In the "Select Analyzer Calibration" dialog of the Uncertainty Calculator there is a "Adapter Removal" checkbox. Can I use this to include uncertainty from 2 port fixture de-embedding?
OK - I am not the uncertainty calculator expert, but I know his number, so here is what I learned.
Cal Expert:
It refers to adapter removal process--essentially two 2-port calibrations. It allows for the different ports to have different calibration kits. Each calibration kit will likely have different specifications.
Nothing to do with fixture de-embedding.
In other words, that check box simply tells the calculator to allow for a different cal kit on port 2 than on port 1, whereas in normal scenarios, the same cal kit is assumed on both ports.
more of our conversation
Me:
ok - so how would someone deal with fixture de-embedding uncertainties?
Cal Expert:
It would be a function of both calibrations, Use the same math as deembeding but take a few partial derivatives when dealing with the first cal.
I could work through it, but I won't be able to get to it for a while.
Me:
ok - I think that was the real question - can the uncertainty calculator take into account cascaded uncertainties when you de-embed a fixture from a port, assuming you know the uncertainties of the s-parameters for the fixture?
Cal Expert:
It doesn't do that.
So essentially if you have a situation where you have done a calibration and want to make a measurement on a DUT where you have to de-embed a fixture and calculate the final uncertainty of the DUT measurement, you have to cascade the uncertainties of the calibration with the uncertainties of the s-parameters of the fixture you are de-embedding. Think of the fixture s-parameters as a separate s-parameter calibration and measurement process for which you can run the numbers through the uncertainty calculator. once you have that, you can then add the fixture uncertainty to the uncertainty that you calculated originally with the calculator. The question is how to add the numbers.
Back to my converation
Me:
ok, so if they want to do that manually, how would they go about doing it? Assume that they have calculated the uncert. of the 1st tier calibration and they have uncert. associated with the SnP file being de-embedded?
Cal Expert:
To a first order, it would be the sum of the two effects.
Me:
so it is a sum and not RSS?
Cal Expert:
That gets to be the issue--strictly speaking since the errors terms are functions of frequency they can't be assumed to be uncorrelated so RSS is not totally legit. However, practically speaking RSS usually gives a closer (although optimistic) estimate.
Here is an example to help understand this idea:
I do a 2-port cal and when I run the numbers through the uncert. calc, I get a +/- 0.2 dB in S21 uncertainty. I then measure a 2-port fixture (with a different calibration and setup), and the fixture S21 ends up having +/- 0.25 dB Uncertainty. now if I measure my DUT with the fixture in place (and de-embedded), the uncertainty of the DUT S21 measurement using the completely legit and conservative estimate is +/- 0.45 dB. However in practical terms an RSS average of the two uncertainties would be more appropriate, so the real answer is closer to:
OK - I am not the uncertainty calculator expert, but I know his number, so here is what I learned.
In other words, that check box simply tells the calculator to allow for a different cal kit on port 2 than on port 1, whereas in normal scenarios, the same cal kit is assumed on both ports.
more of our conversation
So essentially if you have a situation where you have done a calibration and want to make a measurement on a DUT where you have to de-embed a fixture and calculate the final uncertainty of the DUT measurement, you have to cascade the uncertainties of the calibration with the uncertainties of the s-parameters of the fixture you are de-embedding. Think of the fixture s-parameters as a separate s-parameter calibration and measurement process for which you can run the numbers through the uncertainty calculator. once you have that, you can then add the fixture uncertainty to the uncertainty that you calculated originally with the calculator. The question is how to add the numbers.
Back to my converation
Here is an example to help understand this idea:
Hope this helps.