Microphony considerations - test

As promised some time ago, here's a little test for microphony in solid state audio circuits.
This is not presented as a case that such effects do no exist under any circumstances - it is simply a practical test that can hopefully inform further discussion.

When doing these sorts of tests I like to make the conditions as extreme as reasonably possible, on the basis that real world use is very likely to be kinder and any results can be scaled accordingly.

In this instance we are looking at a a pre amp. I have a few models knocking around but to satisfy the above it seems sensible to chose something with complex circuits, containing lots of active and passive components and importantly, something with plenty of gain. My Quad 34 fits the bill - lots of active op amp stages, capacitor coupling, arguably too much gain and with no attempt by the manufacturer to protect the circuits from either structure or airborne interference. It is a circuit screwed onto a light chassis and slipped into a thin steel sleeve.

See here:




The pre amp was placed onto a loudspeaker (Rogers Export) bypassing the rubber feet and place on-end so as to expose the underside to the output of another speaker placed 12 inches away. This ensures that the unit receives vibration from the speaker below, and is exposed to the SPL of the speaker sitting to the right.

Talking of SPLs, as this is a test the speakers were fed with wide band pink noise with the SPL meter measuring around 105dB at the test position.

See here:






The pre amp CD input was selected and volume (remember this is active with gain) advanced to maximum. The output was taken from the pre amp out and captured via line in on a Macbook Pro using Audacity. Line gain on the Mac was again set to maximum.

Once the file was captured into Audacity, a further 50dB of amplification is applied. I did say this was an extreme test!

The result is as follows:

Here is the amplified WAV file showing the pre amp residual noise:




Here is the 30 sec WAV file for download and listening:

http://www.mediafire.com/file/2i9e91cpcwdpj2e/Microphony.wav

The pink noise was applied part way into the recording - so can you hear or see when this happened?
No, me neither.

Food for thought when it comes to the purchase of special feet, cones, spikes, racks and other assorted accessories for solid state gear - certainly at line level.
We'll look at valve circuits, TTs and high gain phono sections in parts 2 & 3.

 

105db? that is LOUD

your neighbours must love you!!

will be interested in hearing what happens with valve amps and phono stages. turntables I suspect being a different result but we will see..

 

- bear in mind the meter is only about a foot away.

 

Great test Robert. Conclusive results.

Any chance of repeating this test with Naim amps? For the benefit of all those user's who've spent £hundreds on Fraims or other supports.


Some valve amps and all turntables (with the possible exception of the ELP one) will be microphinic. It'll be interesting to find out how microphonic.

 

104.7db out of a pair of Rogers Export speakers without completely destroying them? Amazing!

Tony.

PS it might be be worth repeating but with some thing rhythmic rather than pink noise, i.e. fire a four on the floor 909 kick drum or what ever at it, something that might be easier for humans to detect than noise on background noise. I'd not do it with a nice old pair of Rogers Export though!

 

Interesting.

With TTs suports have a mechanical effect but I have often wondered about microphony in circuits. Still some people swear by them (supports) and not all effects can be measured. I bet you could not take any readings from an LP12 fitted with a steel and then Keel subchassis that would provide measured backup to the difference in performance.

Thanks for the time taken to do this Rob, I heard no effect.

 

None heard here either. Now dig out a stand alone phonostage with 60db and do it again, just for a laugh.

 

Remember thats 104.7db at 1 foot.

I'll include some music next time. Though if you play the wav noise file and apply even more gain, there is still nothing on it.
There a huge amount of cumulative gain here, far more than any home system would apply.

Could do, but I'd be amazed if you got a different results given it is still a SS circuit using similar components and gain.

The 34 does have a MC phono card installed. I did have a quick try with that but is still showed nothing, which surprised me a little.
I'll try a WD valve stage next weekend - 3xECC83 and MC input transformers.

 

Hi Andrew,

I think you could but it would be a different test.

Some time ago on pfm I put up some recordings of arm/sub-chassis sound and those were markedly different - the resonant behaviour was very different.
Changing the LP12 sub-chassis definitely has a measurable effect.

 

In what way did they measure differently?

 

Resonant behaviour.

The recordings were made with a modified microphone capsule with the moving element in contact with the arm board/sub-chassis. A recording was then made and on the playback you can clearly hear the sound of the TT structures responding the the energy from the cartridge.
Its an old test done as part of school project back in the 80s and I transferred the results from cassette tape.

If you then plot the spectral content of the file you can clearly see the different energy distribution.

Its on pfm somewhere - I'll look for the link.

<edit> here you go:

http://www.pinkfishmedia.net/forum/showpost.php?p=1162861&postcount=10

 

I don't see why the amplifier was not placed on its feet rather than on its side; surely variables should be kept to a minimum.

Perhaps the test could be re-done using music rather than pink noise. I've done tests like these in the past with regard to turntable suspension/isolation and found music to be more useful than pink noise.

A friend of mine has a power amp built by one of the UK's smaller manufacturers. This amp has a thin aluminium wrap case. If you turn the volume up a bit and strike it on the top of the case a clear 'boing' sound is heard through the speakers. Now this can easily be damped out but, never the less, the transistor amp is picking up mechanical vibration and amplifying it.

Years ago, another friend had a counterpoint pre-amp (valve). If you spoke into the perforated top of that you could hear your voice through the speakers (volume set to normal listening levels).

At Andrews first bake-off Pete (I_should_coco) brought along a valve phonostage he had been working on. Whilst this sounded like a fine unit, it proved to be so microphonic that it behaved essentially like a microphone. On some of the recordings made using it you can hear people chatting on the sofa just before the music starts (they were sat 6ft away or so).

 

Thinking about it, would we be able to pick out the 'noise' anyway? I'm sure you could sneak quite a bit of pink noise past me, it is 'dither' after all.

I think this test falls short, somewhat.

It should really be, output of amp recorded and digitised, and then nulled. But with real music playing, a bit of the 1812 should do it, you can leave the case on, amp on its feet, just point the speaker at it.

Record it twice, once with volume up and once with none. (Obviously not using the amp you record to provide the drive to the speakers, we do need to remove speaker load effects and back MEF from the equation).

To be honest I'm surprised it wasn't more rigorous, hardly suitable for any proclamation of the effects of feedback.

 

I have to agree that pink noise is not a good test since it is basically the same thing as the background noise.

How about Jim Keltner, that drum solo track. Transient should be easier to hear, or see.

 

Rob,

I applaude you efforts in this area, and for our own experiances I firmly believe in microphonany having an audiable effect on sonic quality.

Though may I suggest that a commissioned study by say a recognised University, using an accredited lab and an accelerometer, data capature equipment etc, plus benchmarked equivalent test would be more in keeping with industry professionalism.

I say this simply as it would add proper creedence to any findings you may ecounter.

Tony

 

You forget the levels involved and the cumulative gain - which is enormous in this test.
Given the SPL, the overall nose levels from the pre should have moved perceptibly both on the graph and on the sound recording. More than happy to run it with music, but don't expect to see anything different. There are pitfalls using music as the case or components might be prone to microphony at certain frequencies - in fact that is almost certainly the case. Wide band noise is more likely to to provoke a response where music might simply lack energy in the area where the internal components are most microphonic.

You cannot do a nul test as the pre amp circuit noise is random.

The test was intended to represent an extreme condition. Therefore the rubber feet were bypassed to remove any filtering effect.
Also by placing the case on-end, the entire bottom of the unit is exposed at close proximity to the sound source.

What you have to bear in mind is that the pre would ordinarily be sited many feet away from the speakers and on a support, not a louspeaker, and that the SPL at the listening position would be greatly reduced.
....and you wouldn't have anything like the gain used used here.

On striking the amplifier case, well you don't in normal use and i'd suggest you need insanely high SPLs of the type that would cause fast and permanent hearing damage to impart as much energy into the case. I'd be be interested to see the amp design.

 

Hi Tony and many thanks.

Of course you may suggest it, and when I see such commissioned studies, benchmarked testing and industry professionalism supporting the sale of products designed to reduce microphony in electronics I'll give it some thought.

regards,

 

The above is all true in relation to a Quad 34, but I feel the test could be looked at beyond that scope, i.e. can we prove the concept of microphony in solid state electronics or not? I'd be interested to try the 34 with it's outer case off and standing directly in front of a speaker playing say a loud repetitive 909 kick drum pattern or club breakbeat - we should try and establish if the phenomena exists at all first, and only then try to establish what is going on. Bare in mind there are many active systems with amplifier sections built right into the speaker cabinet, so this test is not as crazy as one might think.

It is also important not to form absolute conclusions as to what (if anything) is happening with stands and other audiophile tweaks. These need assessing separately, i.e. can people identify the presence of say a Mana stand unsighted. It is still a leap to conclude microphony is the only factor at play IMO. It's also worth noting that most stand dems, e.g. the famous 'Dancing Steve' QS show dems always used a CD player, not an amp as the test subject, i.e. a mechanical component and with both Mana and QS the posher top bit of the stand is used to support a source component, the amps tend to slum it lower down the stand.

Tony.

 

Hi Tony,

No problem I can run it with a drum track and no case.
Perhaps on line and MC phono though applying 50dB of gain to the latter might be a challenge.

This is a moving subject and the test is a start, as I made clear in the very first line in anticipation of the naysayers - (not you I should add)

I'll re-run it this afternoon and also try the valve phono stage.

Your point regarding active speakers with amps is interesting and very relevant.
I would say that personally i think that microphony in SS kit in a nonsense and the many good active systems tend to reinforce that view.
Now, you can show a positive result - utterly insane levels of gain and striking an op amp or capacitor with a screwdriver will certainly produce a result. But such testing is so far removed from reality that I can't really so the point of doing it.

On CD players, unlike with the pre amp that is one area where it is possible to do a nul test by taking the digital output and nulling on and off stand. Certainly something we can try. Given the quality and effectiveness of the error correction on stimuli sat greater than conducted through a hi-fi rack, I'd be astonished if we got a positive result. We can do it though and i've a Meridian 206 knocking around.

In the case off test for the pre amp using music I'll need to redice the SPL a little (to protect the Rogers but 95db at perhaps 6 inches should do it.

 

I'd love to see the test repeated with olive Naim CDS-class players, preamps and power amps and a concrete slab floor with various brands of supports. As I've mentioned here before, olive gear is extremely microphonic when lashed up normally and music playing through it. In other words, all stands I've tried make a difference to the sound. Most expensive audiophile stands degrade the sound IME with simple, low-mass and rigid furniture providing the best results or placing the units directly to the carpeted floor.

It's a shame I don't live closer, Rob. I'd happily volunteer the gear and room for an experiment.

regards,

dave