FET Compressor Ratio Question

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

LazyTurtle

Active member
Joined
Apr 18, 2017
Messages
43
Hi,

I´m trying to understand how ratio is controlled in a FET compressor (1176 for example) and still don´t get it.  While having the FET in the ohmic region, its rds can be controlled using a control voltage in the gate. If used as a VCR in a voltage divider the FET can control the output voltage using DC control voltage at its gate.
I understand that, if ratio is pushed to higher gain reduction, this Control voltage (in N channel JFET) must be turned to higher values (from negative to 0V), reducing rds and, therefore, reducing the outpu voltage. As a compressor must be dinamically dependant on the signal this control voltage must be varied in function with the signal asigning more reduction depending on the threshold and signal level.
Is there info on how this works exactly and is there any equation/function/method to easily configure ratio in a fet compressor?

Thank you in advance for your answers.
 
Bear with me for a bit, but this is all important to understand:

A JFET compressor is what is known as a 'feedback limiter'. What this means is that the limiter's output level is examined, and if it's above some threshold level, a signal is generated that reduces the amplifier gain, so that, by the action of feedback, the output level is reduced so that it will generally be below threshold. If the signal exceeded threshold, then the gain reduction signal would turn its level down.

These types of limiters were developed because JFETs, vari-mu tube stages, and photo-resistor (opto) stages do not have a simple way to program their gain given a specific control voltage. So, you can't have a detector circuit that could come up with a voltage that could turn the gain down by some exact amount given an input signal that is such and such dB above threshold. However, by using a circuit that says "the output is too large, here's a voltage to reduce the gain", one does not need to use a gain control device that has a simple, linear relationship between the gain control voltage and the amplifier gain. The basic idea is that "if the output's too big, the input is turned down".

Compressors and limiters using the 'feedforward' topology are built with multipliers or VCAs whose gain can be controlled accurately with a control voltage. Using something like a Blackmer (dbx) VCA, one can examine the input signal, determine how loud it is, and then derive a control voltage that can precisely turn the signal down by the required number of dB.

The idea of "ratio" only makes sense for the feedforward devices. With those devices, it is possible for a dynamics device to provide a very specific input / output gain relationship above threshold. With any feedback dynamics processor, the ratio will always vary from just above 1:1 near threshold to inf:1 deep into threshold.

Yes, I know the 1176 has a ratio switch, but the concept is fraudulent. A true feedforward dynamics processor will give you that ratio everywhere above threshold, but the 1176 will always provide a ratio just above 1:1 just above threshold, and the 'ratio' will rise as the input signal increases.

All the 1176 ratio switch does is to reduce the gain of the output signal above threshold into the detector, which will reduce the 'ratio' when the device is performing a handful of dB of gain reduction. However, the ultimate ratio of these devices will still be inf:1, and right above threshold, it's basically a little more than 1:1 - it's always varying, never a fixed constant.

Another way to say this is that a feedback limiter will never have 'one ratio' - the ratio is always changing depending upon the amount of gain reduction. By contrast, a feedforward dynamics processor can provide exactly one ratio above threshold, regardless of how much gain reduction is happening.

Now, whether it's important to provide a mathematically pure, constant ratio is another thing - people use compressors and limiters to turn loud stuff down and turn quiet stuff up, so does it really matter what the ratio is at any one time? The ratio switches on an 1176 basically allow you to  do more or less gain reduction above threshold, so they sort of act like the ratio control of a feedforward limiter. However, they can never provide a pure ratio, because the JFET attenuator and the feedback detector prevent that. It will always have a ratio of just over 1:1 just above threshold, and then it will approach inf:1 deep into threshold. The ratio control essentially controls how much level above threshold you have to go to get to the inevitable inf:1 ratio.
 
LazyTurtle said:
Is there info on how this works exactly and is there any equation/function/method to easily configure ratio in a fet compressor?
In addition to Monte's words of wisdom  :)
FET's (and optos) have a hyperbolic control law (1/ax+b), which is not too dissimilar to the dBlinear control law of VCA's, except that, being part of a voltage divider, the actual gain cell control law asymptotes to 0dB instead of +infinity.
And the fact that the control voltage is derived from the actual linear amplitude of the signal has for direct consequence that the higher the signal amplitude, the higher the ratio. In a VCA+RMS compressor, a log version of the signal drives the gain cell, so the ratio (above threshold) is the same whatever the amplitude, since it depends on relative variations (a 10dB increase produces the same gain variation whether it's from -10to 0 or +10 to +20).
Usually, manufacturers quote their ratio figures as the slope they measure for a certain amount of GR, but since there is no standard, each one his own...
 
Monte McGuire said:
The ratio switches on an 1176 basically allow you to  do more or less gain reduction above threshold, so they sort of act like the ratio control of a feedforward limiter... The ratio control essentially controls how much level above threshold you have to go to get to the inevitable inf:1 ratio.
I've been following the various compressor threads currently active and this adds a real insight to feedforward vs feedback compressors

It probably explains why I always used the 1176 at 4:1 with the GR meter just kicking all the time. The compressor would be into gain reduction, and you'd hear the sound of it, but it wouldn't sound obnoxious on big peaks

Nick Froome
 
Hopefully not to add to the confusion, the ratio of any feedback compressor/limiter is determined solely by the characteristics of the gain reduction element itself i.e. the FET. The gain around the loop from the output of the FET through to the detector, only affects the threshold. Sometimes you can change the ratio by adjusting the dc bias on the FET so it spans a different section of its characteristic curve.

Cheers

Ian
 
Yes, applying small DC offset to the control voltage of a FET will set the control in a different Vgs interval being able to apply different gain reduction characteristic curves to the gain reduction stage making its behaviour more or less sensitive to input changes to the gate.

thank yo for your answer.
 
ruffrecords said:
Hopefully not to add to the confusion, the ratio of any feedback compressor/limiter is determined solely by the characteristics of the gain reduction element itself i.e. the FET. The gain around the loop from the output of the FET through to the detector, only affects the threshold. Sometimes you can change the ratio by adjusting the dc bias on the FET so it spans a different section of its characteristic curve.
I beg to differ. When you write the equations, you can see that both threshold and ratio are affected by the loop gain. A good example is the 1176, where the compression ratio control affects both the FET bias and the loop gain, with one bank changing the attenuation of signal reaching the side-chain amp and the other changing the reference bias of the signal rectifiers. Without this arrangement, changing only the bias or the loop gain would result in a questionable operation, where increasing ratio would decrease threshold and vice-versa. This is true for all types of feedback compressors, FET, Opto, Vari-MU...
 
abbey road d enfer said:
I beg to differ. When you write the equations, you can see that both threshold and ratio are affected by the loop gain. A good example is the 1176, where the compression ratio control affects both the FET bias and the loop gain, with one bank changing the attenuation of signal reaching the side-chain amp and the other changing the reference bias of the signal rectifiers. Without this arrangement, changing only the bias or the loop gain would result in a questionable operation, where increasing ratio would decrease threshold and vice-versa. This is true for all types of feedback compressors, FET, Opto, Vari-MU...

I beg to differ also. When you write the equations you see that any gain before the threshold affects only the threshold. The ratio is affected only by the law of the gain reduction device. In the 1176 the ratio is altered by changing FET bias but the threshold is also changed at the same time, probably so the 0,0 point remains the same.

Cheers

Ian
 
ruffrecords said:
I beg to differ also. When you write the equations you see that any gain before the threshold affects only the threshold.
Whether the loop gain is before or after the threshold detector, when the signal is above threshold, the ratio depends on the total loop gain.


The ratio is affected only by the law of the gain reduction device. In the 1176 the ratio is altered by changing FET bias but the threshold is also changed at the same time, probably so the 0,0 point remains the same.
Exactly. If only one factor was changed the 0.0 point would move, i.e. changing the ratio would change the threshold and vice-versa.
 
abbey road d enfer said:
Whether the loop gain is before or after the threshold detector, when the signal is above threshold, the ratio depends on the total loop gain.

You would think that but that is not what happens. I had a long discussion with etheory about this and built a spreadsheet to show what happens. Below the threshold, any gain before the threshold just alters the threshold. Gain before the variable gain element, gain after it or gain in the side chain before the threshold - they all just alter the threshold.

Once the threshold is exceeded it's a bit like a servo - more loop gain reduces the error but the function of input vs output is entirely that of the gain control element. I will see if I can locate the spreadsheet that illustrates this.
Exactly. If only one factor was changed the 0.0 point would move, i.e. changing the ratio would change the threshold and vice-versa.
If the threshold was not changed, at higher ratios the output might never reach 0,0. The threshold has to be raised so it does.

Cheers

Ian
 

Latest posts

Back
Top