Redoing a failed test without having to redo the whole sequence test

[Mike Vivanti]

Hi there,

I admit I need to study more the TestStand data structures and also how a test results data base should be maintained, but I thought I'd just ask this question in case someone has a quick answer.

We have a complicated test sequence in Test Stand which takes several minutes to go through. Because we have quite a few failures in production, I would like the operator to after a failure be able to re-test a single line or sub-sequence without having to waste time going through the whole sequence again (which is necessary just for the sake of generating a new complete report showing all passes).

In another test executive we use, the operator is asked for the serial number and the test sequence executed and report saved. However the previous results are remembered until the serial number is changed, such that if the operator redoes any individual steps in the sequence, the remembered results record is updated. The operator is then able to re-save the updated results record as a new report. That is, if a sequce fails on a step it is possible to redo that step only and still generate a complete report showing all passes.

Is it possible to something like this in TestStand? (currently we just save xml reports though it is intended to create a results data base).

Thanks,

Mike V

[Bazza]

Hi Mike,

I am unable to give you the technical answer you require, but was interested in the approach you wish to take with regards to your production testing.  Your idea was put forward by an operator here last year, but we decided not to pursue its implementation.  We felt that retesting only failed steps (for a single serial number), and allowing this to represent a PASS for the whole product test (if the step passed) would compromise the integrity of the test.

For example, say you have a complex board that fails a voltage measurement for a given serial number (but passes everything else).  How do you know that in the process of repairing the voltage measurement other areas of the product have not been effected?  If you had to replace a component to fix the failure you would have had to use a soldering iron and there is every chance that solder balls / splashes could occur else where on the PCB.  If you choose not to retest everyting it is quite feasible that you could ship the product with further undetected failures.

Since what you were requesting was so similar to something we seriously considered, I did think it was worth passing on our thoughts to you.  Of course, I know nothing about the kind of products you are testing and what other processes you have in place, so please forgive my ignorance if this is the case.

Basically we concluded that the complicated implementation required to achieve this woud serve to get products through test quicker, but with a potentially higher undetected failure rate.

Good luck!

Regards,

Barry

[Captain_Jack]

In the Process Model, we added the callback ProcessModelPostStepFailure to allow us to perform some defined actions such as Force Pass, Force Fail, Terminate, or Prompt the operator for the specified action whenever a failure occurred. While we didn't provide an option for retesting a step, depending on how you designed your test sequences, you might be able to use the RunState.Next Step, RunState.PreviousStep, and Runstate.Step properties to get what you're looking for.

Hope this helps.

[JonathanN]

Hi Mike,

There is an awesome example that does what you are looking for in TestStand. This example is located at C:\Program Files\National Instruments\TestStand 3.x\Examples\Callbacks\PostStepFailureCallback. 

Also, here is a link to a KnowledgeBase that has some good information.

Hope this helps!

Best Regards,