Need some help interpreting unexpected Distortion changes

tony22

Registered
Thread Starter
Joined
Apr 29, 2023
Posts
40
I've had no changes to my system since my last set of "reference" measurements. But after a rather nasty electrical incident I decided to run new sweeps and compare them to my references. I set up my UMIK-1 in its standard position (it's always in the same place, +/- 1/4-1/2"), and set the Calibration level at the same amplitude (75 dB(C), +/- a tenth or so). The SPL sweeps looked about the same. I did Trace Arithmetic subtraction of one from the other and got pretty much nothing. But when I looked at the Distortion graphs I was surprised (and now worried). Here's what it looked like before the incident:

1752851816882.png


and here's what it looks like after:

1752851856091.png


To me this looks like something happened, but I'm not an expert in interpreting distortion data. I don't know if slight mic position differences could cause this, or anything else for that matter other than some potential damage. Audibly, the music sounds the same as always... I think. My room is well treated acoustically and there was no outside noise or other intrusions when I did any of these sweeps.
 
Oh, that was with my REL Carbon Special sub on. Here are the same before and after with sub off.

1752852246125.png


1752852279732.png
 
Nothing of concern in distortion, all well below the noise floor. Noise floor is a bit higher and those spikes at multiples of 1 kHz are USB clock breakthrough, if using a laptop they may change depending on whether the charger is connected. Check the UMIK volume setting, if it has been set to 100% turn it down to the unity gain (0 dB) setting. You can get the windows input volume control to read in dB by right clicking it.
 
Thanks John. I will check as instructed.

What worries me is that there's such a significant appearing difference. Nothing has changed in my system between these two measurements.
 
You can get the windows input volume control to read in dB by right clicking it.
John, the level was off. It was a bit higher. Could that explain what I'm seeing here? I'm about to do another run.
 
Last edited:
Hi @John Mulcahy, I hate to keep nagging at this but I'm just confused. Is the UMIK-1 maybe just not a good device for measuring THD consistently? I ask because this morning I set up REW to take 4 consecutive sweeps, spaced 5 seconds apart, and this is what the THD plots look like (in order of capture):

1753031402105.png

1753031420434.png

1753031433807.png

1753031452245.png
 
What's your concern? The plots just show that the noise floor (both acoustic from the environment and the self-noise of the mic) are higher than the distortion at the level of your measurement. The only thing above the noise floor is a little 3rd harmonic between about 500 Hz and 1 kHz. Don't often encounter complaints that a system doesn't have enough distortion:)
 
It's not the magnitude so much. It's the variation from sweep to sweep. Look at the first sweep. There's (compared to the next three) significantly less distortion in the 30-60Hz range. The question is - why?
 
You don't show the system response, so it is difficult to comment meaningfully, but that region, like nearly all of the rest of the response, has distortion levels that are within or below the noise floor, so "distortion" is a combination of whatever actual distortion the system has and the noise. The noise varies a lot from measurement to measurement at low frequencies, which isn't unusual especially if traffic noise might be present. If you want to see actual distortion measure at a higher level.
 
I guess it just bothers me. That's a >10 dB worsening of the THD in that range, when there was no change between sweeps. No HAVC (I'd turned it off before starting sweeps, as I always do). No vehicles outside. No lawn mowers. No airplanes. Nothing. I'd attach the mdat but I'm not sure it would attach. It's about 7 MB.
 
Clearly, I need a bit of learning in this area. :reading: Can you help me understand what could cause this degree of variation in the noise floor?
 
It is most likely acoustic, you could watch the RTA to see if you can correlate the variations with things happening in the environment.
 
Okay. I'm flummoxed as to what it could be. Four sweeps. Five Seconds apart. No external audible intrusions whatsoever across all four sweeps. :scratch:
 
It is very unlikely that you can hear noise at the levels shown on your measurements, the audibility threshold at 32 Hz is around 60 dB. Hence the suggestion to use the RTA.
 
No, you are looking for the source(s) of the noise variation, as that is what is provoking your concerns, so no signal playing.
 
Hi @tony22, Just for the heck of it, upload two of your graphs to an AI, like Perplexity https://www.perplexity.ai, and ask it to analyze and compare the images...

I won't spoil the outcome... Just do it... You will be amazed...
 
Last edited:
Back
Top