R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 2 [3] 4 5 ... 10   Go Down

Author Topic: One thing I don't understand about compressors  (Read 22088 times)

zmix

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2828
Re: One thing I don't understand about compressors
« Reply #30 on: April 13, 2006, 03:16:08 am »

index.php/fa/2671/0/

Ronny

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2739
Re: One thing I don't understand about compressors
« Reply #31 on: April 13, 2006, 04:11:57 am »

zmix wrote on Thu, 13 April 2006 03:12

Ronny wrote on Wed, 12 April 2006 22:24


I'm arguing against zmix's point. Again, compressors only increase gain on release from the compression.



This is not true. A compressor operates on a rotation point and a limiter operates above threshold.


A little reading of Michael Morgan's research (analog Devices application note AN 135) will clear this up. Here is an excerpt:

index.php/fa/2669/0/



Not hardly Z, look at "all four threshold points", unity gain going to all 4. Maybe this will clear things up. The first compressors, probably before your time, didn't have threshold settings, what Morgan is referring to as rotation point was known as the hinge point, this was fixed on early compressors. When they added the threshold setting it allowed for the hinge point to vary. Even in modern times, hinge is a better term, IMHO, as surround mixing environments use rotation to describe surround imaging. Nothing rotates around the threshold, it is a simple hinge as it lowers gain above that point. Your first graph clearly illustrates what I've been relating. Anytime that you have an X-Y square grid to illustrate compression, unity gain is represented by 45 degree lines. For example each square represents 10dB change in gain, left side is output, bottom side is input, unity gain is represented by a line going through each square from bottom left corner to top right corner. Take the first square bottom left of graph, gain is at -50 input and and -50dB output unity gain, at the top right corner input and output gain are -40dB again unity gain, next square going up unity gain line (45 degree diagonally) input gain is -30dB, output gain is -30dB again unity gain and so on, matters not where the threshold level is set, below the threshold there is "always unity gain" no increase, nor decrease in gain. As I mentioned below threshold the compressor is not operating.  
Logged
------Ronny Morris - Digitak Mastering------
---------http://digitakmastering.com---------
----------Powered By Experience-------------
-------------Driven To Perfection---------------

Ronny

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2739
Re: One thing I don't understand about compressors
« Reply #32 on: April 13, 2006, 04:22:02 am »

zmix wrote on Thu, 13 April 2006 03:14

index.php/fa/2670/0/



I prefer to go by Rane's description, although some people consider a limiter a compressor with ratio at 20:1 and above, the guy that wrote the Rane article considers limiting at 10:1 ratio. As I mentioned a brick wall limiter has a ratio set at -oo infinity with a hard knee. Here's the Rane description.



Rane Audio Encyclopedia:

limiter A compressor with a fixed ratio of 10:1 or greater. The dynamic action effectively prevents the audio signal from becoming any larger than the threshold setting. For example, if the threshold is set for, say, +16 dBu and the input signal increases by 10 dB to +26 dB, the output only increases by 1 dB to +17 dBu, essentially remaining constant. Used primarily for preventing equipment, media, and transmitter overloads. A limiter is to a compressor what a noise gate is to an expander. See the RaneNote "Dynamics Processors -- Technology & Applications."
Logged
------Ronny Morris - Digitak Mastering------
---------http://digitakmastering.com---------
----------Powered By Experience-------------
-------------Driven To Perfection---------------

Ronny

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2739
Re: One thing I don't understand about compressors
« Reply #33 on: April 13, 2006, 04:36:56 am »

zmix wrote on Thu, 13 April 2006 03:16

index.php/fa/2671/0/



These next two graphs 2 and 3, also illustrate clearly what I'm relating. Unity gain is a 45 degree diagonal line. Second graph, when the signal reaches the threshold, ratio is applied, unity gain is again below the threshold and ratio does not come into effect "until the threshold is met", this is quite clear when you understand that unity gain is represented by a 45 degree diagonal line. Note gain is at unity before "all threshold levels" are met.  

3rd graph is the same as the first graph.

I repeat, unity gain is always below threshold, regardless if the signal has been compressed or not. Because the signal is lowered in gain when compressed, yes there is an increase in gain from the "compressed signal", but again, not from the "original" signal, when the compressor is released it merely restores the signal to the gain that it was on processor input.


Logged
------Ronny Morris - Digitak Mastering------
---------http://digitakmastering.com---------
----------Powered By Experience-------------
-------------Driven To Perfection---------------

tom eaton

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 3640
Re: One thing I don't understand about compressors
« Reply #34 on: April 13, 2006, 07:51:25 am »

That figure 1b is not nearly as clear as it could be, overlapping three different input to output scenarios (at different input levels) does indeed make the graph appear to "rotate" around a point... it's a lousy way of showing how varying the input and threshold changes the output. The idea of "rotation" is totally misleading, implying that when a signal is driven heavily into compression there will be an increase in level BEFORE the "rotation" point.  That's just not the case.  The problem here is terminology.  In these graphs the "rotation" point is the point at which the compressor achieves the desired ratio.

The term "knee" has pretty much become the default term for the point where output begins to vary from input, though "hinge" is certainly similar.

-t

Ronny

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2739
Re: One thing I don't understand about compressors
« Reply #35 on: April 13, 2006, 09:12:34 am »

TER wrote on Thu, 13 April 2006 07:51

That figure 1b is not nearly as clear as it could be, overlapping three different input to output scenarios (at different input levels) does indeed make the graph appear to "rotate" around a point... it's a lousy way of showing how varying the input and threshold changes the output. The idea of "rotation" is totally misleading, implying that when a signal is driven heavily into compression there will be an increase in level BEFORE the "rotation" point.  That's just not the case.  The problem here is terminology.  In these graphs the "rotation" point is the point at which the compressor achieves the desired ratio.


That's absolutely correct Thomas, graph 1b is showing more than one input level, 3 to be exact although there are "4" threshold levels noted. I've never seen a comp with two threshold levels for one signal. A better way would have all 3 input lines on separate graphs and all inputs starting at the bottom left corner of the X-Y axis with 3 thresholds one on each graph, like 99.9% of the envelope graphs for comps do. Morgans example on 1b looks more like a compander grid, where the width is above the 45 degree diagonal line that goes from square one bottom left to the last square at top right, signifying that gain below threshold "on a compander" does indeed raise gain between the range level and the threshold level, "when the signal falls below threshold". However on a compressor the diagonal line should never exceed 45 degrees. Same on a ducker, expander, gate or range gate.  
 
Quote:


The term "knee" has pretty much become the default term for the point where output begins to vary from input, though "hinge" is certainly similar.

-t


This is also correct. The hinge point is technically at the threshold level, the ratio determines how wide or narrow the hinge opens and closes, the knee determines how fast the hinge closes and the release determines how fast or slow the hinge returns to it's starting position, which is flat open and laying on the 45 degree diagonal unity gain line before the signal reaches the threshold.

BTW, I saw your studio write-up in Sweet Notes just yesterday. Just to let you know that the free advertising that Mitch gave you has reached me down here in Georgia.  

Here's a graph of a compressor with -20dB threshold and a 2:1 ratio showing showing that the signal is outputting -10dB, not -5dB as was mentioned earlier. Hopefully the calibration markings can be seen better than the other graphs. I had a hard time reading the markings on Morgans graphs.

index.php/fa/2673/0/

Logged
------Ronny Morris - Digitak Mastering------
---------http://digitakmastering.com---------
----------Powered By Experience-------------
-------------Driven To Perfection---------------

zmix

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2828
Re: One thing I don't understand about compressors
« Reply #36 on: April 13, 2006, 12:01:19 pm »

What's with the condescending tone? I've designed compressors and these arguments that limiting is the same as compression are becoming pedantic.
index.php/fa/2676/0/
If you read the text carefully you will learn that graph 1a (above)
shows the effect of varying the RATIO while keeping the threshold
constant.
Note that the line diverges from the 45 degree 'unity' as the ratio is increased.
This is a how a compressor circuit differs from a limiter. Look at the operating level and think about where the average signal level is.

A Knee, btw, would represent a cuving of the transfer function.

zmix

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2828
Re: One thing I don't understand about compressors
« Reply #37 on: April 13, 2006, 12:09:42 pm »



There is no piece of equipment which is misunderstood more than the compressor, I think that a study of the various schematics available would dispell a lot of the hype and mythology surrounding compression.

zmix

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2828
Re: One thing I don't understand about compressors
« Reply #38 on: April 13, 2006, 01:38:24 pm »

Ronny wrote on Thu, 13 April 2006 04:36

I repeat, unity gain is always below threshold, regardless if the signal has been compressed or not.

Not true if you use makeup gain.
Quote:


Because the signal is lowered in gain when compressed, yes there is an increase in gain from the "compressed signal", but again, not from the "original" signal, when the compressor is released it merely restores the signal to the gain that it was on processor input.

Not true. A compressor 'compress' the dynamic range as it raises the signal level below threshold and reduces the level above threshold, at the selected ratio.

A limiter will reduce the level above the threshold. If you  use makeup gain with a limiter, the signals below threshold will be 'louder', but not compressed. If you want a limiter to act like a compressor (as you would for parallel compression) you would likely set the threshold quite low so that the signal is always above threshold.

If you are at all interested in learning how you use your limiters, plot the I/O transfer function and compare them to the graphs here.

joeaudio

  • Full Member
  • ***
  • Offline Offline
  • Posts: 112
Re: One thing I don't understand about compressors
« Reply #39 on: April 13, 2006, 02:36:50 pm »

zmix wrote on Thu, 13 April 2006 18:38

Ronny wrote on Thu, 13 April 2006 04:36

I repeat, unity gain is always below threshold, regardless if the signal has been compressed or not.

Not true if you use makeup gain.
Quote:


Because the signal is lowered in gain when compressed, yes there is an increase in gain from the "compressed signal", but again, not from the "original" signal, when the compressor is released it merely restores the signal to the gain that it was on processor input.

Not true. A compressor 'compress' the dynamic range as it raises the signal level below threshold and reduces the level above threshold, at the selected ratio.

A limiter will reduce the level above the threshold. If you  use makeup gain with a limiter, the signals below threshold will be 'louder', but not compressed. If you want a limiter to act like a compressor (as you would for parallel compression) you would likely set the threshold quite low so that the signal is always above threshold.

If you are at all interested in learning how you use your limiters, plot the I/O transfer function and compare them to the graphs here.



I just put all my compressor/limiters on ebay because I
now realize I have no idea what they do.

Joe
Classic
Logged

Ronny

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2739
Re: One thing I don't understand about compressors
« Reply #40 on: April 13, 2006, 02:42:45 pm »

zmix wrote on Thu, 13 April 2006 13:38

Ronny wrote on Thu, 13 April 2006 04:36

I repeat, unity gain is always below threshold, regardless if the signal has been compressed or not.

Not true if you use makeup gain.
Quote:


Because the signal is lowered in gain when compressed, yes there is an increase in gain from the "compressed signal", but again, not from the "original" signal, when the compressor is released it merely restores the signal to the gain that it was on processor input.

Not true. A compressor 'compress' the dynamic range as it raises the signal level below threshold and reduces the level above threshold, at the selected ratio.

A limiter will reduce the level above the threshold. If you  use makeup gain with a limiter, the signals below threshold will be 'louder', but not compressed. If you want a limiter to act like a compressor (as you would for parallel compression) you would likely set the threshold quite low so that the signal is always above threshold.

If you are at all interested in learning how you use your limiters, plot the I/O transfer function and compare them to the graphs here.


I haven't been condescending, I'm trying to clear up your misconceptions, showing you with your own graphs and if you can't see what I'm relating by now, there is no sense in me continuing this discussion. You are incorrect, a compressor does not operate when the signal is below threshold, that's plain and simple to understand and there isn't an engineer alive that can refute what I'm saying. I already know how to use limiters and compressors, I've been working with them for over 30 years. Your comment about "if you are at all interested" now that's condescending. If I weren't concerned about and knew exactly how my compressors and limiters operate, I wouldn't have taken the time to correct your misconceptions and your commments on what output dB's are, relative to the ratio applied, for example -20dB threshold at 2:1 ratio is a -5dB gain reduction, my graph clearly shows that you are wrong and the statement that you continue to try and enforce about gain being increased below threshold. It just ain't so Z. I can't give you any more proof than I already have. I suggest that if you want to get into discussions about compression and limiting that you post your own experiences and not what you read on the internet.
Logged
------Ronny Morris - Digitak Mastering------
---------http://digitakmastering.com---------
----------Powered By Experience-------------
-------------Driven To Perfection---------------

zmix

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2828
Re: One thing I don't understand about compressors
« Reply #41 on: April 13, 2006, 04:12:59 pm »

Ronny wrote on Thu, 13 April 2006 14:42

I haven't been condescending, I'm trying to clear up your misconceptions, showing you with your own graphs and if you can't see what I'm relating by now, there is no sense in me continuing this discussion. You are incorrect, a compressor does not operate when the signal is below threshold, that's plain and simple to understand and there isn't an engineer alive that can refute what I'm saying. I already know how to use limiters and compressors, I've been working with them for over 30 years. Your comment about "if you are at all interested" now that's condescending. If I weren't concerned about and knew exactly how my compressors and limiters operate, I wouldn't have taken the time to correct your misconceptions and your commments on what output dB's are, relative to the ratio applied, for example -20dB threshold at 2:1 ratio is a -5dB gain reduction, my graph clearly shows that you are wrong and the statement that you continue to try and enforce about gain being increased below threshold. It just ain't so Z. I can't give you any more proof than I already have. I suggest that if you want to get into discussions about compression and limiting that you post your own experiences and not what you read on the internet.



I'll happily send you the full application note. I certainly wouldn't be so arrogant as to say that Michael Morgan is incorrect. You may be right in saying that there isn't an engineer alive who can refute what you are saying. I know exactly three including myself who even understand the difference. The majority don't care, which is fine, until they start quoting sales brocures as evidence of understanding the subtleties of design topologies. I suppose any information is meaningless if you don't read it clearly. You really shouldn't feel so offended by what I said, I didn't suggest that you needed to see how to use your limiter, I stated "If you want to see how YOU use your limiters, plot the transfer function.

zmix

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2828
Re: One thing I don't understand about compressors
« Reply #42 on: April 13, 2006, 04:58:33 pm »

Ronny wrote on Thu, 13 April 2006 14:42

... -20dB threshold at 2:1 ratio is a -5dB gain reduction, my graph clearly shows that you are wrong and the statement that you continue to try and enforce about gain being increased below threshold.


Falacious argument #952: You show a graph of a limiter and attempt to refute that a 'compressor' even exists.
index.php/fa/2677/0/


If there is a rotation point at 0dB (as the 1a graph shows) a signal 10dB below this point will have a net gain of 5dB.  THere is nothing incorrect about my assertion.


Quote:

I suggest that if you want to get into discussions about compression and limiting that you post your own experiences and not what you read on the internet.

Now that's just rude...

Ronny

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2739
Re: One thing I don't understand about compressors
« Reply #43 on: April 13, 2006, 06:14:26 pm »

zmix wrote on Thu, 13 April 2006 16:58

Ronny wrote on Thu, 13 April 2006 14:42

... -20dB threshold at 2:1 ratio is a -5dB gain reduction, my graph clearly shows that you are wrong and the statement that you continue to try and enforce about gain being increased below threshold.


Falacious argument #952: You show a graph of a limiter and attempt to refute that a 'compressor' even exists.
index.php/fa/2677/0/


If there is a rotation point at 0dB (as the 1a graph shows) a signal 10dB below this point will have a net gain of 5dB.  THere is nothing incorrect about my assertion.


Quote:

I suggest that if you want to get into discussions about compression and limiting that you post your own experiences and not what you read on the internet.

Now that's just rude...




Your explanation that you weren't trying to be rude to me first is good enough for me, sorry that I took it wrong and was trying to be rude back.

I don't remember saying Morgan was incorrect, I said that he's not plotting his graph in the typical manner. The rotation point is merely 0dB input, 0dB output, the same 45 degree unity gain line that I'm talking about. I fail to see the rotation that he's talking about and I've never heard the term rotation used with compression. I assumed, maybe incorrectly that he was talking about the hinge point, where compression is applied.

I can't read the ratio numbers, but on graph 1a the middle input line, the hinge is at -20dB threshold, the gain difference is 10dB, this represents a 2:1 ratio maybe 3:1 ratio as it's not clear. 10dB gain reduction at -20dB not 5dB. Gain reduction "does not start at -0dB input, -0dB output", Z. It starts at the threshold, you can forget about the rotation point in his graph as being pertinent to what I'm relating it's unity gain, meaning the input and output are "exactly" the same, "no gain reduction", the diagonal 45 degree "unity gain line" intersects the rotation point.

You said my graph is a limiter. No, it's a compressor with a 2:1 ratio. At -20dB threshold and 2:1 ratio, the signal will be reduced in gain by -10dB when the threshold is met and the compressor comes on. You may be associating the hard knee with a limiter. Knee on a brick wall limiter is hard, as I mentioned, but I use hard knees on lower ratio's all of the time, it's entirely material dependent and what I want to do that determines knee settings for me.

Your graphs clearly show that no gain increase is below the threhold on all 3 input lines. Please understand that a line that intersects the bottom left and top right corner of each plot square signifies "unity" gain, "no compression". The graphs don't support your argument that gain increases below threshold, it does not and they clearly show that I'm correct. What Bob is saying and correctly so, is that gain increases "on release" of the compressor NOT when the signal is below threshold and the release time is over. The signal is again at unity gain and the compressor is no longer effecting the signal.  
Logged
------Ronny Morris - Digitak Mastering------
---------http://digitakmastering.com---------
----------Powered By Experience-------------
-------------Driven To Perfection---------------

zmix

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2828
Re: One thing I don't understand about compressors
« Reply #44 on: April 13, 2006, 06:30:18 pm »

Ronny,
Thanks for the reply.


Ronny wrote on Thu, 13 April 2006 18:14

 What Bob is saying and correctly so, is that gain increases "on release" of the compressor NOT when the signal is below threshold and the release time is over. The signal is again at unity gain and the compressor is no longer effecting the signal.

This a clumsy description at best. A compressor is not like a noise gate... the 'release ' is not a triggered event, but simply the discharging of the timing cap as the CV diminishes. The maximum 'gain' of a typical GR element in a limiter is unity and the action of a limiter is only to reduce gain. The only way that a limiter adds gain is through makeup gain..
Pages: 1 2 [3] 4 5 ... 10   Go Up