Schematic 'new' Roger Mayer RM58

[ruffrecords]

The control voltage on a sustained note is almost DC. So the voltage that reaches the gate evolves from 50% right at the beginning to 71% after 1TC (100ms) to 90% after 4TC (400ms). Since the control voltage law is the typical 1/x, that means the gain reduction evolves in a ratio of 2 over the first 100ms. That may not be what the user expects, particularly if he chose a fast attack setting.

Agreed but I thought your mod was trying improve transient response i.e to not detract from the set attack time. If the attack time is significantly less than 100mS then the cap does not have chance to charge.

Yes there is; it's not a typical RC delay, it's a R/R+C delay.
The principle of operation is to superimpose on the control voltage an AC component that is exactly half the voltage across the FET (from source to drain); this indeed for reducing 2nd-order distortion. Picking up the compensation signal from the output of the opamp indeed makes sure there is no direct injection of the control signal into the FET's channel.

Agreed. Picking up the compensation from a very low source impedance like the op amp output is a definite improvement and it means you can use a relatively large capacitor so the time constant with 1Meg is seconds. This is exactly how it is done in the 1176.

Cheers

Ian

 

[JohnRoberts]

Isn't AC voltage at the gate in your mod equal in level as d-s voltage?

There are some lines going off the page so I am unsure of the entire schematic but yes the design target is -6dB from DS AC voltage. Perhaps R3 could be 750k, but who wants a clean RM gate?  :


JR

 

[abbey road d enfer]

Isn't AC voltage at the gate in your mod equal in level as d-s voltage?

That's correct. My mistake. The ratio of resistors should be halved.
Thanks for mentioning. 

 

[Bo Deadly]

Just for kicks I ran the mod through the puter.

Looks like something a little funky is going on. If I hold the CV at -6dB and run a square wave through it, it's actually adding some delay that is causing transients to get through that do not in the unmodified circuit.

Change attached .txt to .asc for opening with LTSpice.

 

[Fuzz Face]

There are some lines going off the page so I am unsure of the entire schematic but yes the design target is -6dB from DS AC voltage. Perhaps R3 could be 750k, but who wants a clean RM gate?  :


JR

I attached the full schematic in question on page 1. It is based RuudNL’s modernized version of the RM57 compressor.


Squarewave - Thanks for doing that LTSpice sim. I’ve just been playing with Eagle for a few months, I should learn LTSpice

 

[Fuzz Face]

RuudNL - I apologize for hijacking your thread

 

[abbey road d enfer]

Just for kicks I ran the mod through the puter.

Looks like something a little funky is going on. If I hold the CV at -6dB and run a square wave through it, it's actually adding some delay that is causing transients to get through that do not in the unmodified circuit.

Change attached .txt to .asc for opening with LTSpice.

Your test signal is unrealistic; 1ns rise/fall time is bound to create problems in audio circuits.
Indeed, the lag due to the finite response of the opamp creates issues on very fast transients. This can be compensated by adding a small cap (1.1pF) across R2 (470k). You must change Trise and Tfall from 1ns (300MHz BW) to 1us (300kHz).

 

[RuudNL]

By the way: 24 replies later, I still have no idea how the gain reduction meter was implemented in the "new" RM58...
(And that was my question...)

 

[abbey road d enfer]

Agreed but I thought your mod was trying improve transient response i.e to not detract from the set attack time. If the attack time is significantly less than 100mS then the cap does not have chance to charge.

If the signal sustains after the attack, the cap will charge and GR will increase. This has almost nothing to do with the attack time of the rectifier, which is typically much smaller than the 1Meg/100nF TC.

Picking up the compensation from a very low source impedance like the op amp output is a definite improvement and it means you can use a relatively large capacitor so the time constant with 1Meg is seconds. This is exactly how it is done in the 1176.

There are so many variants of the 1176, I don't know. Which capacitor?