Polarity anomaly - UMIK-1 vs. UMIK-2

[AustinJerry]

When measuring the same speaker using first a UMIK-1 and then a UMIK-2, the measurements are showing opposite polarity. Please help me understand why this would be so. Am I missing a setting?

 

[sm52]

Make sure both have Excl selected in the Preferences tab.

 

[AustinJerry]

I simply unplugged one mic and connected the other. Both were using Excl with identical settings.

 

[thothsong]

You posted the same thing two months ago. You said you were going to open a support case with miniDSP; did you get a response from that?

 

[AustinJerry]

I never heard back fro MiniDSP. The reason I re-posted this issue is that this time my test was extremely simple. Swapping the two mics keeping everything else the same should IMO result in the same polarity mesured by the impulse response. I am hoping that someone else who has both mics will run a similar test.

I also have not measured the polarity of all of my speakers (a 7.1 setup) to see if polarity matches all around. If all are the same, then the polarity doesn't really matter.

Do you have a theory why I am seeing different polarity?

 

[thothsong]

As I posted previously, I have both mics, and they produce the same polarity.

 

[AustinJerry]

Interesting, thanks for reminding me again. I wonder why I am getting different results. I tried different USB cables, but that didn't make any difference.

 

[thothsong]

Have you looked at old mdat files to see if the polarity has always been different, or if it changed at some point?

 

[AustinJerry]

Here are two impulses from prior measurements that don't show the polarity anomaly. The REW version used to take the measurements are likely different, if that might cause a difference.

 

[thothsong]

The Info button will show you the REW version under Dated: and also show the driver used under Source:. Might be worth searching your mdat files further to see if you can narrow down when the polarity flipped.

 

[AustinJerry]

Well, I have an MDAT with measurements using both mics, and it shows the polarity issue. I can't see anything in the info file that would suggest why this is happening. I was hoping that JohnM would shed some light on the issue.

 

[sm52]

One measurement was made at 48k, the other at 96k. But this is not the main question. Your REW version is 5.20.13. The latest version of REW has many bugs fixed for now. Perhaps your case has been corrected. So it's better to upgrade to the new version of REW. And check the measurements on the new version.

 

[thothsong]

Jul 5 2023 is the date that you previously first posted about the problem. So you don't have any evidence of it occurring earlier?

 

[AustinJerry]

I have 63 folders containing unique REW measurements after fresh Dirac calibrations. Some use the UMIK-1 and some use the UMIK-2. It is tedious to search each of these folders to see if there is a point where the polarity anomaly first appeared. And I don't know what this would prove anyway.

 

[AustinJerry]

Well, I took another pass at trying to resolve the polarity anomaly between the UMIK-1 and UMIK-2 in my setup. I made sure that I was using the latest REW version for the measurements.

Still the same result. I measured all seven speakers in my 7.1 configuration, and all are showing the same polarity.

So, as long as all speakers have the same polarity, I won't be concerned.

And by the way, for those who are not familiar with the effectiveness of Dirac in improving the impulse response, consider this comparison. It is quite a difference! And the time alignment is equally impressive.

 

[phofman]

IMO your device sends inverted data. Can you just check the "invert input" box in REW prefs?

 

[AustinJerry]

IMO your device sends inverted data. Can you just check the "invert input" box in REW prefs?

Thank you for the suggestion. Here are the areas where a setting might result in an inverted result:

As you can see, neither setting has been selected. Is there another setting that you think I might have missed? Regardless, wouldn't such a setting apply to both mics, not just one?

 

[phofman]

Have you checked in some other capture application (e.g. audacity) that your second does not really fetch inverted signal?

 

[AustinJerry]

No. Are you suggesting REW may have an issue? Again, if it were an REW issue, it would likely affect each mic the same. I really have no other way of measuring anyway.

My theory is that there is something different with the two mics. I just haven't isolated what the difference might be. Next step will be to measure with different cables to rule out whether a cable might be wired incorrectly.

 

[phofman]

I do not think there is a problem in REW when using one mike vs. the other. That's why IMO the next step should determine if one of the mikes does not fetch a reverted signal. Just play the measurement signal in REW and capture mike A and mike B e.g. in audacity. The polarity should be directly visible from the initial pulse right away, IMO.

IIUC both mikes are USB so no cables to flip polarity.

 

[sam_adams]

This is becoming more puzzling every time this issue is raised. In this blog post from 2016 the author states that the Umik-1 has inverted output when compared to a DPA4007 using the SMAART software. There was an issue raised in the MiniDSP forum in 2015 about the Umik-1 polarity that got only a single response from MiniDSP about opening a ticket.

 

[AustinJerry]

I opened a ticket with MiniDSP in early July but never received a response. I opened a second ticket this week, reminding MiniDSP of the many years that I supported their products by developing guides that they have published on their web site. I reminded them that I expect an answer this time. Let’s see what happens.

 

[AustinJerry]

MiniDSP has responded and they are looking into this issue. I ran a quick analysis to try and identify when the issue first appeared. The analysis results are in the attached file. The issue first started appearing for me in July of this year.

I am also attaching a REW MDAT file that compares measurements using both mics that clearly shows the issue.

 

[thothsong]

It's certainly curious that it switched polarity in a stable build (5.20.13).

 

[thothsong]

Was the 2/24/2023 measurement done with the Java driver and EXCL device?