Strange EQ related speaker distortion

[slartibartfast]

This might be of interest @dominikz
Until now I have always used REW by playing a file from my server with a timing reference. I have always avoided playing directly from my PC as I thought my only option was Bluetooth. I finally remembered I have a Behringer UF202 which I can use to connect my PC to the Line In of my WiiM Amp and it seems to work OK.
I made some more speaker distortion measurements and noticed something weird when I disabled EQ.

This is the left speaker with RoomFit and EQ

This is the left speaker with EQ disabled to see if the slightly increased distortion between 200 and 300 Hz was due to the EQ boosts at those frequencies.
Above 2kHz you can see all the harmonics heading swiftly up before stopping. What on earth is happening there, it only happens with EQ disabled.



It can be seen more clearly in this image

 

[dominikz]

This might be of interest @dominikz
Until now I have always used REW by playing a file from my server with a timing reference. I have always avoided playing directly from my PC as I thought my only option was Bluetooth. I finally remembered I have a Behringer UF202 which I can use to connect my PC to the Line In of my WiiM Amp and it seems to work OK.
I made some more speaker distortion measurements and noticed something weird when I disabled EQ.

This is the left speaker with RoomFit and EQ
View attachment 28110
This is the left speaker with EQ disabled to see if the slightly increased distortion between 200 and 300 Hz was due to the EQ boosts at those frequencies.
Above 2kHz you can see all the harmonics heading swiftly up before stopping. What on earth is happening there, it only happens with EQ disabled.
View attachment 28113


It can be seen more clearly in this image
View attachment 28116

This is interesting, but may be difficult to point out the root cause of the difference between first and second plot without some deeper troubleshooting.
Is the behavior always consistent between measurement attempts?
Is everything aside from EQ on/off identical between the two attempts (i.e. all WiiM and REW settings)?

The easy part is to explain why HD rise stops after a certain frequency in the second plot - this is related to the sample rate used for the measurement. Note that with 44.1kHz sample rate you can only capture a little above 20kHz, and then consider that 2nd harmonic of 10kHz is 20kHz, 3rd harmonic of 6.7kHz is 20.1kHz, etc. Now check the frequencies where the THD rise stops in your second plot.

Try to measure with e.g. 96kHz or higher sample rate and you should see the rise continues past these frequencies.

 

[slartibartfast]

This is interesting, but may be difficult to point out the root cause of the difference between first and second plot without some deeper troubleshooting.
Is the behavior always consistent between measurement attempts?
Is everything aside from EQ on/off identical between the two attempts (i.e. all WiiM and REW settings)?

The easy part is to explain why HD rise stops after a certain frequency in the second plot - this is related to the sample rate used for the measurement. Note that with 44.1kHz sample rate you can only capture a little above 20kHz, and then consider that 2nd harmonic of 10kHz is 20kHz, 3rd harmonic of 6.7kHz is 20.1kHz, etc. Now check the frequencies where the THD rise stops in your second plot.

Try to measure with e.g. 96kHz or higher sample rate and you should see the rise continues past these frequencies.

Yup the only change between plots was that EQ was disabled. I even repeated the plot without EQ and saw the same result. I don't think I can change the sample rate of the UMIK-1.

 

[dominikz]

Yup the only change between plots was that EQ was disabled. I even repeated the plot without EQ and saw the same result.

Did you also try to repeat the measurement a few times with EQ enabled?
And did I understand correctly that you kept RoomFit enabled and just toggled the separately configured EQ?
What filters were configured in the EQ profile you toggled?

I don't think I can change the sample rate of the UMIK-1.

Yes, I believe UMIK-1 only supports 48kHz sample rate.

 

[slartibartfast]

Did you also try to repeat the measurement a few times with EQ enabled?
And did I understand correctly that you kept RoomFit enabled and just toggled the separately configured EQ?
What filters were configured in the EQ profile you toggled?


Yes, I believe UMIK-1 only supports 48kHz sample rate.

I have made lots of measurements with EQ and RoomFit enabled and many with both disabled and never seen this before. Generally the distortion at the end of each harmonic plot does the opposite, it decreased.
Yes I only disabled EQ

 

[dominikz]

I have made lots of measurements with EQ and RoomFit enabled and many with both disabled and never seen this before.

Indeed, this is why I wonder if it could be some kind of 'temporary' artifact that perhaps may not persist with every attempt.

 

[slartibartfast]

Indeed, this is why I wonder if it could be some kind of 'temporary' artifact that perhaps may not persist with every attempt.

I did repeat it because I didn't believe it but got the same twice. Can't do any more testing at the moment as TV is on.

 

[canard]

Resolution input line?
"Fixed" resolution of Toslink or coaxial output?
You make mesureament with output and input of this small sound card? Thd thd/n vs frequency? Rew or easy multitones?

 

[slartibartfast]

Resolution input line?
"Fixed" resolution of Toslink or coaxial output?
You make mesureament with output and input of this small sound card? Thd thd/n vs frequency? Rew or easy multitones?

Amp Line In resolution 24/96
It's a WiiM Amp so no toslink or coaxial.
Input UMIK-1 and output through the sound card.
I used the REW sweep for measurement. I may try stepped sine later.

 

[canard]

I thought it was a plus or an ultra, hence the proposal for measures on line out...
Strange story....
(sensitive to gain variations on the pre-gain?)

 

[EddNog]

Perhaps something odd is happening between the room correction algorithm and the sweep results. Try using a third party application to perform reviews before and after applying correction. I use HouseCurve; it's free, but you need an iPhone or iPad. It supports external mics and calibration files. It's a very, very useful tool for me to confirm my results before and after calibrating/applying DIRAC Live.

-Ed

 

[slartibartfast]

Perhaps something odd is happening between the room correction algorithm and the sweep results. Try using a third party application to perform reviews before and after applying correction. I use HouseCurve; it's free, but you need an iPhone or iPad. It supports external mics and calibration files. It's a very, very useful tool for me to confirm my results before and after calibrating/applying DIRAC Live.

-Ed

Does it display distortion? I don't have an iPhone anyway. The strange thing is the affected frequencies are way above the filter frequencies in RoomFit. When I get a chance I'll run the same sweeps from file instead of directly from the PC.

 

[EddNog]

Does it display distortion? I don't have an iPhone anyway. The strange thing is the affected frequencies are way above the filter frequencies in RoomFit. When I get a chance I'll run the same sweeps from file instead of directly from the PC.

Ah, good point. That I am not sure. This is definitely odd.