Need help removing odd order harmonics

[Lafore]

Hello! I'm looking for help removing odd order harmonics from my stereo. I noticed a problem when certain things sounded harsh, so I started measuring. My audio chain is a digital source (TV, computer, doesn't matter) -> minidsp flex ht ->Buckeye Hypex NC502MP -> Revel F206. My microphone is a UMIK-1.

I recorded the output of my minidsp directly to my computer's line in, and the spectrum is clean, so the distortion must be introduced after that. I swapped my buckeye amp for an AIYIMA A20, changed line level and speaker cables, and changed speakers to Triangle BR02. The distortion remains.

The only other things that I can think of is that my microphone is introducing the harmonics, and the problem is just my ears or something. If anyone has more ideas I'd really appreciate you sharing.

Example harmonics are from a 333 Hz tone, I see an additional 1 kHz peak, and for a 350 Hz tone, I see an additional 1050 Hz peak.

 

[LTig]

I think you see the speaker distortion. How high is it?

 

[ThatSoundsGood]

Can we see the measurements?

 

[Redacted]

Changing speaker cables isn't going to help.
I assume you are injecting single digital tones from a file into your DAC.

You are measuring speaker output into the room using a mic. when you get the "problem", but when you measure the electrical signal at line level, there is no problem. What you don't yet know is what the signal going to the speakers looks like. You can use a simple voltage divider to feed the signal from the terminals on the speaker to line-in on your PC. I'm betting the signal is clean, just like you see post miniDSP....
If so, this leaves two possibilities - the speakers or or the microphone are introducing distortion. I think your method is suspect as you seem to be comparing an analogue signal coming into your PC's mic input (then going through the PC's ADC) to a digital signal coming in via USB from the Umik.
Could also be you are overloading the input on both amps.

 

[Lafore]

Thanks everyone for your responses.
I’m happy to get so much help so fast. I really didn’t expect it. I can post more, like measurements, later today. I’ll also try a different mic.

I don’t remember the distortion level exactly, but it looked about the same with both speaker+amp pair. I’ll post it later when I’m back home.

The distortion is there with the UMIK’s ADC. It’s not there if I record directly from my minidsp’s line out to my computer’s line in. I didn’t use my computer’s mic input for anything. I don’t have an analog mic. My other mic also has its own ADC.

 

[Lafore]

I assume you are injecting single digital tones from a file into your DAC.

I used both https://www.szynalski.com/tone-generator/ and REW to generate tones.

Could also be you are overloading the input on both amps.

Is it possible to do that even when volume is set low, like 50%? I can hear the distortion no matter the volume.

Can we see the measurements?

You bet.

Here is a 440 Hz tone recorded with a Blue Yeti instead of a UMIK-1. The tone was generated by REW 16 -11.6 dBFS, and the minidsp's master volume set to -18 dB. I set the input volume in windows to about 50%. You can see the 1317 (3x) harmonic. EDIT I'm posting new measurements now that the air conditioner stopped running.

Here's a baseline.

 

[LTig]

Let me suggest to use the RTA window in REW. It is visually much better and shows the value of each harmonic component in either dB or %.

Spoiler: See this example from my review of the Arcam Black Box 3.

 

[Lafore]

I think you see the speaker distortion. How high is it?

Even though I changed speakers? I'm not sure I understand how that's possible, unless I have 3 speakers with the same defect.

 

[Lafore]

Let me suggest to use the RTA window in REW. It is visually much better and shows the value of each harmonic component in either dB or %.

Excellent suggestion. Thank you. I think I did it.

Spoiler

 

[kemmler3D]

I would also suggest looking at the spectrum in REW instead of Audacity, the visualization in Audacity is really sensitive to settings. If you have the same unexpected distortion from multiple amps and multiple speakers and the output of the DAC is clean then it's either the mic, or a measurement error. But if it somehow turned out to be the room, I wouldn't be surprised.

Based on your RTA above the distortion is highish but not insane. I wonder if you are hearing a problem with the room and attributing it to distortion? Looking at the waterfall might reveal any resonances contributing to a harsh sound.

 

[LTig]

Even though I changed speakers? I'm not sure I understand how that's possible, unless I have 3 speakers with the same defect.

Speaker distortion is several orders of magnitude higher than distortion of electronic devices. REW's RTA window should show differences in level of distortion for 3 different speakers.

 

[LTig]

Excellent suggestion. Thank you. I think I did it.

Spoiler

View attachment 543145

Yep, and I think the distortion values are to be expected.

A few hints:

• 32k or 64k sample size is enough
• use a lower range for the y-axis, e.g. +80 to -50 (or higher)
• use a lower range for the x-axis: 20 Hz to 20 kHz
• use averaging to reduce the noise floor (e.g. 16 or 32)
• move the distortion window elsewhere so that it does not hide the curve

 

[Lafore]

Here's a waterfall for a measurement in REW. I'm also attaching the measurement itself, but as a link since I had trouble uploading it. I'm also attaching a recording of the windows calendar reminder, which is a good example of the distortion.

Spoiler

measurement
windows calendar reminder

 

[LTig]

Here's a waterfall for a measurement in REW. I'm also attaching the measurement itself, but as a link since I had trouble uploading it. I'm also attaching a recording of the windows calendar reminder, which is a good example of the distortion.

measurement
windows calendar reminder

Was the mdat file created by the windows sound? That makes no sense. Use a sinus signal or a multitone (created by REWs signal generator). Try first in loop back mode, to get a feeling for the settings.

 

[Lafore]

oh. Well I can just share a recording of a tone with the distortion.

 

[Lafore]

Oh, I see what you were saying. The mdat was created by a REW measurement. A frequency sweep. The recording was just for a demo.

 

[Hapo]

...you could come up with more creative terms for the distortion and embrace it...???...

 

[Torin Krell]

Hello! I'm looking for help removing odd order harmonics from my stereo. I noticed a problem when certain things sounded harsh, so I started measuring. My audio chain is a digital source (TV, computer, doesn't matter) -> minidsp flex ht ->Buckeye Hypex NC502MP -> Revel F206. My microphone is a UMIK-1.

I recorded the output of my minidsp directly to my computer's line in, and the spectrum is clean, so the distortion must be introduced after that. I swapped my buckeye amp for an AIYIMA A20, changed line level and speaker cables, and changed speakers to Triangle BR02. The distortion remains.

The only other things that I can think of is that my microphone is introducing the harmonics, and the problem is just my ears or something. If anyone has more ideas I'd really appreciate you sharing.

Example harmonics are from a 333 Hz tone, I see an additional 1 kHz peak, and for a 350 Hz tone, I see an additional 1050 Hz peak.

Before you do any additional troubleshooting I would suggest that you simply tighten all the screws holding the Revel F206 drivers in place - especially for the midrange.
The crossover frequency from the woofer to the midrange is at about 275Hz, so the "problem frequencies" you designate (330Hz & 350Hz) are close to the bottom of the 5" midrange driver's range.
Lower frequencies in the midrange driver's range are more likely to push that driver into distortion if you play the F206 speakers loudly.

 

[mhardy6647]

...you could come up with more creative terms for the distortion and embrace it...???...

ahh -- the audiophile solution.