RMS detector necessary for compression??

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

loss1234

Well-known member
Joined
Aug 28, 2009
Messages
48
I have been reading and building quite a few compressors in the last 2 weeks.

i keep bumping into the THAT chips and their RMS detector.

what I am mainly wondering is whether or not RMS detection makes a big difference in the sound of the compressor or if it mainly
just effects the way the compressor reacts to transients (well i suppose that WOULD effect the sound)

i havent seen any designs that use an RMS detector with a Vactrol or JEFT though. only with VCA's.

also...are there other uses for the THAT RMS detector?

thanks so much
 
loss1234 said:
what I am mainly wondering is whether or not RMS detection makes a big difference in the sound of the compressor or if it mainly
just effects the way the compressor reacts to transients (well i suppose that WOULD effect the sound)
It makes a big difference in sound because it reacts to loudness just like the hearing process does. An envelope (peak) compressor controls the amplitude of a signal, an RMS compressor controls its loudness.
I haven't seen any designs that use an RMS detector with a Vactrol or JEFT though. only with VCA's.
Vactrol or JEFT, used as gain cell, exhibit a not particularly linear dB/v  behaviour; for this reason, they are very often used in a feedback configuration (the detector senses the post gain cell output). OTOH, RMS detectors and Blackmer VCA's have very linear and repeatable dB/v characteristics, which lend them to feed-forward operation.
It is indeed possible to use an RMS detector in conjunction with an FET or an optoresistor, but the advantages of such a combination are not enticing enough. The character of a FET or opto compressor comes in part from the unpredictable interaction between the non-linearities of its gain cell and the rudeness of a peak detector.
also...are there other uses for the THAT RMS detector?
Many:
Compressor
Limiter
Expander
Compander
Seriously, the only other application I know is some extended range audio level meter
 
The only DOWNSIDE i see to most of the RMS based designs i see is that they have no ATTACK and Release controls.
(except for design note 115)

however, I seem to remember seeing commercial schematics that had RMS detectors AND attack and release (was it on a DBX?)


what i see so far in the app notes is that the path is:

RMS detector, FWR, THRESHOLD, Ratio (and then in one case an attack/release section)

i will start trying to build some of the app notes



is it possible to use the RMS section of the THAT 4301 as a regular rectifier?

sorry for all the questions

thanks so much


 
I have already talked myself out on this subject in a thread over on the PICO forum.

I am not completely convinced that RMS is significantly different than average with similar att/rel time constants. In fact back in the day a friend of mine made a tape NR with average responding side chain that nulled very well against RMS DBX channels.

When I was researching a microprocessor based version of my simultaneous Peak/VU meter, I went through the effort of coding in a RMS conversion (SQRT of X^2) on the VU. The x^2 was easy but i had to write a little routine to do the SQRT without burning too many clock ticks.

To my (60 YO) eyes I couldn't see any difference between Average and RMS with same time constants.  [EDIT- in all fairness I was looking at complex final mix waveforms. There may be subtle differences on very percussive individual instruments between RMS and average. Still probably manageable with adequate att/rel controls.  /EDIT]

I am very familiar with DBXs claims about RMS being less error prone in their old tape NR (IMO it wasn't).

The DBX rms circuit is a good match to their VCA for simple low cost dynamics.  IMO Nothing magical about RMS but perhaps keeping the customers from tweaking knobs too much is useful  :eek:

JR


 
maybe i will head on over to the PICO thread and look for your old posts!
still trying to decide if adding more complexity to the level detector is even worth it.

thanks
 
If you need to maximize a channel which clips, you want a Peak|Peak detector.

If abused, it will sometimes do things which are odd to the ear. And Peak|Peak is only valid until you hit a phase-shift. But it is easy and widely used.

There's many ways to detect level. There's many ways to control gain. Some combinations fit-together better than others.

VCAs may be linear or not-linear. A tube VCA can be very-roughly V/dB linear, but not so good that you can blindly feed-forward; anyway a linear detector does not make V/dB output. They are usually used feedback.

LED+LDR can be quite linear over 10:1 range, but errors mount if you want more range. And ideally we use the lowest resistance part of the range, but errors rise fast. Again works best feedback.

Drawmer and others developed silicon-junction circuits. The Si junction can be remarkably predictable over a vast range. The control law is linear V/dB, not suited to simple peak-detector in feed-forward. But the same guys developed exponential converters which turn simple linear into V/dB. And also the RMS circuit, which bypasses the linear detection process (and some of its flaws) to directly give V/dB from a complex signal.

And dBx held patents on both VCA and RMS schemes, and saw it was elegant and reasonably affordable (dirt-cheap today), and made millions of them.
 

Latest posts

Back
Top