IMD and compressors

I've been curious about intermodulation distortion measurements on compressors.  There are very few manufacturers who list this in their specs.  So, is this because its not a flattering test and most don't really won't to advertise that, or is it not a really meaningful test for compressors?  I'd be curious to see what the IMD measurements are for some of the common ones.  Like the GSSL, G384, C1 and C2.  I'd like to see testing with compressor in and no compression as well some normal use settings.  Anyone done this?

Regards,
Jeff

It would seem meaningless unless you could capture it dynamically, since a static test won't at all show what occurs when GR is asked to do something with a specific and unspecified dynamic source.  I'm sure you could produce a number through a large number of averages, but how do you define and explain the averages?  I'm not sure it's meaningful, since something is being changed; a bit apples to oranges.  I expect IMD would tend to go up most during shifts; they would need to be well defined in the test, and I don't think the real world allows that.  Unless you get everyone to agree to a standard, yet compressors are designed to do different things, so even that seems impossible. 

The classic 60Hz+4KHz IM test signal is sure to embarass any fast limiter. 60Hz is just too close to fast attack constants.

Most VCAs run naked devices open-loop. The IM effects can be calculated from the HD effects, no complication from NFB. Traditionally, classic IMD tests on simple full-band amps showed a number 4X higher than the THD number.

If you really look at it, any not-slow comp/limiter MUST "modulate" the signals. Play a steady flute tone. Then bring in the kick-drum. The flute must duck each time the drum kicks. This is the same action as speech AM-modulating a radio carrier. Well, only downward modulation for comp/limiters, but peak-expanders (yes, we used to do that!) will up-modulate.

If it does not grate, forget about it. If it does grate, stomp less.

We won't bring the fame of the famous fast comps down by exposing their IMD.
Best just let sleeping dogs lie and enjoy them and the music.

Bitman, I won't bring down the famous ones ;D.  My intention was purely to make sure my compressor was the best it could be!  This whole thing started when my wife came in while I was starting to setup some mixes for her.  She mentioned she didn't like the buss compressor engaged.  After a short discussion of why it was there, I decided not to ignore her comment as she may be hearing something I wasn't.  I tested my C1 clone again and found higher IMD than I remembered from my initial tests (probably due to the settings).  After searching the internet for IMD measurements on compressors, I found very few had included these in their specs as well as little information on using this particular test on compressors.

I did find that even with the compressor engaged with minimal compression the IMD levels were very respectable and the IMD levels only increased upon more aggressive settings.  THD+noise% was excellent at all settings.  It would seem that this test could be meaningful during manufacturer production line testing if a specific repeatable setting is used (per emrr).

Thanks for the info PRR, I do use my ears but like to verify as sometimes we can be subjective and hear what we expect/want to hear.

regards,
Jeff

IIRC I thought the old SMPTE IMD was 60 Hz and 7kHz, but not important. To characterize linearity of a compressor path you would need to defeat the side chain gain modulations.

Without going inside the unit, if you have a side chain jack you might inject a very HF fixed level tone (say 50-100 kHz).  If you can get the control voltage stable enough you could perform THD and IMD testing at different amounts of gain reduction. Note: perhaps LPF the signal at say 20khz before sending to the analyser to filter out remnants of side chain drive.

This might be useful to find a really squirrely circuit path, but in general the sound character of dynamics processors is dominated by control voltage manipulations. IMO only of academic interest to find a clunker. It will not reveal much about why something sounds good.


JR 

Yes John you are correct, SMPTE IMD is 60Hz and 7KHz.

The reason this peaked my interest is that I didn't hear any issues with the compressor yet IMD measurements appeared to indicate problems.  In fact I believe my C1 sounds excellent!  It is aggresive but even on higher compression ratios and fast attack/release settings it still sounds good.  While doing initial testing on my tube microphone preamp, higher IMD settings did reveal an issue that allowed me to find and resolve.  Also, it was very audible and not pleasing.

Sounds like this test could be used to find "clunkers" or major flaws but I would assume that THD tests would reveal this just as well. 

regards,
Jeff

In similar band passes and within reason, you can roughly predict IMD from THD and vice versa,,,  I stopped measuring SMPTE IMD a few decades ago since it was mainly reflective of 7 khz linearity. I switched to 19:20 kHz IMD as more stressful while still using in band test stimulus. With 19:20 kHz 1:1, we see slew rates approaching 40khz sine, with distortion products that occur in band so are more likely to be audible. The equivalent THD testing at 40 kHz would be compromised by GBW and frequency response stepping on distortion products. 

The sound quality of dynamics will be dominated by nonlinear tricks in the side chain control voltage manipulations. I have done a lot of work with companding noise reductions and in my experience CV modulation dominates over even very modest (IC) gain cell performance. The current generation of THAT VCAs are quite good so not suspect, unless misapplied.

JR