I am using the ± operator on some of my variables to include tolerances on a variable. This is an example:


When I combine this plus other variables with their own tolerances I end up with resulting variable which contains several entries. I can access each element as if it is a vector element, but I cannot use the variable itself as a vector which would allow me to use the built-in functions. To date I have been manually counting how many elements are in a variable and then using a for loop to convert the elements into a single column matrix, and then performing a min/max function on the resultant matrix. This can be error prone if I add or remove terms in the preceding formula, either creating an error as the for loop is referencing elements that don't exist or even worse, not including elements which should be included!

Is there a way to get the min/max values? If not is there a way to get the number of elements so I can auto-update by for loops.

I'm also interested in know what variable type this is.

From reading your request, your approach is not correct.
In some way, each component belong to the Engineering result.
Then the resulting ± error is the square root of the sum of the
components expressed in term of result.
As an example: for a T/C [Thermocouple]
we sum the T/C itself + wire extension + converter
then take the square root of that sum.

Stat Incertitude.sm (21 KiB) downloaded 35 time(s).

Thanks for your reply, it sent me into a deep dive into statistical tolerance research and though I've only just touched the surface I'm looking forward to reading more!

I think I understand what you are saying. Instead of using the tolerance directly in the calculations itself, I should be calculating the inaccuracy as it's own term. More importantly the analysis I was performing previously was the worst case instead of the statistical case.

In the example below I performed the RSS (Root Sum Squares) on the tolerance terms which I returned to the nominal calculation at the end. I wasn't sure how to perform it to the inductor/capacitor values themselves and retain proper dimensions so I opted to do it on the percent tolerance instead. The result was the min/max terms I am looking for. Does this look correct to you?

Tolerance examples.sm (10 KiB) downloaded 42 time(s).

Wrote

Thanks for your reply, it sent me into a deep dive into statistical tolerance research and though I've only just touched the surface I'm looking forward to reading more!

Hi. This isn't the statistical procedure, but it's the usual propagation of errors using the gradient of the functional dependency. (Actually, the absolute values of the gradient components). With only one measure, I'm not sure that you can apply other method.

Tolerance examples.pdf (108 KiB) downloaded 56 time(s).
Tolerance examples.sm (25 KiB) downloaded 46 time(s).

Best regards.
Alvaro.

The technical approach from Alvaro is the right one.
It may turn out horrible in some/most cases.
Measuring the mass-flow via orifice plate is the best example.
All those complexities are included in the ISO-5167.
It results to ± 0.06% of the reading, as long as the remaining
of the installation is all conformed ...
pipe length, roughness, fabrication, XTR [transmitter] ... etc.
Once your circuit will be wired, carry enough frequency measurements,
plot the histogram, attempt a PDF fit.
Denser the histogram, more reliable the attempted PDF will be.

Cheers ... Jean