Impedance calibration issue

[oliver]

Hi John,
I am on OSX 10.13.6, with REW latest build.
I do use a RME Fireface UC as audio interface.
I wanted to perform a more accurate impedance calibration for my measurement system, taking into account also the test leads.
My test system has a test box where I can select different test procedures and other stuff for my lab, so the whole sytem has a 1.9Ω resistance as Rleads.
When I perform the "Short circuit cal" with an R leads of 1.9Ω the calibration ends with this warning:

Input and Ref channels are correct:

If I short before my box, at the ampli output after the Rsense, the calibration is good, but doesn't take into account the 1.9Ω resistance of the box.
The Rsense I use is 4.7Ω and the RME input impedance is 10K.
My headphone amp has an output impedance of > 0.2Ω, so the difference between the Ref and the input is not really high, and maybe REW is expectng a higher difference.

Thank you,

Oliver.

 

[John Mulcahy]

1.9 ohm in the interface box? Do you have a km of wire in there?

With such a low Rsense and high lead resistance the "shorted" level is only about 10 dB below the ref level. REW's 'sanity check' minimum difference is 12 dB, you would need an Rsense of 5.7 ohm or more.

 

[trobbins]

Oliver, is there a reason you need to use a 4R7 Rsense ? I use a 100R nominal Rsense. I find that the accuracy of making measurements of external low resistances (or impedance with low magnitude at low frequency like a coil, or conversely the ESR of a large cap) depends on using a calibration reference resistor that is similarly low (I use 10R 0.05% for that application). It is worthwhile using a few reference resistances and capacitors (and air-core inductances) to confirm your setup is performing as anticipated, especially during initial testing of a new jig.

If you are measuring up to 96kHz, then also worth checking if Rsense has a datasheet that shows impedance with frequency. It's unlikely that a small MF will show reactance, but some MF resistors (eg. Vishay PR02) can start to show a slightly capacitive impedance above 100kHz for low values like 4R7, and of course your test jig should be wired and screened for low influence of parasitics.

 

[oliver]

Hi John,
thanks for your answer.
Yes, due to some plugs and switches, I do have a quite high contact resistance..

Hi Trobbins,
I am following your suggestion, thanks, I have increased the Rsense to 47Ω and performed a new calibration, but I have found that with a calibration reference resistor of 4,7Ω I am out of several Ohm, instead with a reference resistor of 99Ω the measure is very close to the real value.
This is Balanced Armature receiver, with a 4-wire milli ohm meter it measure 12Ω, so the result with a 99Ω reference resistor is almost identical ( now with the Rleads included):

:

Thank you,
Oliver

 

[trobbins]

Are you able to independently confirm the Rsense value, the reference resistor value, the Rinput value presented by your soundcard, and the Rleads value, and are you using those measured values in the Impedance calibration setup?

Are you using a signal level setting that you know by loopback test is sufficiently far below your known clipping level ?

When you do a calibration using the 4R7 resistor, can you then do a measurement of the 4R7 resistor and do you get a flat line at 4R7 ?

 

[oliver]

Hi Trobbins,
I can confirm that all values are correct, but I think I have found where I went wrong ..

Due to how my system is set up, the reference resistor was after the Rsense but before the leads, so the calibration didn't take into account the Rlead value.

I performed a calibration with the ref R connected to the test leads and it is now correct:

Thank you for your help,
Oliver.