'Sprung plank' circuit boards and mechanical feedback

[suzywong]

Been there, 25 years ago, I worked on a project to build RF frequency generation for use in aircraft. We designed and built them to our usual “space” standards, and tested the units on our vibration rig to simulate being in an aircraft, and watched as the phase noise got worse and go out of spec. Needed several weeks of trials to get it sorted. IIRC, capacitors & particular diode were the culprits.

Took us by surprise. Normally the stuff we built had merely to physically survive launch (powered down), so components would be bonded to the printed circuit board, and the boards mounted in milled aluminium boxes, the theory being that once on orbit, vibrations would be so low as to be irrelevant.

After that, the industry started analysing for on-orbit “micro-vibrations.

 

[wd40addict]

I once read an article in a professional magazine back in the 90s where they visited Naim and been shown the spring suspensions etc. The journalist expressed skepticism, but was then shown how the noise level fell on an AF spectrum analyser. After that he believed. I would assume this was inside a CD player.

 

[Mynamemynaim]

Some of Their CDs streamers and phono stages were "sprung"
A heavy brass plate on springs

 

[PigletsDad]

While testing power amps, you can sometimes hear them "singing" at full output for higher audio frequencies. There is a reciprocity theorem that means that anything that converts electricity to sound will do the same thing in reverse.

 

[a.palfreyman]

Interesting @PigletsDad because when I ran it with test tones, looking at input and output on a 'scope to see if there was anything obvious, I could hear a very quiet 'overtone' behind the main frequency. Need to check my sig. gen. through headphones to see if it is the sig.gen. though before I draw any conclusions.
I was only running at 0.56vrms to speakers as any louder was too loud!

 

[martin clark]

While testing power amps, you can sometimes hear them "singing" at full output for higher audio frequencies. There is a reciprocity theorem that means that anything that converts electricity to sound will do the same thing in reverse.

Yes I've experienced that. Most often the capacitors - no surprise, the electrostatic forces involved are surprisingly large to huge.

Also its exactly why I prefer tightly-wound, industrial -quality, perhaps boxed & resin-filled types - over any variety bijou hand-rolled aftermarket nonsense. Even non-nonsense supplier Wima sells a -LN variant line of some of their products - where 'LN' is expressly Low Noise in a mechanical sense; compare datasheets for MKS4, and MKS4-LN for example. Conversely, flexing a pcb that has large-format SMT caps bonded on can do similar - esp. the really high-K types, like X7R*

I've seen hints of vibration-sensitivity on a few DAC pcbs - likely the X7R issue above - but only at an impact level level that is far higher than /not representative-of background vibration (i.e - tapping a part with rubber end of an old pencil). Never seen anything like that in pre- and power amps I've ...played with.


*Add that to a list of MLCC and esp. X7R sins, like excruciating capacity-derating with both applied-voltage and aging.

 

[PigletsDad]

See https://www.allaboutcircuits.com/te....-a-concise-guide-to-ceramic-capacitor-types/ The X7R and even worse X5R dielectrics are piezoelectric. Any capacitor will have a value which changes under mechanical stress, as the dielectric gets thicker or thinner. This obviously changes the voltage across it, as charge is conserved. At a lower level, you expect resistors to show the strain gauge effect - if you change the dimensions, the value must change.

All these effects should be small, but will be worse for circuits with large standing voltages on caps and resistors - like valve gear. Don't get me started on microphonics in valves!!!

 

[bugbear]

This has been one of the interesting, informative, cheerful threads I've seen in a long time.

Well done all!

 

[a.palfreyman]

Yep, seen this effect with Wima stacked MKS 0.68uF cap on the front of a MM phono stage. Tap the cap and you get quite a low frequency 'pulse' at the output.

 

[Dowser]

In the Naim CDI the mech is rigidly bolted to main PCB, and the PCB is mounted on springs in the chassis.

 

[laverda]

A few 'decades' ago I used to suspend the transformers of my amps on a sprung sub chassis damped by rubber grommets. The sub chassis and transformer were able to 'float' on the 4 springs beneath. To my astonishment there was an audible difference, for the better.

I should pull my finger out and implement this mod on my current amps.

 

[bugbear]

A few 'decades' ago I used to suspend the transformers of my amps on a sprung sub chassis damped by rubber grommets. The sub chassis and transformer were able to 'float' on the 4 springs beneath. To my astonishment there was an audible difference, for the better.

I should pull my finger out and implement this mod on my current amps.

I wonder if you were protecting the transformer from external vibration sources or the rest of the amplifier from the vibrations caused by the transformer. Both seem plausible!

 

[laverda]

I wonder if you were protecting the transformer from external vibration sources or the rest of the amplifier from the vibrations caused by the transformer. Both seem plausible!

I believe it was a bit of both but the latter predominantly, the vibration from the transformer can in many cases be felt with the hand, some Naim 250's I've worked on are very bad indeed, you can feel the transformer objecting to being asked to do its job. If something is making a noise its vibrating, simple physics.

My thinking at the time and still is...eliminate the noise and your eliminating the vibration which can and does impact on the rest of the equipment attached to the same housing/chassis as the transformer.

 

[Mynamemynaim]

I guess sorbothane is the "modern" equivalent to springs

 

[a.palfreyman]

Well I checked it again today running as before off the sig gen. Amp placed on it's feet direct on shelf. Same setting as before (0.56V RMS at speaker) and couldn't hear any overtone...
However it did remind me of how badly I get different nulls/peaks at the listening position with changes in frequency. An inch or too in any direction causes changes in the effects of the nulls. Maybe that's what I'm hearing. Only plausible explanation I can think of. I'll set it up again with my low distortion signal source and filter with o'scope and try tapping the circuit boards. If I can I'll also try one channel with that set up with a dummy load whilst playing music through the other channel into a speaker.