[AndreaT]

Dear All, I have a problem of calibration of my soundcard embedded in my notebook DELL D830.

I can perfectly calibrate the amplitude but not the phase response.
I am using REV 5.01 beta7 and I used a known good loop cable.
Windows 7 32bit is my OS. Soundcard OS settings are for the best performance 24bit 192 kHz max.
(By the way, I have also the problem that REW 5.01b7 has only options for 44.1 and 48 kHz sampling... is it a bug or is so limited by design?)

The problem I have seams also dependent by the samling frequency.

In order to explain the problem I am providing 3 screenshots:
1) 44.1 kHz NOT calibrated amplitude and phase response


2) 48 kHz NOT calibrated amplitude and phase response


3) 44.1 kHz Calibrated response showing anyway a big phase error near 30Hz


My opinion is that there is a bug in the REW 5 or a problem in my notebook sound card or in its anti alias filter.

Anybody know what's the problem and if it is fixable?

Thanks in advance for your support
Kind regards,
Andrea.

 

[EarlK]

> If you are able to make the error go away by matching the sample rates of REW & the Soundcard ( set to either 44.1K or 48K for both ) then it's not a bug in REW . ( ie ; then it's likely a resampling error of some sort )

> REW ( or maybe it's JAVA ) doesn't support higher sampling rates .

:sn:

 

[AndreaT]

Dear EarlK,

thank you so much for your prompt and resolutive help!!! :sn::sn::sn:

You got the problem! It is needed to set all the audio related settings under Windows and within REW at exactly the same sampling rate.

I am not sure this is highlighted in the manual, but it should be.

Kind regards,
Andrea.

 

[jtalden]

Andrea,
All is good. This "phase error" is not really an error.
Your soundcard is just inverting the signal (as some do) and therefore the phase appears at 180° instead of at 0°.

Since by default the phase chart is set to wrap at +180° and -180° it can jump back and forth between those levels with only a minuscule change in actual phase as it did for you. [Just select "Unwrap Phase" in the "graph controls" box to see that the line is actually flat at 180°.

It is a easy fix to tell REW to invert the signal so the phase line will be at 0°, just check the "Invert" box found on the soundcard tab in preferences.

 

[AndreaT]

Dear Jtalden,

yes, I know, my displayed graphs also show an inverted intup response, but it is not the problem that I was reporting.

As I replied to EarlK, the problem was only in the sampling rate mismatch between the Windows setting (in my case I set 192 kHz) and the REW settings (44.1 and 48 kHz).
The error I reported is the one related to the sudden 180 deg phase change around 1.5 and 2 kHz.

So, thank you for have signaled to take care also of the inverting input, but for all the reader I repeat to be very careful about the sampling rate setting of Windows driver.

Regards
Andrea

 

[jtalden]

Andrea,
I did not explain well enough. There is no "sudden 180° phase change around 1.5 and 2 kHz" in any of your charts.

The +180° and -180° are the same value when “Wrap Phase” is turned on. A change from -179.99° to -180.01° will cause the chart to jump as shown in your chart. [It would be nice if the phase chart did not show the vertical connecting line the -180° to the +180° levels as the phase does not really follow the vertical line that is drawn. The vertical lines make the chart easy to misunderstand.]

To see the true phase of these charts, select the graph controls drop-down (the gear icon in the upper right corner) and select "unwrap phase". You will then see that the phase is flat at either -180° or +180° (depending on the position of the cursor when the button was pushed).

Note that, I am not suggesting that it is not a good idea to matching the sampling rates in REW with the settings in the soundcard. I am only correcting the misconception that the phase is jumping between two different levels as a result of a mismatch of sampling rates.

 

[AndreaT]

Dear Jtalden,

I already deleted all my past measurements, so I cannot do what you suggested.
Anyway, you are correct, there wasn't a 180° phase shift, but only a strange phase behavior.

However, again, I confirm that all the problems went away once correctly set the same sampling rate in Windows and in REW.

Doing so, the natural (own card) phase response follows the expected rotation at the beginning and end of the Nyquist band, as well as the compensation procedure worked perfectly.

If I will have time tomorrow, I will reproduce again my problem having care - this time - to invert the input to give a more easily readable problem graph.

Kind regards,
Andrea.