Passive Compressor

[joaquins]

Hey, I found this scheme in the web, just want to hear some opinions...

I've tried it with the guitar amp in a protoboard, pick up a lot of noise (hi-z and no shielding) but it do compress and with variable attack(R2), release (P1) and ratio(R5/R6/R3)... maybe could be controled the symmetry (R5/R6) (to look for asymmetrical compression)

Of course it isn't the choice for who is looking for a transparent gain contol, but for a colourfull compressor may work. It would probably need some optimization and in/out buffer (no more passive) but it's at least courious.

JS

 

[alexc]

Here's a schematic of an Elektor compression pedal which appears related  ...
Would be fun to try it out.

Remind's one of several other limiters out there  ;D

 

[JohnRoberts]

About the only similarity is both use diodes.

Neither is very clean so more of an effect than useful limiter (IMO), but nothing wrong with seeing if you like the sound.

The first one appears designed to feed a very high impedance guitar amp input, so that output needs to be buffered for line level use.

JR

PS: Back a few decades ago I looked at a variant on using diode impedance with changing current, as a gain element but dismissed it as too noisy and nonlinear, for professional audio use. 

 

[PRR]

The first one is 0.050V 470,000 ohm output for 0.5V _66_ohm input.... hardly "passive" because you will need gobs of amplification to make-up the massive loss of power.

The second is a straight-forward diode-bridge with appropriate amplifiers.

> Neither is very clean .... I looked at a variant on using diode impedance with changing current, as a gain element but dismissed it as too noisy and nonlinear.....

The classic CBS radio broadcast dynamic-range controllers were very much like the second one, except discrete transistors and lots of transformers instead of opamp inverter and diff-amp. They were very good in their days (1970s), not just AM quality but also widely used for FM, even "good" FM. Yes, the dynamic range does not run a lot more than 70dB (output), but well-hit that's really very good. It was very hard to make them sound "bad" even with massive input level and deep gain-reduction.

 

[alexc]

About the only similarity is both use diodes.

My bad - in the original post, I saw what appears to be a a circuit fragment using a diode bridge for the purpose of level limiting with the possibility of adjusting time constants, for use as an effect, with buffering as a likely requirement.

I thought that essentially described a diode ring compressor limiter pedal  ;D

The operating principle obviously must differ in the first example, having additional resistors in 2 of the diode  ring arms as well as what appears to be balance-able dc blocking/clamping with caps at the top and bottom nodes.

As well as some feedback from output to input and ground referencing on one input.

How does the circuit in original post work?

 

[JohnRoberts]

I was looking for full range VCAs so my opinion diode impedance was not in the same ballpark with even the early VCAs. FWIW I have used lower performance gain cells in limiter applications where they are only active while reducing the gain of a hot signal so noise floor and linearity is not much of a concern... OTOH for a console VCA channel feeding unity gain the noise floor needed to be much better, even back then...

=======
How does the circuit in original post work

From a quick glance... at low voltage levels, the diodes are all cut off and the 470K R is the full-on path from input to output, thus the need for a high impedance input circuit following this.

For larger signal voltages, D1 and D2 conduct and charge the two capacitors to + and - DC voltages, roughly AC peak less a diode drop each way. These capacitors when charged high enough start conducting through the two 22k resistors and diodes D3 and D4. When the input voltage is high enough to bias up and turn on D3 and D4, the 470k R will be shunted by 11k to ground (the two 22k resistors in parallel, as a simple pad/divider

The 100k pot is like a release time adjustment, the faster it discharges the caps  the quicker the limiting releases.

This is pretty high threshold since it needs +/- a couple diode drops to start working,,,  clean as a whistle when fully released, and could be relatively linear when hard limiting,,,I can't vouch for the transition region on the edge of limiting.

Hopefully there is some useful impedance range between open circuit and 11k provided by the diodes as they turn slightly on...

JR

 

[alexc]

Thanks JR - I can follow it better now. Wasn't quite clear on a few things.

Cheers

 

[emrr]

The classic CBS radio broadcast dynamic-range controllers were very much like the second one, except discrete transistors and lots of transformers instead of opamp inverter and diff-amp. They were very good in their days (1970s), not just AM quality but also widely used for FM, even "good" FM. Yes, the dynamic range does not run a lot more than 70dB (output), but well-hit that's really very good. It was very hard to make them sound "bad" even with massive input level and deep gain-reduction.

A 1RU CBS Volumax still sounds real fun in a rock recording these days too.  Kind of a one trick pony, but pretty hard to fault on sonic effect.  I cut a few demos for myself using nothing but a Volumax hit hard, and it sounds good.   

 

[PRR]

I see JR beat me..... see if we agree on anything.

First: the circuit shown uses Germanium diodes. It will work with Silicon, perhaps more predictably, but at higher voltages.

I started with the ass-umption they were Silicon and I'll plow onward.

D1 C1 D2 C2 form a peak-to-peak rectifier. The P-P voltage is held across C1+C2, bled by P1+R4.

Signal won't pass through the "ring" because C1 C2 shunt it all to ground.

R3 passes signal to the output.

R5 D3 D4 R6, with C1 C2, form a variable impedance shunt to the signal coming through R3. When the C1+C2 voltage is zero, the shunt is near-infinite, no attenuation. When diode current is 0.1 microAmp, each diode acts like 300K. So the shunting is like (300K+22K)/2= 161K. This will give significant attenuation against R3 470K (about 12dB).

The input level for this point is around 2V-2.4V p-p for Silicon, 0.5V-1.0V for Germanium.

When diode current is 1 microAmp, each diode acts like 30K. Ah, but there's 22K in series with each. So the shunting is like (30K+22K)/2= 26K. And when diode current is 10 microAmp, each diode acts like 3K, the shunting is like (3K+22K)/2= 12K.

The plan shows 10V input, 28V p-p. This would cause almost 14V/22K or 500 microAmps in each diode, which should drop to 60 ohms, but the 22K series resistor has been limiting the limiting (hah) since about 1uA diode current. So it would seem to max-out at 32dB limiting (which is pretty good for the simplicity).

Yeah, it will work, and is probably pretty optimum for a thing that does so much without amplifiers. The input charging current is brutal, and the output impedance is awful high and variable; that's what comes of avoiding amplifiers.

I *suspect* this goes between a TV set speaker (good low impedance) and the hi-Z mike input of a PA amp, to knock-down loud commercials so listening is tolerable on a better speaker. So while "passive" it requires a loudspeaker-like source and a 4-tube amplifier behind it; but you may have such gear available and the limiter itself is simple and un-powered.

These days you can put a resistor and LED on the gitar amp output, shine on a LDR and resistor at the guitar input or some low level stage.

 

[PRR]

Oh.... the output of these 2-diode limiters can only be ~~20mV peak clean, 50mV peak bent. (CBS used 4 diodes for another 6dB higher.)

The signal voltage appears on the diodes, adding on one and subtracting on the other. 20mV will double diode current and half diode impedance. So say 1mA each, 30 ohms each, 15 ohms effective. Now unbalance 20mV, one diode drops to 15 ohms, the other rises to 60 ohms. There's some cancellation but not perfect, we have 12 ohms on 20mV peaks instead of 15 ohms at tiny-level. So what comes out (cleanly) is much weaker than typical guitar levels, and hiss is a real problem.

And John: at the time CBS had a seriously staffed lab, and the original patent on these devices is about an intercom for military tanks. We still feared Russia would roll tanks across Europe. More money than you ever shook a stick at. And as you say, a limiter is very different from a mixer VCA.

 

[Dan Kennedy]

Qoute:  These days you can put a resistor and LED on the gitar amp output, shine on a LDR and resistor at the guitar input or some low level stage

Back in the day you could put a flashlight bulb on the output of a 20W Eico and an cad cell on the input and get the same result.

KRAP AM in the flesh.

Did you know that 4 metal 6L6's could crank out a lot of AM? Especially upside down in Mom's wash tub.

Don't use the plumbing when on the air.

 

[opacheco]

.......When diode current is 1 microAmp, each diode acts like 30K. Ah, but there's 22K in series with each. So the shunting is like (30K+22K)/2= 26K. And when diode current is 10 microAmp, each diode acts like 3K, the shunting is like (3K+22K)/2= 12K.........This would cause almost 14V/22K or 500 microAmps in each diode, which should drop to 60 ohms, but the 22K series resistor has been limiting the limiting (hah) since about 1uA diode current. So it would seem to max-out at 32dB limiting......

How or where do you find the resistance values for these bias current ??.....Do you have a Vd versus Id graph ??; Please show us how do you did it.

Opacheco