"Compression Quotient" of AIAB / Distortion pedal

[Vivek]

Many guitarists praise AIAB / Distortion pedals by using words like "Responsive", "Dynamic" and "Not Compressed"

Upon thinking about this, I decided to look at some of the LTSPICE models I downloaded / input and tried to calculate the ratio of

(Output signal peak for 600mv peak input signal) /(Output signal peak for 60mv peak input signal)

Is that a valid / useful way to quantify dynamism  / touch sensitivity ?

[Steben]

All is subjective.
All overdrive / saturation / clipping distortion circuitry is compressive because it takes a part of the signal away.
"Responsive" and "dynamic" may refer to the change of sonic character according to the attack / amplitude with the pure amplitude compressive effect still present.
"Not compressed" is impossible with clipping unless it means "less compressed".

[Vivek]

Yes

Still, I know that a distortion box with compression Quotient 5 will behave different than one with compression Quotient 3

[Steben]

What about the difference between very gradual overdrive response eg ac30 amp and a harsh opamp clipper? The voltages you choose will be crucial.

[PRR]

> Output signal peak for....

Your ear is almost deaf to peak amplitude. It is more impressed by overtone spread.

[Rob Strand]

The way it's done (for compressors) is as follows,

    Compression Ratio CR   =   Input Change in dB  / Output Change in dB

You can compute "Input Change in dB"  as 20 * log10(V2/V1).  However you can also compute in dBV subtract the result,

     dB1 = 20*log10(V1), dB2 = 20*log10(V2),   dB change = dB2 - dB1

or relative to the threshold or unity gain voltage,

     dB1 = 20*log10(V1/Vref), dB2 = 20*log10(V2/Vref),   dB change = dB2 - dB1

Vref cancels in the dB change subtraction.

By computing the change in dB you also factor out any gain in the system.     
For simple 2:1 compressors  using an NE570,

    Input          Output            Output
                       (no gain)        (with gain)
   0dB             0dB                 5dB
   20dB           10dB               15dB

Input dB change = 20dB
Output dB change = 10dB   for both cases of gain

CR = 20 / 10 = 2:1

Real compressors might have a soft knee or a limiting zone.    What that means is the compression ratio varies for different inputs.
The simplest case  is 1:1 below the threshold and n:1 above threshold.  However for a soft knee it's not a black and white transistion between below and above the threshold.   The CR is near 1 just below the threshold and gradually changes to the final n:1 compression ratio.      Over the knee the CR varies from 1:1 to n:1.

In order to handle this case the "change"  is taken over a small spacing of input signal.

                   dB1  = 20*log10(V)   and dB2 = 20*log10(V + small value)

The result is a CR *curve* as a function of Vin.    It is not the dBout vs dBin curve.
A CR curve like this is often called the incremental compression ratio (and can be defined more formally with a derivative).

To get broad representation of CR over some span of input it is desirable to have one CR number which kind if incorporates some local behaviour.      In this case we simply pick two input levels sufficiently far apart to capture the broad local behaviour.     This more along the lines of what you want.

The risk using this method blindly is you can fall into the trap of not comparing apples to apples.      If one compressor has a threshold at 100mV and another at 1V.    Then using test voltages of say V1=1V and V2=2V is going to give misleading results as the test voltages are way over the threshold for compressor 1  and just hitting it for compressor 2.    In some case it makes more sense to use the output levels as the common reference point.  In others it is completely bogus.   In fact if you have two compressors with different softness, different gains and different thresholds.    How to tweak the input  and output scaling to compare two compressors isn't so clear.    You have to decide how you will pin down points for comparison.   For apples to apples comparisons it starts to make sense if you choose say a constant dB change of the input.

Your ear is almost deaf to peak amplitude. It is more impressed by overtone spread.

I was assuming he will be using sine waves so the difference between rms and peak is largely a scaling factor.    If two compressors are compared the peak vs rms measure could make a difference if we consider different frequencies.   Not all compressors behave the same at 100Hz vs 1kHz.

[PRR]

.....I was assuming he will be using sine waves so the difference between rms and peak is largely a scaling factor.    If two compressors are compared the peak vs rms measure could make a difference if ....

He said " AIAB / Distortion pedals".

Amiga in a box? Italian Organic growers??

IAC (in any case), considering where and who the question is, I don't think this is clean compression.

[Rob Strand]

He said " AIAB / Distortion pedals".

Amiga in a box? Italian Organic growers??

IAC (in any case), considering where and who the question is, I don't think this is clean compression.

In all honesty I've got no idea.   I hardly use acronyms.

I get what he is trying do because I've tried to compare compressors in the past.

[willienillie]

"Amp in a box"

Also, not an acronym unless read as a word, ie RADAR (radio detection and ranging).

[Steben]

I think Rob explained thoroughly the problem. A number can mean different origins which means it can not point out a character of a pedal.

[Rob Strand]

Here's a test set which should give the same amount of (non-linear) compression.
The only difference is the input and output scaling.

[Click to Enlarge]

[Vivek]

I am basically a Chemical Engineer (with minor in Electronics)

Too many things that we need to study have no real deep understanding at molecular level. Nobody really knows what is going on. For example, Nobody can predict flow of a particular molecule of water when it flows in a turbulent manner.

So we make some assumptions

and we arbitrarily set some measurement standards

Then we all use the same measurement standards to compare things.

And thereby, we have some level of insight and some way of comparison of a complex phenomenon


Hence I arbitrarily proposed low level input sine wave signal of 1KHz at 60mv. It could have been anything at all. But I suggested this as possible start of clipping

and other upper input limit at 600mv. Again, it could have been anything at all. But I assumed and proposed this value as possible maximum distorted signal at output.

If we use these arbitrary points to get some insight into a complex phenomenon, we can say something like this

Pedal A has 600/60 of 3 meaning that output with 600mv input has peak that is 3 times larger than peak with 60mv input)
while Pedal B has 600/60 of 5

That will never completely explain a complex phenomenon.

But it might begin to give us an insight, a method to measure and compare


I only made an initial suggestion to get the ball rolling,
I request the gurus of this forum to suggest better way to compare the "openness" and "compression" of distortion pedals.

[Steben]

I am basically a Chemical Engineer (with minor in Electronics)

Too many things that we need to study have no real deep understanding at molecular level. Nobody really knows what is going on. For example, Nobody can predict flow of a particular molecule of water when it flows in a turbulent manner.

So we make some assumptions

and we arbitrarily set some measurement standards

Then we all use the same measurement standards to compare things.

And thereby, we have some level of insight and some way of comparison of a complex phenomenon


Hence I arbitrarily proposed low level input sine wave signal of 1KHz at 60mv. It could have been anything at all. But I suggested this as possible start of clipping

and other upper input limit at 600mv. Again, it could have been anything at all. But I assumed and proposed this value as possible maximum distorted signal at output.

If we use these arbitrary points to get some insight into a complex phenomenon, we can say something like this

Pedal A has 600/60 of 3 meaning that output with 600mv input has peak that is 3 times larger than peak with 60mv input)
while Pedal B has 600/60 of 5

That will never completely explain a complex phenomenon.

But it might begin to give us an insight, a method to measure and compare


I only made an initial suggestion to get the ball rolling,
I request the gurus of this forum to suggest better way to compare the "openness" and "compression" of distortion pedals.

Just think of non inverting TS like overdrive.
If you put the gain on zero, the gain is actually x11 which becomes x1 when the diode start to conduct.
With the gain at max it s 116, which still becoms x1 when the diode start to conduct.
This means the gain setting alone already changes the compression ratio.

[Vivek]

Here's a test set which should give the same amount of (non-linear) compression.
The only difference is the input and output scaling.

[Click to Enlarge]

First Circuit on left:

Peak output signal for 600mv Sine in = 581.26mV
Peak Output signal for 60mv Sine in = 510.16mC

Ratio = 1.14


For circuit in the middle:

Peak output signal for 600mv Sine in = 1.162V
Peak Output signal for 60mv Sine in = 1.0199V

Ratio = 1.14


For circuit on the right:

Peak output signal for 600mv Sine in = 599.8mv
Peak Output signal for 60mv Sine in = 534.6mv

Ratio = 1.122


For circuit on the right but with 5K compliance resistors added on each diode limb:

Peak output signal for 600mv Sine in = 1.172V
Peak Output signal for 60mv Sine in = 580.7mv

Ratio = 2.02


It's obvious that compliance resistors make the clipping more soft, leading to more dynamic range.


Pete Cornish Crunch CC1 at Max gain

Peak output signal for 600mv Sine in = 0.887V
Peak Output signal for 60mv Sine in = 0.232V

Ratio = 3.82 == seems to indicate mild distortion, high dynamic range pedal


MRX Distortion III at max gain

Peak output signal for 600mv Sine in = 444mv
Peak Output signal for 60mv Sine in = 444mV

Ratio = 1.00 == seems to indicate High gain, low dynamic range pedal



The concept of a ratio of outputs for 2 fixed inputs begins to tell us about the dynamic range and compression of that pedal.

[Vivek]

Just think of non inverting TS like overdrive.
If you put the gain on zero, the gain is actually x11 which becomes x1 when the diode start to conduct.
With the gain at max it s 116, which still becoms x1 when the diode start to conduct.
This means the gain setting alone already changes the compression ratio.

We can measure at max gain setting for each pedal, to standardise readings

If by "Gain setting alone changes the compression ratio" you mean that there is no other factor, I would say that design of the pedal will also decide on the compression quotient of that pedal.

Are there compliance resistors ?

Are signal levels driving the diodes deep into clipping?

[iainpunk]

the volume of a sound is determined by the rms value. you can have a signal that looks like this:



it has 6-ish volt peak to peak, but it sounds really thin and soft. while this :



sounds the same volume, while only being 1 volt peak to peak

i don't think you should measure peak to peak voltages to determine volume (i believe LTspice has RMS value measurements as well)

cheers, Iain

[Vivek]

That makes sense, iainpunk !!!

Thanks


On graph, Hold CTRL and click on name of trace, a little window pops up with the RMS value of that trace

[iainpunk]

you are welcome

cheers

[amz-fx]

The simplest case  is 1:1 below the threshold and n:1 above threshold. 

This is a peak compressor, and it is the way that a lot of software compressors work. Not all hardware compressors are designed that way. The NE570, for example, does not have this type of transfer.  Above the threshold, the gain is reduced and below the threshold it is increased. This has the effect of reducing the dynamic range.

For simple 2:1 compressors  using an NE570:
Input          Output         
  0dB             0dB       
+20dB           +10dB     
-20dB           -10dB   

This is more of the standard compressor design that is squashing the dynamics into a smaller range.

OTOH, the Dynacomp is a peak limiter, not a peak compressor. Below the threshold it is linear and above the threshold it is a high ratio limiter. The Sensitivity control changes the threshold point where the limiting kicks in. The ratio is not adjustable.

Best regards, Jack

[Rob Strand]

The concept of a ratio of outputs for 2 fixed inputs begins to tell us about the dynamic range and compression of that pedal.

There's no doubt your method would produce a reasonable measure of compression.

If you think about it, it's not that much different to the more common dB method I posted.   Your input voltage ratio is fixed so the dB change for the input is fixed.   Your measure is a ratio that too is essentially the same as the output dB. ie. 20*log10(V2)-20*log10(V1) = 20*log10(V2/V1).   A small advantage of the dB version is the numbers aren't so close together for example for your measure a change from 1.122 to 1.14 increases the measure by 1.6% and the dB version produces a 14% change in the measure.

For circuit on the right:
Ratio = 1.122

While the non-linearity of that circuit is the same as the other, you could argue that *in use* the guitar input level is fixed so it will drive the diodes higher and compress more.  So the low value is an indicator of what will happen.

It's obvious that compliance resistors make the clipping more soft, leading to more dynamic range.

FWIW,  the addition of diode series resistors is largely the same as blending the clean signal (or an EQ'd version of it) back with the diode signal.

As an extreme case imagine blending the output of a a high gain pedal and a clean signal.  The high gain pedal will have a high compression and the clean will hand none compression measure 1.    If we dial the mix back toward clean the compression will remain largely 1.  However the sound will be the clean signal with an objectionable and unnatural fizz in the background.

In small doses, like the Cornish CC1,  the clean and distorted components become one and it sounds quite natural.