Microphony considerations - test

[RobHolt]

Tony, I think a certified lab is required if you want to have results meeting industry standards. However, if you just want to test if something works or not to within quite a large tolerance it's not really necessary. I can't see a way that doing Rob's tests in a lab will cause there to be anything measurable on the audio output when there wasn't before. If he wants to start testing the capability of different isolation devices then maybe something more is required.

What would people like to see as the next step?

I think the ideal test signal would be a square wave with something like a 4Hz frequency. It contains all frequencies and has a repetitive transient pattern to spot easily in noise.

On the first point, yes absolutely right.
What matters is how products behave under domestic conditions and that doesn't require the hiring of a lab. Industry standards here don't exist - I wish they did - standards are sadly absent from high end audio work.

The one final follow up of course would be to do a couple of recording with the amp back in the usual place. I'll do that over the next couple of days but I guarantee the result will absolutely nothing audible, even using the high gain MC stage.

This type of test is all about putting things into perspective.
I don't need to hire a lab to determine that capacitors, ICs and wirewound components, for example, are microphonic. Home tests can demonstrate this and if anyone wants me to show microphony in a capacitor, I can quite easily do so.
This is simply demonstrating some basic electronics and tells you nothing about performance in situ.
What matters is how these components respond under typical and extreme domestic conditions, and what we've shown here is that SS circuits need very extreme conditions in order for microphony to be audible.

The ultinmate test for any equipment when modifying the design is to ask the question 'does the output change in any way post the alterations?'
You can damp cases, change the feet, fit 500 regulated lines and bodge away to your heart's content, but if nothing changes on the output it all amounts to nothing.

Of course some people really do maintain that they can hear things that just aren't there. Not much can be done about that!

 

[RobHolt]

It's not mine; as I said before, a friend owns it. As you stated earlier, one doesn't go round banging on ones amps so it doesn't really mattter. However, the fact that any sound is produced at all is the surprising thing.

A SS power amplifer that produces clearly audible noise through the loudspeakers when you tap the case has some clear issues with the design. That is my point.

What it does in use would need to be tested. Such an ampifier might not be ideal for placing close to the loudspeakers. Anyone doing so might well conclude - in fact they probably would - that the shorter cabling is having some effect, or that the amplifier gets into trouble at high levels. It could be a microphony issue and a fairly simple test will deliver the answer.

Don't bother to test it and you have a range of sujective responses open to you, all of which could be wrong in this instance without some basic facts to inform the opinion.

 

[bottleneck]

- Lastly the line stage. You can make up your own minds but I for one will never be the slightest bit concerned about these things, or having such electronics inside the loudspeaker cabinet.

This certainly dispells some of my worries about 'active' loudspeakers having solid state amps inside them.

My other main bug-bear (the quality of a loudspeaker hidden inside a case) remains of course and is not relevant to the subject in hand.

 

[RobHolt]

To conclude, this is what we get by placing the pre amp back in the rack and running the drum test again. Same playback level.
The amp sits in a little wooden Quad rack to which it is bolted, and the rack sits on a large record storage unit. There are no special supports, cones or other attempts at isolation. Distance from the left hand speaker is about two meters.

Here is the Line input - full gain on the pre and recorder:

This is the sound file:

http://www.mediafire.com/file/dxt6ro24rjn9fsv/line drum at rack.wav


Same thing but using the MC phono card:

The sound file:

http://www.mediafire.com/file/iwu8zvrxe7f5b0e/MC drum rack b.wav

Nothing but residual circuit hiss.

 

[TonyL]

Nothing but residual circuit hiss.

If you crank it to insane levels on headphones there is a slight rhythmic quality to the noise, just a little interference really, and so far down into the noise-floor I can't for the life of me see how it could ever be relevant. I tried normalising it to get a bit more amplitude and so I could look at the waveform. I am a geek.

Tony.

 

[Werner]

Not that I am a subscriber to the theory that everything matters and needs expensive solutions, but if you do this effort you might as well do it well.

The above tests looked at vibration-induced noise. That's one thing.

Another thing is to see what happens when vibration induced changes to the values of components, such as resistance and capacitance. Think silicon strain gauge and you know that I mean.

For this to be detected you need to excite the device with a signal, and then with acoustic energy. They you perform distortion analyses on the signal, for all types of distortion.

 

[Tenson]

Do you mean the components sensitivity to vibration will change with a signal going through them?

 

[RobHolt]

If you crank it to insane levels on headphones there is a slight rhythmic quality to the noise, just a little interference really, and so far down into the noise-floor I can't for the life of me see how it could ever be relevant. I tried normalising it to get a bit more amplitude and so I could look at the waveform. I am a geek.

Tony.

Same noise exists if you mute the speakers. You'll notice that it persists even when there is no druming. Just noise from huge amounts of gain.
60dB from the MC stage, 31dB from the Quad line stage, plus whatever the Mac is adding at full pelt (the sound card was running at max) - call it 20dB for cash.
100dB of gain + whatever you apply when listening.

That's.......a lot!

 

[RobHolt]

Not that I am a subscriber to the theory that everything matters and needs expensive solutions, but if you do this effort you might as well do it well.

The above tests looked at vibration-induced noise. That's one thing.

Another thing is to see what happens when vibration induced changes to the values of components, such as resistance and capacitance. Think silicon strain gauge and you know that I mean.

For this to be detected you need to excite the device with a signal, and then with acoustic energy. They you perform distortion analyses on the signal, for all types of distortion.

Do you mean the components sensitivity to vibration will change with a signal going through them?

Well worth investigating though there is little you can do to control internal component vibration using external means on certain components.
Valves are a good example. You can damp the glass but you can't do anthing to the internal plates and structure. You can take measures to prevent noise reaching the internals but you cannot directly control the internals.
Same with caps really and silicon components to some degree.

Testing individual components is quite easy. You need a scope, signal generator able to provide the required single or swept tones, a means of replaying the tones at sufficient SPL and some software able to perform analysis and generate the distortion stats.
I've that to hand as does Simon so it isn't difficult to do some testing. It shouldn't be necessary to hire college facilites.

I think it will largely tell us what we already understand though.
Just about every component can be shown to display some microphony, from severe in the case of certain tubes to the barely detectable and then only with extreme levels of stumuli, metal film resistors spring to mind.

As a very simple example of a more focused test, you could measure the resonant behaviour of a valve's glass tube with a swept test tones. At some/various frequencies the glass envelope will be excited. You can then apply different damping materials and observe the result.
You can do the same with silicon components by applying damping or pressure to them and observing the output, though you'll have to look a lot harder in some cases!

Where it gets tricky is relating all of this to the real world.
I'm sure a thesis could be produced illustrating the degrees of microphony in different components but what matters is the degree to which these effects are present in normal use - or even pretty abnormal use as in these tests.

On the dual method of looking at distortion on a signal passing through a component that would certainly be worth doing IMO. This is one area where nul tests would be useful. Tone recorded with external vibration present - tone recorded without - attmempt a nul.
Gets difficult when using lots of gain because of all the randon noise generated but probably do-able with digital hardware.

 

[Werner]

Do you mean the components sensitivity to vibration will change with a signal going through them?

No.

For the sake of argument imagine an amplifier that changes its transfer curve (i.e. distortion) under vibration. How would you know without a signal passing through it?

 

[wadia-miester]

I think the ideal test signal would be a square wave with something like a 4Hz frequency. It contains all frequencies and has a repetitive transient pattern to spot easily in noise.

Just as long as you do not put it through a TVC pre-amp LOL