Test Report shows Data value within limits but Test Step Reports Failed

[Rick1458]

I am using TestStand 3.5 on Windows XP system.  A test report is attached (.doc file) showing a 'Multiple Numeric Limits' Test Step with strange behavior. For Measurement 0 and Measurement 2, a Data value > 0 is expected.  The Data = 0 but the report shows Passed.  For Measurement 3, a Data value <= 25 is expected.  The Data = 0 but the report shows Failed.  For Measurement 4, a Data value = 0 is expected.  The Data does = 0 but the report shows Failed. This test step is used in multiple places within the application and usually displays and evaluates data correctly.  In this one case, I got the data shown in the report and I can't come up with any explanation for how this could have happened.  The test step itself is straight forward.  The Data Source for each measurement is a separate Local numeric variable.  The test limits are hard-coded in the Multiple Numeric Limit Test Step.  There are no Pre, Post or Status expressions defined for the test step nor are there any Pre-conditions or post actions defined.  A 'stripped-down' sequence file showing just this step is attached. If anyone out there has any possible explanation for this, your help would be greatly appreciated.

[j_dodek]

Hi Rick,

I have opened your file on my TS4.0 maschine and executed it

it works like suspected.

I assume your NI-StepType has been modified or you are using it from user folder. Try to run it on a "clean" maschine.

and it will work.

Maybe a reinstall of Teststand helps

Juergen

[StevieZ]

I also tested out your code on TestStand 3.5 and found it to function properly. Please take Juergen's advice and try running this on a clean machine.

Regards,

Steven Zittrower

Applications Engineer

National Instruments

http://www.ni.com/support

[Rick1458 · Accepted solution]

After a little more investigation and some experimenting, I have been able to create a situation that produces the peculiar results shown in the test report.  The attached 4-line test sequence will produce the results seen in the report.  The 1^st and 3^rd lines in this sequence were lifted directly from the original failed test.  The 2^nd line was added by me to simulate specific data returned from the measurement instrument.  The 4^th line is a modified version of the evaluation step that would have shown the test data being evaluated more accurately.

. 

As it turns out, if the measurement value is an exact multiple of 2³² the test sequence will evaluate the data comparison correctly (Pass / Fail) but the test report will display 0.  This is because the test step that displays the data in the report formats the data as an unsigned integer.  In TestStand™, an unsigned integer can only hold a value of (2³² – 1) (0xFFFF FFFF).  The upper digits would not show up in the test report so the displayed value would be 0.  Note that any rollover digits above 2³² would get displayed correctly so the data value would have to be an exact multiple of 2³² to have test report display it as zero.  (i.e.  If the data value is 2³² + 1, the test report would still not display the upper digits but would display the 1 value in the report.). 

By modifying the evaluation step to format the displayed values using the TestStand™ ‘default’ display format, the test report would have shown the full data value including any digits above 2³².  This is demonstrated in the 4^th line of the sequence file.

.

Thank you to everyone who responded to this post with helpful suggestions.