CMC higher than calibration uncertainty
László Sleisz
Hi Michael or Juerg,
We calibrated a 10 dB attenuator and after that we switched on the CMC Entry in the Data Explorer.
Surprisingly it shows that uncertainty (especially the phase) of 10 dB attenuator calibration less than calculated CMC Entry in Database.
I used the same cal. kit, VNA, cable, connector during the CMC calculation and attenuator calibration.
Is it possible or I made a mistake during somewhere?
Of course, if I add CMC uncertainties in the Data Explorer, it became correct.
Regards,
László
Juerg Ruefenacht
Hi László,
what would be interesting is to see, is a graph where you normalize the two traces and then show the uncertainty regions (first trace without and second trace with additionally added CMC uncertainties).
This would give us an idea of how much the actual uncertainties are smaller compared to your calculated CMC entries.
Using the CMC calculator and sampling the complex plane does assume that your component has no length and the system is perfectly matched. Depending on the used calibration scheme and due to correlation effects it is possible that the uncertainties can get a little bit smaller (some contributions will cancel each other). This is one of the reasons why the 'add CMC uncertainties' functionality is available.
Like with an SOLT calibration and if the DUT length is long enough it can easily happen that the uncertainty contributions do highly correlate and therefore drop out almost complete.
Looking forward to see the above described graph of your measurement example.
Best Regards,
Juerg
László Sleisz
Hi Juerg,
You are right the differences between the CMC and actual uncertainties are very small.
I inserted two graph where I normalized to trace1 (without additionally added CMC uncertainties), trace2 (with additionally added CMC uncertainties).
First graph the full Sxx measurement, I zoomed the first 50MHz frequency of the S11 measurement in the second graph.
Best Regards,
László