0

How do I write a procedure for a one-sided spec?

A lot of devices that we calibrate have a +/- specification, usually of equal value. But what's the  best way to write for a one-sided spec sothaton the Met/Cal test report, we have a nice, clean set of values. If there's no minimum or no maximum then those are blank or some other clear value and not an absurd one, such as +104 dB.

I'm familar withthe EVAL statement that does a comparison toa value and then gives either a PASS or a FAIL statement, butthat leaves the columns under Lower Limit, Upper Limit and Units blank.

Is there now a better way to write for one-sided specs? If so, please share your skill with us. 

 

Steve Reeves

 

4 comments

Date Votes
0
Avatar
James Smith

MetCal is not equipped to handle one sided tolerances very well. As far as I know it is always looking for an upper and lower value to compare nominal value to. If there is a method to remove these criteria, we would like to know as well. However, I am unaware of any so I use a work around. 

 

One preferred method is if I am looking for any value I know is positive I use nominal of the tolerance and the tolerance -nominal. It will generate values of 0 - X on the report. It’s not really one sided, but it works. I use this in scenarios like jitter is <2.3nS etc...

Another preferred method is In the event I am running a RF measurement I will use a similar method but use the difference value to reach a value where one expect to run into noise floor plus a little. For instance, spec is <60 and the instrument at best would only read -100, I would say nominal is -110 tol + 70 or nominal is -60 tol is -70 depending on how you want it to look at nominal and the test report to appear. You could do the same on the positive end of the scale as well. 

Other less than ideal way to deal with it is to go to the results table and just remove the values manually. It is generally regarded as less than ideal to intentionally edit the test records, even if it is to correct an inadequacy of software. 

Another less than ideal way that will require a little more effort and has potential to cause you a headache later would be to pick a very obscure value that you do not think any test tolerance would be. Then go to the report writer and put in a statement referencing if value = obscure value do not display. Save that as a one off report that is only to be run with that model/procedure etc... I do not know the formula off the top of my head, you would have to go look up crystal report writing. Then when you write your procedure do the math for the tolerance you do not want to display to equal the obscure value. In the event you EVER end up with a test with that obscure value, you will be kicking yourself unless you have written this particular clause for this one particular report aside from your other general reports. You will also have to keep in mind if tolerances change, other reports update etc . . . that this report may cause errors.

Those are the only 4 ways I know to work around metCals double sided requirement. If it does not exist, and enough people ask, MetCal support may write something in to work on a single sided tolerance. I personally just use a work around that give my reports mildly erroneous tolerances like noise floors that the instrument would never see in its life, but not so low it is obviously wrong. Please, if anyone else has better methods I’m sure most of us would like to know. I imagine most of us users have asked this same question.

 

0
Avatar
Steve Reeves -X (stevreev - SIMCO ELECTRONICS at Cisco)

Thanks James for your reply.

Can you provide a snippet of code to illustrate your first method, of the Jitter Test?

0
Avatar
James Smith

These two attachemnts show that same thing. In this scenario I am using 0 as nominal and the tolerance as a + value. When I use one of these two types of methods I always check the test report to verify it came out the way I want. Sometimes I use one method or the other, depending on the scenario.

 0.000V + 0.1U

or

0.100V -0.1U

 

Attachment not imported: gage.png
0
Avatar
James Smith

0.000V 0.1U

is the same as

0.000V +0.1U -0.1U

 

0.000V -0.1U 

is the same as 

0.000V -0.1U +0U

 

The + or - makes it apply the value in one direction. No value assigns it as two sided. 

Please sign in to leave a comment.