 100 frequency measurements made in this block.

Multiple-Block Measurement Mode

The Multiple Block Measurement Mode is an example of the Femto's powerful arming modes and DSP processing power. The multiple-block mode is most effectively used where a test is to be repeated several times consecutively. The Femto can take each measurement block sequentially without stopping to process the data in each block or having to transfer the results of each block prior to the next. The DSP's process the multiple blocks simultaneously and the 5 statistical results: mean, standard deviation, minimum, maximum, and count size of each block, are generated and passed to the tester in a single GPIB transfer. Test time in most cases is reduced by as much as 80%.

Multiple-Block mode can be illustrated in the case where the propagation delay of a buffer has to be measured at eight different VDD voltage levels. The conventional technique would be to repeat the measurement for each VDD level and the total test time would be 8x the time it takes to test a single VDD voltage level.

As an example, let’s assume that 1 block of 100 measurements at a single VDD level takes 8ms to measure and read back the data via GPIB using the Femto TIA. Therefore, the test time for measuring the propagation delay for one buffer would take a total test time of 8ms multiplied by the 8 iterations for one VDD totaling 64ms. See table 1 for single-block versus multiple-block measurement mode test time comparisons.

Table 1: Test Time Comparison (Single-Block vs. Multiple-Block)

Multiple-Block mode can be used to capture the measurements of the eight VDD levels in eight consecutive blocks with only the VDD settling time between each block capture. The eight measurement blocks are processed in one DSP operation and the results for each block are transferred to the tester/test program in one GPIB transfer. The total test time for 8 blocks has an overhead of about _ of the test time needed for a single VDD measurement block.

Figure 1, illustrates a conventional test program setup to measure multiple VDD levels. After the measurement and levels have been setup then the loop count is set. At the end of one measurement loop, the program resets to the beginning of the program while the levels are changed on the device. Then the next iteration of the measurement is executed and repeated for each VDD level.

In contrast, in the multiple-block measurement mode, there are no loops associated with capturing the data after the device and levels have been set. Each block of measurements is triggered by the pattern of the tester. See figure 1, program flow comparison.

In both programming techniques two GPIB commands are executed, one to start the measurements and the other to send the data back to the tester. The difference between the two techniques is that the start-measurement and read-data GPIB commands are executed every time a loop is executed. In the multiple-block measurement mode the start-measurement command is executed one time via GPIB, the subsequent measurement blocks are triggered from the pattern of the tester and then the read-GPIB command is executed one time at the end of all the blocks to read the statistical data (mean, stddev, min, max, count).

Figure 1. Program Flow Comparison

Measurement Test Time Results

Figure 2 illustrates typical test times for measuring multiple-blocks on a single Femto channel. The graph also indicates the test times for measuring 8 VDD levels using a conventional measurement flow.

NOTE: the results are typical but not guaranteed.

Figure 2. Test Time Graph

Programming Example

The example code in this application note only shows portions of code that are relevant to the multiple-block measurement mode.

In this partial example, setup, measure, and reading data from three blocks of frequency from one channel of the Femto will be shown.

NOTE: the highlighted portions of the code indicate the items that enable multiple-block measurement mode. In addition, all Femto programming commands can be referenced in the User-manual or the GT_drvr.h file.

Setup

Measure

gt_start_measurement( GT_RD_CHAN_0A );
gt_check_errors();
:
:

Read

Once all the data has been read back via the GPIB, the results can then be sent to a tester datalog or a summary file.

Conclusion

GuideTech’s Femto TIA’s have a number of features to solve your critical timing test needs. Programming the multiple-block measurement mode is simple. Furthermore, using this powerful feature exclusive to the Femto TIA, you are able to reduce the test times for AC measurements by as much as 80%.

For further applications support or questions
please contact the
Applications Department of
GuideTech
(408) 733 - 6555