General Metrology Forum
Electronic Leads effect on the Uncertainty of a Measurement
Measurement Error due to the connectors and leads in electrical measurement comes in two varieties:
- Bias Error: or offset error, such as lead resistance in an ohm measurement which affects the measurement with a "bias" equal to the resistance value of the leads at the moment of measurement.
- Random Error: a randomly moving variable during the measurement. This can be caused by electrical noise, poor grounding, connector problems, inappropriate leads for the task, etc.
When analyzing Measurement Uncertainty, we deal with both bias and random type influences in a measurement result. When bias can be estimated, and it is significant in its effect on the quality of the measurement result, it should be corrected. But even then, we are left with the uncertainty of the correction.
Their are many sources of information regarding testing wire "bias", including thermal effects experienced when measuring small DC Voltages. I referred to Keithley in my other post with a link.
For example, if two types of leads are allowed for a given process, set up the process, repeat the measurement, as an experiment, several times to determine your process repeatability (standard uncertainty). Swap out the lead types every other measurement, making a note of which leads were used for which measurement. Compare the two data sets in a two sample t-test to determine if a discernible difference exists. If you are seeing differences, then investigate them. Keep it simple, direct, and most of all, make sure you are dealing with a significant contributor to your measurement uncertainty before you invest a lot of time and resources to study it.
The list of potential sources of uncertainty is near infinite, find the significant ones, estimate them appropriately. Meaning, the more significant the source of uncertainty, the greater effort it deserves in adequately estimating its effect.
If you are estimating a minor contributor, use a method that addresses the worst case scenario, find a reasonable, albeit somewhat conservative, containment limit value and move on. Values believed to be conservative containment limits, require less evidence to support them and if they are NOT significant with respect to the other values in your uncertainty budget, it won't make a difference in your Total Uncertainty.
This allows you more time to properly and thoroughly evaluate the things that matter, the significant contributors. I speak from experience, as I have found myself on more than one occasion, spending more time than I should have on a contributor that just did not matter to the final value.
Have a Nice Day!