Code behind the icons drives test languages
Benchmarks
Staff -- Test & Measurement World, 4/15/2001
Benchmarks
Meaningful benchmarks can be difficult to establish. Certain programs may excel at certain tasks, but those tasks might not relate to the real-world test applications you are running.
Consequently, we're undertaking these three approaches to test-program performance: program development, raw computational speed, and real-world computational speed.
• Test program development
We will develop programs to perform the simple diode test outlined in "Automated Instruments Smooth Rapid Test System Development" (Test & Measurement World, August 2000, www.tmworld.com/articles/2000/08_automated_instruments.htm). For this "benchmark," the required number of programming minutes won't be of particular value, as it would depend on the enthusiasm or fatigue level of the programmer, which can't be duplicated from one effort to the next. We can, however, provide a step-by-step, screen-by-screen description to indicate the programming ease of each development environment.
The effort will follow these steps:
1. Provide an indication to an operator to insert a diode-under-test into the test fixture, observing a particular polarity (cathode marking on the right, for example).
2. Deliver 20 mA of forward current, limiting applied voltage to 2 V.
3. If 20 mA can't be achieved with 2-V applied voltage, then instruct operator to tag DUT as "possibly labeled with incorrect polarity"; to reinsert DUT in opposite polarity; and to return to step 2. If DUT is already so-tagged, reject and record rejection in spreadsheet with associated DUT number.
4. Measure forward voltage drop with 20 mA applied; store answer in spreadsheet with associated DUT number.
5. Instruct operator to reinsert DUT with reversed polarity.
6. Apply a 2-V reverse voltage; measure and record reverse current; and store result in spreadsheet with associated DUT number.
7. Compare stored results for the DUT with predetermined pass/fail limits; instruct the operator to place the DUT in either a pass or fail bin; and continue.
• Raw computation speed. This test will run a sieve algorithm to locate prime numbers less than an input integer.
• Real-world computation speed. This test will perform digital-signal-processing computations on large arrays of data.
Stay tuned to this Web page as we gather and report results. We welcome your comments and invite any suggestions you might have on the types of comparisons you would like to see. Contact Rick Nelson at rnelson@tmworld.com.
