Crazy results using Cable or Mic attached to sound card

[RobertoG]

OK, I cleared calibration by clicking two buttons as on below picture:

Next I measured impedance for correct input connection and next for swapped input. Results are presented in attachment. When I changed Rsense to 100ohm I obtained similar results. When I selected SPL measurements resulst seemed to be reasonable.

 

[John Mulcahy]

For completeness, also make SPL measurements for the other input channel (the one used as the reference for impedance) for the 3 cases open/10/short.
Edit: and attach the mdat files for the SPL measurements.

 

[RobertoG]

All measururements were done with using small amplifier to ensure low driving impedance. Rsense = 100ohm. Details of each connection were descibed in .mdat files. Thank you for your time.

 

[John Mulcahy]

I don't understand your descriptions of e.g. "amplifier connected to the R input - measurement output is loaded by 10ohm resistor and connected to L input". Can you attach a sketch of how everything is connected?

 

[RobertoG]

Simplified schematic was presented below. For second group of measurements inputs were swapped.

 

[John Mulcahy]

The idea of the second group of SPL measurements was to leave the connections as they are and only change the input channel selection on the Measure dialog. Currently the set of measurements in SPL.mdat all use the R channel as input so the L channel behaviour can't be seen.

Do you see any different behaviour without the amplifier? Output impedance has a very minor effect on impedance results.

Are you still using the UCA202? It should be used at 48 kHz (set 48 kHz in REW and as the default format sample rate in the Windows sound properties for the input and output). The monitor switch should be off.

With the connections shown above the measurement channel in REW should be the L channel, the ref should be the R channel.

Difficult to understand what can be going wrong.

Edit: Also worth turning off the REW noise filter to eliminate that.

 

[RobertoG]

I appied settings as you recomended. I removed also amplifier. Results are similar. I'm still using UCA202, but some time ago I used Mini DSP 2x4HD as as input and results were silmilar. I did the new calibration without amplifier and new results I presented in attachment.

 

[John Mulcahy]

Note that input and output need to come from the same device for impedance measurements.

You have selected very large buffer settings on the soundcard preferences, I suggest you set them back to 32k.

The short circuit calibration result is really very strange. I'm struggling to think of a way to produce that, and I'm away from home this week so don't have access to an impedance rig to see what might give such a result. With UCA202 there should be a message about a 1.00 sample timing offset between the channels when the open circuit calibration is performed, I guess you did not get that message, there is no record of it in the cal result.

Perhaps if the input were configured as mono somehow it might explain these behaviours? That is partly why I was asking for SPL measurements with the channel selection changed (not changing any connections), to make sure that the reference channel looked correct and different to the measured channel for the short circuit and ref calibration measurements.

 

[rmodelski]

Hi, I am also approaching very similar issue as described by RobertoG. Has the setup based on laptop + miniDSP2x4 + REW.
Is there somebody here who was successful with the impedance measurement? What was your setup (detailed equipment used)? I will try then to recreate similar scenario.
I am not motivated now to go into Dayton solution to measure 2 speakers. Maybe in the future when I will have more speakers to measure.
All the best.

 

[John Mulcahy]

I have used very cheap 2-channel soundcards with reasonable results down to 10 Hz or so, and an RME ADI-2 Pro using the high power headphone output as the driver with excellent results down as low as REW can measure. The attached is from the ADI-2 Pro made just now, the various cal steps then a measurement of a 100R bulk foil 0.005%.

 

[John Mulcahy]

I did the new calibration without amplifier and new results I presented in attachment.

I have been trying to find a way to reproduce the flat result at 100R you got for the short circuit calibration. If I leave the test lead terminals open instead of shorting them together I get a very similar result, flat at 100R. I haven't found any other way to produce that behaviour, though connecting the inputs together might give a similar result. The short circuit calibration is the same as measuring a zero ohm load, so the only impedance is from the test leads themselves. The measurement input gets connected to its own ground terminal through the test leads. The sense resistor remains in circuit as it does for any measurement. How were you making the short circuit calibration measurement?

 

[RobertoG]

Your results are fine, and I would like to achieve the same . You asked me, that I should try to change input, from LEFT to RIGHT. Should I do it at the place marked at the picture (red arrow). If yes, it didn't bring any difference in my measurements. They looks similar. I can add that in my opinion proper output connection (red and black cable to speaker) shoud be connected to LEFT input. Then it is possible to observe differences in level on input level analuser (marked at the picture). If i connect it as instrcuction suggest, I see constant signal oninpul level analser for opened as well shorted measurement output. Hmm, it is strange... My cable connections are presented in attachments.

 

[John Mulcahy]

Yes, use the channel selector to swap. The channel you select there should be the one the load is connected to (your red wire). The other channel, which is connected directly to the output, provides the reference signal. That one should have constant levels when measuring regardless of the load. To watch the levels on both channels while measuring open the Levels meter before pressing Measure.

Note that it is best to select the actual device, input and output being used in REW's soundcard preferences and not leave anything set to default.

 

[RobertoG]

Hmm. What can I add... I see that input level indicators are always equal. Sholud they show LEFT and RIGHT input channels level? If swap the inputs in the software measure window, mentioned before, there is no differences in result and also on the input meters. It looks like that software can't take control inputs or inputs behaves as MONO. I have no any other ideas...

 

[John Mulcahy]

Yes, they should show left and right, one showing the measurement channel the other the ref channel. Worth checking the Windows Audio properties for the input to make sure the default format is 2 channel and that Enhancements are disabled. Also may as well check that the sense resistor is not faulty (short) and there are no shorts between the channels in the wiring,

 

[RobertoG]

Thank you very much for your support and help. The problem was finally solved. Windows Audio setup had the input selected as MONO input. When I changed them to STEREO 16bit, 48kHz sampling, everything worked fine. Now I will be able to focus on my passionate measurements. Thank you for the great measurement software.

 

[John Mulcahy]

Good to hear! Turn the noise filter on again for cleaner measurements.

 

[Jasnochytron]

Thank you very much for your support and help. The problem was finally solved. Windows Audio setup had the input selected as MONO input. When I changed them to STEREO 16bit, 48kHz sampling, everything worked fine. Now I will be able to focus on my passionate measurements. Thank you for the great measurement software.

I love you man.
Thanks for following up on it till the end. I had exactly the same issue and exactly the same cause:-)
I still get straight 100 ohms on the first calibration step, doesn't matter what value I put and use.
Does your first calibration return the value of the Rsense resistor? Mine returns 100 ohm. The actual value of the resistor (plus leads) is 97.9 Ohms.
But the other two steps are good and the impedance measurement for a woofer looks good.

 

[John Mulcahy]

The first calibration step is not measuring the sense resistor, it is measuring gain differences between the channels. The result is actually a percentage, so 100 means 100% (i.e. channels have the same level).

 

[Jasnochytron]

Ah! I got confused because the value of the resistor is around 100 so I thought this was related. That makes much more sense
Many thanks again!