How does that explain the TI links from post #6 81?
It was post
81.
Yes, all sorts of vibration-sensitive nonlinearities and odd effects exist in audio devices. Doing things like tapping on semiconductors and passive components causes varying degrees of artifacts. Lightly tapping on a mechanical relay in a preamp with headphones is an example of a seemingly mundane component that is sensitive to vibration. In some high-vibration applications the artifacts from hammering and pounding matter. Home audio can't be considered high-vibration.
Similarly - microphonics of ceramic capacitors is understood. Though I've never seen any demonstration of audible effects in audio gear.
And again all these examples involve tapping the equipment - resulting in effect very much higher than you will get with any audio device sat on your shelf at home.
At risk of repeating my self - if you are worried about audible microphonics from tapping your gear, don't do it.
I couldn't agree more.
Tapping until you get an audible response is completely unrepresentative of actual use.
So at the risk of opening a can of worms, how about trying with sound?
I have a QA403. It's made of the same stuff as a typical audio device, with the ability to self-test, and a very low noise floor.
I configured the left channel in loopback mode. I grounded the inputs of the right channel. This way if I see artifacts due to vibration, I can maybe isolate them to input vs. output circuitry.
I used three sources of excitation. A Genelec 8361A, a JBL M2, and a Technics 11" woofer that I had lying around. I tried a number of signals, including multitone. I tried a number of volumes all the way up to eleven
, and a number of distances including
point-blank and
sitting-on-top:
More on the sitting-on-top configuration later.
First, here is the noise spectra of the QA403 with no audio playing, only the background sound of the room.
In all of the graphs, L channel is blue and is measured in IO loopback, R channel is red and is measured with inputs grounded.
It's clean with no input to the speaker. The loopback measurement being slightly higher floor than with inputs grounded.
I mounted the QA403 on a mic stand and put it 5 cm in front of the front baffle of the 8361A and blasted the analyzer with music at maximum volume while measuring the spectra of the QA403 as it was vibrated by the speaker's output:
That was loud. The QA403 is unaffected by the vibrations, even at only 5 cm from the speaker playing it's limit lights flickering amber.
I did the same with a JBL M2, in this case I put it in front of the woofer which should excite it up to 800 Hz or so.
No effect, despite being terrifyingly loud at ridiculously close range.
This is good news. I assume that the QA403 doesn't have any dodgy or cost-saving applications, like ceramic capacitors with well-known piezoelectric properties in circuit applications that are sensitive to capacitance changes. The Genelec have amps built into them, so this is also good news.
I wanted to try multitone excitation. I used an 11" flat woofer from an old Technics speaker. It allows me to place the QA403 directly on top of the driver's planar surface.
Here is the QA403 at 5 cm above the flat surface of the woofer:
Nothing, to see. This woofer doesn't play as loud as the other speakers, but I thought it would be interesting to directly excite the analyzer by putting it on top of the woofer like the photo above.
I placed it directly on top of the woofer and repeated the experiment, feeding it well over 100 Watt multitone signal, the limit of what the woofer can do.
Interesting. You can see evidence of the multitone signal at 40 Hz, 50 Hz, 64 Hz, 80 Hz, etc. at -140dB, rising above the floor by ~10 dB. Also interesting, I can get the same or worse by holding the analyzer just above the speaker, or holding the cables I use for the IO-loopback measurement on the left channel just above the speaker. The question, is it the vibrations or the stray field from the Voice-Coil making all of this -140dB noise???
To investigate this, I repeated the above test of the analyzer sitting on top of the woofer, but physically pushed the driver deep into the magnet until the voice coil was no longer in the gap. At this point, the driver barely vibrates because the VC is no longer in the motor. Here is the noise spectra with the same high-power signal but with no significant vibration:
It got worse, even though the woofer is no longer vibrating. This isn't vibration causing the measured noise, it's the oscillating field from the woofer's voice coil inducing noise in the analyzer's inputs - note that it shows up in the left channel loopback test but not in the right channel with the grounded inputs. It gets worse when I push the woofer in because the analyzer is closer to the motor's core.
It also made the woofer's voice coil very hot, enough I could smell it. It's also extremely unrealistic stress, not representative of audio. It's also something we already know, inductors couple to each other and their surroundings. The old speaker-building advice is
don't mount large-bore inductors near each other since they will couple, which is also another audio-overstatement; the effect exists but is minor except in extreme cases.
Where does that leave us?
Even absurd volume at point-blank range has no measurable effect on the circuity of a high-resolution audio analyzer.
By the time you get close enough to a driver to observe a signal, the field from the voice coil has a comparable or larger effect than the vibration caused by a speaker. Likely much larger. Even if the signal was vibration (it isn't), -140 dB isn't audible.
So aside from not tapping on our audio gear, keep it away from large electric motors.
Of course some devices may have higher levels of vibrational susceptibility than this analyzer. Especially if they use components like ceramic capacitors with piezoelectric properties in the feedback loop of a high gain device.
A cheap example from AliExpress was
recently tested on ASR. I bet it is also vibration-sensitive, but certainly not the worst of it's problems.
edit: typos and added a legend to the graphs of the left and right channel measurements
