Alright, I was wondering how this is currently done in the industry. Usually, when we produce some hardware, we have a few testing options available (Automatic Optical Inspection, Flying Probe, In-Circuit-Testing, Function Testing, ...). Usually the customer will ask us what the advantage of one over the other is and we have a document explaining those which is fine. The next question usually is, if his board can be effectively tested using the one or the other method. What he usually would want to hear is: Using this test method we are able to find 75% of all errors, using the other one we can reach 90%. If you did this up front, we could reach 95%.

Now for the (engineering) question: How do you create such numbers? Have you done this in the past already? Do you require such numbers from your manufacturer. I know there is no universally available answer to this question but I would like to get some insight how this is handled in other companies. We simply want to provide a reliable and dependable source of information to our customers. Any insight is highly appreciated.

  • \$\begingroup\$ Good question. I would like to hear a qualified answer as well. (/me believes that these numbers are mostly made-up..) \$\endgroup\$ – Eugene Sh. Aug 10 '15 at 14:59
  • \$\begingroup\$ Even if I have to say that this is in many cases certainly true, I still have hope for some insight. I could also guess that many of these numbers come from previous experience but ... well ... \$\endgroup\$ – Tom L. Aug 10 '15 at 17:19

When I worked in this field, years ago, building functional test systems, one of the things I'd do was sit down with a printout of the schematic, and some colored pens. For each test, I'd color in the part of the schematic that it verified. Power up test - color in the power supply section. Device programming - color in the flash memory and related programming connections. Boot up - color in the MCU, Flash, and memory bus. etc.... I'm sure the tools available have improved in the intervening years.

Keep in mind that different test methods have tradeoffs, usually related to cost or speed. Flying probe is slow, but has low tooling cost. ICT has high tooling cost, but is fast. ICT or flying probe may not be able to physically reach every circuit node. Functional test may allow you to verify circuits that you can't probe with ICT or Flying Probe.

  • \$\begingroup\$ This is what you will also get from boundary scan testing (from experience in my past). These tools will usually spit out a list of "tested" nets and/or components - basically what you would get with coloring the schematic parts. \$\endgroup\$ – Tom L. Aug 11 '15 at 4:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.