JR's Variable Time Constant Integrator

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
Post Reply
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

JR's Variable Time Constant Integrator

Post by mediatechnology »

John posted this here some time ago and I couldn't find what post it was in.
Going through my drive I found it filed in a place I wouldn't normally be looking for it in.
Having stumbled upon it I thought I better post it.
Now if I can just remember what I was looking for when I found this...

It's a variable TC integrator. I should build one.

Image
Conceptual variable time constant integrator by John Roberts.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: JR's Variable Time Constant Integrator

Post by JR. »

IIRC this was posted in connection with the THAT RMS chip as a way to vary the apparent cap size. So this would be like a speed control making a THAT based comp faster or slower with one knob.

I didn't melt any solder on this so it does need to be vetted...

In hindsight there may be some way to use steering diodes and more parts to generate different attack and release characteristics, but not this early in the morning.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: JR's Variable Time Constant Integrator

Post by mediatechnology »

What does Rx do?
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: JR's Variable Time Constant Integrator

Post by JR. »

mediatechnology wrote:What does Rx do?
So this is like a test ... to see how good my memory is? Not that good.. :o

From inspection Rx is doing nothing when pot is set for fast..

So the AC current flowing in the first .1UF creates a unity inverted voltage across the .1uF feedback cap. at the output of the first opamp... so ideally an inverted version of the voltage at the top of the first cap. With second opamp + input grounded it is an unity gain inverter so voltage on bottom and top of the output side .1uF should be the same so no load on the input cap= fast.. only .1uF.

When the pot is turned toward the slow end, the second opamp gets non-inverting voltage gain, so instead of the output cap looking like no load on the primary or input cap, it is swinging in opposition to the input cap trying to charge up with the voltage gain multiplying the impact of the opposition cap.

CAVEAT.. I have never built this so so there may be stability issues, and Rx is an X because I have no idea how much gain it needs to be useful. With Rx=open circuit at slow end of pot, it is unity gain non inverting so input cap looks like 3x (ballpark guess). With Rx 10k or so, .1uF could be effectively 12-15x .

Or it may not work... This need to get tried out to make sure it doesn't saturate in normal use.etc. I think I saw that THAT was doeing EOL on one of their RMS engines. Hope they still have stuff to use this on... :lol:

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: JR's Variable Time Constant Integrator

Post by mediatechnology »

Thanks John.

OK so I have a clue about what Rx may do. It's worth stuffing some parts on a protoboard to see what happens.

The 2252 is going EOL but the 4316 RMS detector is very easy to apply as a replacement. It's low enough cost the 4316's VCA could simply not be used and the 218X family or 2162 used as the higher-performance VCA half. Since they're both current in 6.1 mV/dB out and have a timing cap they're essentially the same with the exception of the +2.5V bias of the single-supply 4316. I'm not sure the RMS detector has the same dynamic range but in most apps I don't think it would matter. I think the rest of the 43XX family could also benefit from this if it works.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: JR's Variable Time Constant Integrator

Post by JR. »

mediatechnology wrote:Thanks John.

OK so I have a clue about what Rx may do. It's worth stuffing some parts on a protoboard to see what happens.
The things that need to be determined is scale.. range of how small to how large of a cap is desirable. The circuit may be more stable with another .1uF added from the top of the first cap to ground just to always be there. Rx is like a scale factor for how large the effective cap gets.
The 2252 is going EOL but the 4316 RMS detector is very easy to apply as a replacement. It's low enough cost the 4316's VCA could simply not be used and the 218X family or 2162 used as the higher-performance VCA half. Since they're both current in 6.1 mV/dB out and have a timing cap they're essentially the same with the exception of the +2.5V bias of the single-supply 4316. I'm not sure the RMS detector has the same dynamic range but in most apps I don't think it would matter. I think the rest of the 43XX family could also benefit from this if it works.
It doesn't look as hairy as some things I come up with. That old Tape NR (522- schematic around here somewhere) shows how far this approach can be taken, with more dynamics processing tricks.

enjoy

JR.
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: JR's Variable Time Constant Integrator

Post by mediatechnology »

Doug sent me this link some time ago: http://burnit.co.uk/sdiy/images/boxotri ... ressor.gif

This was an interesting circuit fragment for a variable slew limiter.
Seems like it fits here in this thread.
Note that U5A is connected as a comparator and that C3 is referenced to -15V so that it can remain polarized.

Image
Variable Slew Limiter Circuit from http://burnit.co.uk/sdiy/images/boxotri ... ressor.gif
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: JR's Variable Time Constant Integrator

Post by JR. »

Perhaps while a different animal.

The THAT RMS detector is a rectified current feeding a cap (similar to the old NE570 series NR). My cap multiplier, (if it works) scales both attack and release similarly.

The slew rate circuit if made slow enough in one direction of the other could give some control of attack independently from release.

I was thinking some similar diode steering inside my circuit to alter the up multiplication factor vs the down multiply might provide some att/release differential, but that is putting the cart ahead of the horse since we don't even know if my approach works without adding extra tricks..

Since the THAT detector and side chain is operating inside the log domain a linear slew rate control, would provide different dB/sec rate of change, so perhaps not crazy. The pots do mimic attack and release controls. Most side chains are exponential RC characteristic but linear dB/sec could be OK. I kind of like the way exponential ATT/REL (that mine does not have either) slows down for very small changes. Perhaps adding a dead band could reduce distortion from small level changes (and LF ripple).

Have fun... I'm wrestling with microprocessor code today. I may post in the Peabody lounge about things I tried that didnn't work...Interesting to me at least.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: JR's Variable Time Constant Integrator

Post by mediatechnology »

I was thinking some similar diode steering inside my circuit to alter the up multiplication factor vs the down multiply might provide some att/release differential
That was the reason I posted it.

Don't know if you followed the original link but this circuit was fed by the RMS section in a 4301 having a switch-selectable timing cap (or NLC.)
It's more similar to the variable attack-release published by THAT in DN-115 http://www.thatcorp.com/datashts/dn115.pdf which doesn't work that well as published.

By the time it reaches the slew limiter stage it's a voltage and indeed a different animal than the "variable" NLC of your drawing.
Yes, I need to try it.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: JR's Variable Time Constant Integrator

Post by JR. »

As I mentioned I took this concept to exhaustion in my old tape NR based on a NE572 where I had all kinds of fast attack, slow release, and dynamic smoothing, not to mention other games to help reduce mis-tracking for tape recorder use. I actually had the fast attack borrow current from a long term average cap for best long term tracking accuracy.

viewtopic.php?f=6&t=252&hilit=522

Hopefully I described how the 522 works in that other thread because there are many moving parts and more than needed for a simple comp, not to mention some thresholds that aren't obvious (like rail voltage for saturated opamp outputs, etc). .

JR
Cancel the "cancel culture", do not support mob hatred.
Post Reply