Phase in impedance measurement

[jschwender]

I wonder why the phase is limited to +-90° in impedance measurement? That means the result is not a complete system description, right?

 

[John Mulcahy]

If the phase of a passive component's impedance was beyond 90 degrees it would mean it had negative resistance.
That particular measurement also suffers from the input swap problem, since it is an inductor but the phase of its impedance is negative instead of positive.

 

[jschwender]

Thanks for the reply. Now that you remind me, i withdraw my embarrassing question about fundamentals....
A more interesting point is, that if you look at the captured data, the phase between Captured and Ref Captured runs from 0° towards 90° with increasing frequency. The diagram shows a constant phase, which to me seems not related to the Captured data.

 

[John Mulcahy]

At very low frequencies the phase is close to zero as the impedance is mainly resistive. Phase is not constant for valid measurements.

 

[jschwender]

Right, your example looks reasonable. What i get is totally flat, and that is not what i see when i look to the captured data:

 

[John Mulcahy]

That's probably a calibration problem, the phase also has the wrong sign. Perhaps a channel swap during calibration?

 

[jschwender]

That channel swap problem is extremely annoying, and i cannot figure out where it comes from. As i utilize a Laptop with Linux, i am more or less limited to USB sound cards. Mine here is Asus Xonar U7, which has excellent low THD and noise. I also tried that card unter Windows 10, but Windows sound system is a serious nightmare: it is noisy, seems limited to 16 bit, and i could not get any sampling rate beyond 48kHz, no matter what i chose. So i will probably pick another one to test, unless you have another hint?

 

[John Mulcahy]

You should be able to get the same performance on Windows using ASIO drivers as you get on Linux, using FlexASIO if the interface has no native ASIO drivers.

 

[jschwender]

I tested a Asus Xonar U7 MKII instead, and it shows the same channel-swap failure. I tested a Steinberg UR22mkII and it does not show such a failure. Both tests on same Linux PC. I also tested recodings with plain alsa tools: the U7 worked perfect with this, it never shows such a channel swap. At the end it looks like root cause is a incompatible combination of the U7 sound cards with the Java Audio Interfacing. Solution for me and recommendation for all users is: don't use Asus Xonar U7.

 

[jschwender]

Just bought a Steinberg UR22mkII interface: headaches are gone - that one works as expected, no channel swap issue. Utilizing an amplifier the measuring results are surprisingly precise: I have a reference Inductor with 352 µH and the impedance measurement calculates the component to 351 µH. A fairly small reference capacitor of 23 nF gets also precisely calculated as 22.6 nF, although it's ESR value is calculated to 30 ohms, which is far too high. For smaller capacities precision gets less, and also for very large inductors (1 H) the values are less precise, but those are less relevant for audio components. I must praise John for the very good implementation of this component analyse, very good job!. Thank you!

 

[John Mulcahy]

For small caps (below 100 nF or so) the series R element of the model will often be large as it has very little impact on the fit over the range being considered, since the impedance at the 20 kHz end of the range is still in the hundreds of ohms. Something for me to think about though. Glad you are getting stable results now.