Impedance Measurement process - comments

[trobbins]

Hi John, I'm just setting up to do some impedance measurements and have some process comments on V5.20 Beta 55 on Windows 10.

When going through the impedance calibration in the Make a Measurement dialog box, the help notes (http://www.roomeqwizard.com/betahelp/help_en-GB/html/impedancemeasurement.html#top) don't clarify how to input a value for the Reference Resistor calibration and that is only identified in the third step of the calibration process. The dialog box also shows a panel for R input which is a bit confusing as that is not described in the help and it is a bit strange as it defaults to 20,000 and although the value can be modified it just defaults back to 20,000 whenever any action occurs.

When I 'capture' the plot screen it doesn't allow the equivalent circuit values for the DUT to be saved in the image (the values seen when the cursor is over the plot) - that capability would be very helpful as I certainly use the saved image as the measurement record.

My last comment is that I'm thinking there is an issue with the REW calculation for the equivalent series impedance (R+L) representation, and was hoping you could indicate what I am missing in this regard - some details:

I'm pretty sure I have an ok test setup as measured results are accurate for various precision resistances (from 10 to 10k) and air-cored inductances (I have some second grade references of 10u and 100uH) and capacitors, but there is a noticeable error with an audio transformer winding that has a 230 ohm DC resistance (measured by a few meters and the nominal spec from the manufacturer), but the impedance plot shows the winding as having a low-frequency impedance of 315 ohm (in the flat impedance section below 1kHz).

The DUT is a valve amplifier output transformer primary winding - with large inductance (10's of Henry at the low signal levels used for measurement). The test setup doesn't allow the DC resistance to show up for open windings as the impedance of the inductance is greater than the DCR down at 2Hz. But my interest is the leakage inductance of the winding (all other windings on the transformer shorted), where the winding DCR should measure as expected at low frequencies. I have repeated the measurement a few times and also added a 100 ohm resistor in series and it shows the impedance increasing from 315 to 415 ohm. The equivalent series impedance shown by REW is 315 ohm + 11.9mH at low frequency. The impedance and phase plots of the DUT are pretty noiseless up to 90kHz and show the high frequency resonances nicely that I am interested in (and have confirmed by other measurement means as well). The winding is not being saturated or operated in a non-linear region, and I get the same REW calculated low frequency resistance when the drive signal (EMU0404 USB headphone output) is lowered.

 

[John Mulcahy]

When going through the impedance calibration in the Make a Measurement dialog box, the help notes (http://www.roomeqwizard.com/betahelp/help_en-GB/html/impedancemeasurement.html#top) don't clarify how to input a value for the Reference Resistor calibration and that is only identified in the third step of the calibration process.

The help says to press the Reference cal button which then asks for the value, seems fairly straightforward.

The dialog box also shows a panel for R input which is a bit confusing as that is not described in the help and it is a bit strange as it defaults to 20,000 and although the value can be modified it just defaults back to 20,000 whenever any action occurs.

It is described near the top of the help in this passage:

The load appears in parallel with the input impedance of the soundcard and the impedance of the test leads. Before calibration those effects are partially corrected for by using the parallel equivalent of the sense resistor and soundcard input resistance in the calculation, the soundcard input resistance figure is entered on the dialog.

The valid range is 1k to 1M. It isn't used after calibration though.

When I 'capture' the plot screen it doesn't allow the equivalent circuit values for the DUT to be saved in the image (the values seen when the cursor is over the plot) - that capability would be very helpful as I certainly use the saved image as the measurement record.

Probably better using the Component model to get an equivalent circuit, otherwise the way that overlay is generated means you would need to use Alt+PrtSc to capture the window.

My last comment is that I'm thinking there is an issue with the REW calculation for the equivalent series impedance (R+L) representation, and was hoping you could indicate what I am missing in this regard

Seems difficult to argue with the measurement result, 315 ohms and phase pretty much zero. All the overlay is showing is the equivalent values for the impedance magnitude and phase.

 

[trobbins]

Thanks for responding John.

The particular inductor I measured doesn't cause the 'Component Model' tab to be enabled, as the phase is not >45deg at the end of the plot at 96kHz (which seems to be the test condition. If I limit the test frequency span to 20kHz max then I do get a Component Model assessment

The concern I have is that the measured impedance at low frequency is a resistive 315 ohm, whereas the measured DCR of the winding is 230 ohm - so there is an inconsistency creeping in that I can't reconcile, given that all other discrete component measurements have given accurate impedance levels

 

[goginux]

@trobbins follow this link:

I was successful with a USB card, stereo input and output. The measured parameters were close to the manufacturer, morel, seas, vifa, peerless

 

[trobbins]

Thanx goginux. I have been through the impedance measurement calibration process many times now, and tried a wide variety of variations of passive loads to confirm that the test setup is accurately measuring resistor, inductor and capacitor values over a good array of ranges. With respect to inductive DUTs I have measured low value chokes where the DCR is showing accurately down at low Hz before the inductive component starts to become significant, as well as measuring electrodynamic speakers. I have used independent measurement techniques for the DCR measurement to know that for the transformer DUT result above it is 230 ohm and not the low frequency 315 ohm reported by REW. I have the latest beta rev and calibrated with that rev, but no change.

I have just measured another somewhat similar 'output transformer' for its leakage inductance and it shows a shelf change in low frequency impedance, dropping from circa 260 ohm (> 200Hz) to 160 ohm (< 10Hz), with a noticeable phase hump (+13 deg) centred around the middle of the shelf (circa 40Hz), and with a known DCR of 120ohm. I will try some other similar transformers (these are valve amplifier output transformers with substantial audio power ratings - eg. > 10W) to see what low frequency characteristics they show.

The measurement process is certainly at the lowest end of frequency capabilities, as my EMU0404 starts to roll off below about 8-10 Hz and so impedance calibration compensation from 2Hz to 10Hz would be significant down at 2Hz as would be the droop in soundcard excitation voltage applied to the winding (-2dB at 2Hz). It seems that there may be a very low frequency impedance change well below 2Hz for the first transformer I measured which was not showing itself by any droop in impedance or change in phase above 2Hz. I will now try and see if I can observe impedance magnitude and phase changes below 2Hz using the X-Y scope technique, and hopefully they clear up the situation. Such is the joy of having REW as a measurement tool, as it certainly extends commonly available bench measurement capabilities.

 

[John Mulcahy]

If you use the latest V5.20 release candidate you can change the lower limit of the frequency axis in the View preferences to 0.1 Hz and see if a suitably long sweep (ensuring the Analysis preference for IR truncation is set to Keep Full IR) shows any different behaviour at lower frequencies. You will probably need to calibrate with longer sweeps as well.

 

[goginux]

i have 5 audio cards,the best measuring,with 99% tolerance i have with cheap usb 'axagon ada17',with in stereo,out stereo.

 

[trobbins]

Thanks John for the Preference menu adjustment guidance. I have been using a 4M FFT, and have reduced the frequency span to 100Hz for the plot below.

I did a re-calibration as you advised with testing down to 0.1Hz. Prior to these changes, as shown in my first post's plot, the result plot traces start just above 2Hz. Now the traces start right at 2Hz, and that extra data does suggest the impedance magnitude is starting to slightly droop as frequency reduces through 2Hz, and the phase is continuing to slightly increase. As indicated I will try and set up an X-Y scope plot to 'view' the response below 2Hz. Given the shelf response of two other output transformers, where the shelf is at a higher frequency, I am better appreciating that the very large inductance of these transformer windings still influences the winding impedance at surprisingly low frequencies when measuring a leakage inductance test configuration (all other windings shorted except for the driven winding).

Thanks goginux, yes I have obtained excellent passive part impedance plots including for precision resistors and inductors, as well as appropriate component models (when I ensure the sweep stops at a frequency where the phase is sufficient to trigger the modelling).

 

[John Mulcahy]

The 2 Hz limit wasn't intentional, I'll send you a link to a test build that should measure to 0.1 Hz.

 

[trobbins]

Much appreciated John (once again ). The impedance does now measure down to 0.1Hz. I did a recal and confirmed the preference settings. The impedance does show it is still falling below 2Hz. My soundcard bandwidth is starting to get ragged below 1Hz, but the characteristic is showing through.