Noise at 192kHz that maybe doesn't matter


#1

When reading this please note, this is not to spark an outcry, it is merely to ask if I need to fix something in my system.

So I was minding my own business as much as anyone can on the internet looking at linear power supplies when I clicked on this article. So if you read down a little bit in the link you run into some downloads of .wav files that you can play on your system to see if you have “nonlinearity causing audible intermodulation of the ultrasonics”. Whatever that means.

After a bit of trepidation about downloading from some random site on the internet, I finally did. I played them all, and the one that stood out for me was the 192kHz. I got a lot of noise playing that in my normal setup. So just to make sure I tried a few different ways, Roon to DirectStream, mcontrol to DirectStream, mcontrol to integrated, and finally usb stick straight into integrated, and still the noise like static.

My question is whether the file is really pushing the extremes of high frequency and not something realistic so not to worry, or do I need to try isolating different parts of my system to see if I can find the source of this “intermodulation”.

I like the sound of my system, I enjoy high res audio, and really like dsd. I never hear this static when playing normal files. But I do want to figure out the noise thing and if I should try to fix it.

Thanks!


#2

Playing sounds with too high of a signal in the ultrasonic frequencies can destroy (and I mean that literally) your tweeter or worse, your speakers, even if your system is perfect (perhaps even more so with better systems that amplify the ultrasonic tones accurately.)

It is the case that non-linearities can cause frequencies show up that weren’t in the original signal, but that can happen at times even with no non-linearities in the ear itself. Here’s a little more about it: https://en.wikipedia.org/wiki/Combination_tone

In any “real” sound the levels of higher frequencies are smaller than the levels of lower frequencies. e.g. with any real recording you’ll not get levels near the maximum even for 5k, let alone 30k. (The ear is very sensitive in the 3k region so not a lot of signal is needed there anyway.)

Non-linearities in a system can be benign at reasonable levels but cause really bad things if the levels are too high. Be careful running tests without knowing exactly what’s in the tests. If you must run such tests start at a very low level and increase the levels slowly. I have wrinkled my metal dome tweeters by playing 80k too loud (and they cost $1500 each.) IMO don’t go looking for trouble.


#3

Thanks Ted, as always, great explanation. My pocket thanks you a lot as well!