Umik-1 measurement problem

[kenniGT]

Hi guys, I am new here.

I have umik-1 microphone with calibration file. Last couple days I did lots of measurements of my speakers (focal sopra 2) but I always get funny results in higher frequency area. It does not matter where in the room. I even put blanket on speaker and microphone and measure it from close distance. Still the same.
Than I did another measurement of my headphones in other room (using new REW), and the result was similar, some strange higher frequency measurements. Is there something wrong with microphone? Or there are some settings in REW I can adjust?

Thanks
Tom

 

[AudiocRaver]

You have some high frequency noise going on somewhere in your system. It almost looks like the REW sweep triggers a noise source that takes a little time to die down afterward. Does that make sense?

I do not believe it is the UMIK-1.

edit: Or that the REW sweep makes it worse. It looks like the noise is always there at least a little. The blue area at the bottom-right of the first sweep indicates this.

 

[kenniGT]

You have some high frequency noise going on somewhere in your system. It almost looks like the REW sweep triggers a noise source that takes a little time to die down afterward. Does that make sense?

I do not believe it is the UMIK-1.

edit: Or that the REW sweep makes it worse. It looks like the noise is always there at least a little. The blue area at the bottom-right of the first sweep indicates this.

Thank you AudiocRaver, it makes sense.
Strange thing is, those two measurements were taken in two different rooms. I can not hear any strong background noise beside fridge and mechanical vent unit. I have turned them off, once for measurement, but this anomaly was still there. I will venture to find possible noise sources

And second measurement was taken having microphone inside one ear of a headphone (HD650). It is transparent though. I have 2.5m tall windows, not covered yet by curtains, in both rooms. Can they get exited enough by swipe to generate noise like this?

 

[Wayne A. Pflughaupt]

It’s definitely a steady-state signal – maybe from a power supply? A desktop computer, or one of your audio components? Notice with the black-and-white waterfall, it changed in level “mid stream!”

In any event, it shouldn’t be a problem if you can’t find the source. Waterfalls aren’t really relevant above 2-300 Hz or so, and as long as your SPL measurement is higher than the noise, you can get good frequency response measurements.

Regards,
Wayne

 

[kenniGT]

It’s definitely a steady-state signal – maybe from a power supply? A desktop computer, or one of your audio components? Notice with the black-and-white waterfall, it changed in level “mid stream!”

In any event, it shouldn’t be a problem if you can’t find the source. Waterfalls aren’t really relevant above 2-300 Hz or so, and as long as your SPL measurement is higher than the noise, you can get good frequency response measurements.

Wayne, that could be it. My Hegel H300 produce small hum, which is not audible from distance of 2.5m for me, but microphone is much more
sensitive. And with second measurements - yes, desktop computer is a source of some noise, plus mytek brooklyn is connected to it via usb, so maybe some high pitch noise can travel trough it even to headphones?

Good to know waterfalls they are no use above 300Hz. I was more worried there is something wrong with my speakers. I could not sleep after first measurements

Regards
Tom

 

[kenniGT]

There is another possible source. My room is before adaptation and all walls and ceiling are made from plasterboard. It is insulated inside so it should be better than concrete wall - takes some energy but... maybe lower bass during measurements makes some of those plasterboard vibrate, and they give away energy in higher frequency. Is that possible? I will make measurements without lower end this evening to check this theory.

 

[Wayne A. Pflughaupt]

maybe lower bass during measurements makes some of those plasterboard vibrate, and they give away energy in higher frequency. Is that possible?

Doesn’t seem feasible. If that were the case, the high freq signal would fade away with the low freq signal. But it doesn’t.

It's not all bad though. You got an awesome avatar out of it.

Regards,
Wayne A. Pflughaupt

 

[kenniGT]

Doesn’t seem feasible. If that were the case, the high freq signal would fade away with the low freq signal. But it doesn’t.

It's not all bad though. You got an awesome avatar out of it.

Yes avatar is funny I do not get so many arms anymore. It is just one arm. What I discovered is when I do measurement from 25hz to 20khz it is closer to main graph, from 500hz it is higher, from 2000hz it is gone...

 

[AudiocRaver]

Agreeing with Wayne P. that waterfalls are generally not used above 300 Hz or so, except that they can give an alternate perspective that can help you see a problem you could not catch as easily via a different measurement. (Like the HF noise above.) There is valid and there is useful.

Noise in the HF range like that will not be from physical resonances in walls, windows, etc. It would almost have to be electronic.

Have you tried to eliminate any ground loops? Is it possible there is any kind of volume-sensitive processing active in AVR or computer?

 

[kenniGT]

AudiocRaver I think I have checked all possible noise sources - 'unfortunately' non of it was creating this issue, but finally correct result...accidentally I have found where is the problem, but the story is a bit strange.
I have laptop dell xps 15 (1year old) , Umik1 is connected to laptop. Laptop send sweep signal to linn klimax streamer/dac via wifi (using linn songcast soft which can send any sound to my hifi from laptop) and than via xlr to amplifier. Simple setup. And here is the thing. There was something wrong with sweep signal delivered to speakers. I realized that when I measured laptop speakers (to see if I get the same anomaly) , and I realized that by the sound I heard. The sweep signal on laptop was continuous and waterfall was clean. When I was doing measurements via linn, sweep sound was going to the end, and repeating itself 3-4 times at very top. I thought all the time this is how sweep signal should sound. Now. After my laptop experiment I tried to go back to linn songcast stream. It did not work. REW could not see linn songcast as soundcard. I restarted REW couple times, switched on and off songcast and when it started to work I did measurement. And to my surprise, the sweep sound I heard from my hifi speakers was the same as I heard from laptop speakers. It was not looping back at the end anymore, hence normal results... I am new to REW so I did not know how correct sweep signal should sound, well, now I know No looping back at very top.

 

[AudiocRaver]

Very good! It was a bit of a first for me, glad you figured it out! Good detective work!

 

[kenniGT]

Very good! It was a bit of a first for me, glad you figured it out! Good detective work!

Thank you for helping me!
Yes, it was difficult one to pinpoint, and maybe it will help others in the future