THAT corp.'s "Non-Linear Capacitor" for improved Envelope Detection ? (DN114)

[effexfreek]

Hi ! First topic ever...


Couldn't find anything on the subject, beside some mentions in jan. 2012 "The BEST bass compressor", so thought it'd be best to start anew :

   Has anybody toyed around with THAT corp.'s "Non-Linear Capacitor" circuit (from THAT Design Notes DN114 & some of their "new" pedal schems.) to improve their own Envelope Detection/Follower design ?

A 2012 quote from gritz :

"It's early days, but it does seem to be working like it should. Sorry for the hijack, but as has been mentioned, getting a control signal that's free of ripple but still responds quickly to transients is the Holy Grail of single band compressors."

Would love to hear back from him (as his schems. are down), if he's still around...


   I'm particularly curious hearing about using it with say, a NE570 comp./FWR, for example (as it could become an almost-nearly self-contained, side-chain processor, for whatever...). Anxious to try it out myself but, breadboarding has always been a crap shoot for me    wouldn't want to "throw away the baby with the bathwater", if someone can confirm being able to (I feel compelled to add "dramatically") improve an Envelope Detector/Follower with such a circuit, as I'm stumbling (like everyone has, I guess...) on "two-note interference" from "note onset detection" à-la Anderton's AMS-100, Stompboxology's Attack-o-Matic V, as well as Compressors, Envelope Filters & the likes... (sure : high sensitivity on single notes BUT multiple triggers on "clusters" of notes -VERSUS- almost tolerable chord triggering BUT low sensitivity on single notes)

I get it : "there's no free lunch"... or is there ?


Thanks in advance !

P.S. : I know about Harry Bissell's Envelope Detector... couldn't for-the-life-of-me make it work...  also saw oscilloscope shots from some other forum (MuffWiggler, methinks ?) that showed "mixed results" with it... probably just unlucky like me, I guess ? Oh, well...

[StephenGiles]

I could never get Harry Bissell's envelope detector to work either! 

[EBK]

Not an envelope detector, but take a look at pp. 9-11 of this app note:
https://www.onsemi.com/pub/Collateral/AND8159-D.PDF

If I recall correctly, it is the same sort of NLC in a couple NE570 circuits, so you may find it encouraging.
 
I've build THAT's Pedal Design 5 (again, not an envelope detector, so I don't know if this comment is helpful), and I love it, but I don't know for certain how much of my love for the circuit comes from the NLC.

[Digital Larry]

I've always like THAT Corp stuff, having incorporated one of their early VCAs into a commercial design in about 1990.

Here's DN 114 http://www.thatcorp.com/datashts/dn114.pdf
and related:
http://www.thatcorp.com/datashts/dn03.pdf (go to page 7 for extra credit on the non linear capacitor)

Even with my increasingly rusty analog skillz I more or less get it. Now I'm wondering how to adapt it to a DSP implementation.

Couple other comments.
THAT 2252 is obsolete, but they have some recommendations for other chips to use incorporating the function, including a full blown analog engine which incorporates the VCA as well. 

I would say that the concept applies perfectly to the NE570 as the "magic" is done externally with stock linear circuitry.   You might find yourself needing to adjust gains and ratios perhaps?

[Transmogrifox]

I have played with Harry Bissell's concept in software and had results that work out better than a simple FW rectifier w/ RMS detector or peak hold detector, but it was not straightforward -- I had a lot of weird stuff to work out.  In DSP land it's much easier to deal with weird cases, but the main things I needed to do:
1) Envelope detector pre-emphasis.  Bringing up the peak levels of the higher frequency strings reduces that "two-note interference" you're talking about.  No envelope detector without some kind of pre-emphasis is going to handle that well without using advanced DSP algorithms to analyze the signal and adapt over time.  Multi-band processing of the envelope could help here.
2) Post-filtering:  I added attack and release filters after the envelope detector set with minimum time constants that always lag the "stair-step" effect.  In other words, you tune the release time so you always hear an exponential decay, and not stair-stepping decay.
3) Cycle frequency should be as fast as you can tolerate it without noise.  #2 helps this situation.
4) Put a precision full-wave rectifier ahead of it so you are detecting off both polarities of the incoming signal. This helps #3.

Finally, the nonlinear capacitor could be applied to the Bissell style detector.

Also note you could run an RMS detector with nonlinear capacitor ahead of a Bissel circuit, and the Bissell circuit would be used to further reduce the detector ripple.

Don't totally cross the Bissell circuit off the list, but it does need some pre- and post- conditioning based on the story of how my DSP implementation went.
#2 was by far the most important part of getting the Bissel styl envelope detector to sound good.  #1 just deals with a problem common to all envelope detectors.

[effexfreek]

Hey, thanks Stephen, D. Larry, EBK & Transmorgrifox for the fast replies :  sooo much appreciated ! (I don't feel so alone anymore !)



   Yeah EBK, I'd copied those NE570 app. notes "hi-fi compressor" schems. from the net somewhere but, had no accompanying text.... I had an inkling the extra 2 op-amps were a NLC circuit of some sort, but wasn't real sure about it !

   O.k., I won't give up on Harry Bissell's just yet, Transmorgrifox...  it's just : the more involved the circuit on the breadboard, the more chances of bad contacts/pilot error/impatience-frustration... etc. God knows I've been there 


   On the breadboard as I write... keep them comments/suggestions/(dare I say : schems.?) coming ! Much obliged...


On a side-note : just recently got a MOOER Slow Engine for cheap... to open it up, of course (SG clone ?), only to be MIND-BLOWN : it's an ENTIRELY DIGITAL platform ! AND they couldn't improve upon the ol' SG ? (the note onset detection part, I mean) Well, I'll be.... !?

[effexfreek]

   Upon further reading https://www.onsemi.com/pub/Collateral/AND8159-D.PDF (bottom page 9 & page 10) from EBK's post, the tacked-on, dual opamp/capacitor circuit doesn't seem to be used as a NLC per se, I gather ? More like : "speeds up the compressor attack time at low signal levels. The response times of the simple expandor and compressor (Figures 1 and 2) become longer at low signal levels."

Am I reading this right ?

[Transmogrifox]

speeds up the compressor attack time at low signal levels. The response times of the simple expandor and compressor (Figures 1 and 2) become longer at low signal levels."

Am I reading this right ?

You are reading this right.  The NLC is affected by a rate of change of the incoming signal, the 8159 app note refers to a circuit that is affected by the signal level.  I don't see any reason both techniques couldn't be employed simultaneously to solve different problems, yielding a more intelligent sidechain system.

[effexfreek]

100% agree with that ! Just worrying they won't play nice with each other ? (how would you go about it ? Any suggestion would be wholly welcomed as I'm not much of a mathematical/theoretical/technical wizard... more of a "poky-poky here, poky-poky there, smells funny all-of-a-sudden ?" kinda-guy        )

[Transmogrifox]

It sounds like you understand the main differences between what each technique offers.  If you "get it" how to employ a circuit to do each one independently then I think you can do this like a real engineer:  "Poky-poky here, poky-poky there, smells funny all-of-a-sudden ?" kinda-way.  All the math stuff is just to convince your non-musical colleagues it's technically sound -- if you don't have to do that then you can just stop when it is right to your ears.

Your ears will tell you when you nailed it.  The main thing is to do each piece, one at a time and get a good feel for how each one helps, how each thing sounds.

Then when you combine, your ears will tell you if something is just plain wrong with the circuit since the tweaking with both together should not be totally different than tweaking them individually.

[effexfreek]

Al-righty, then !  If you think they can peacefully co-exist,  I'll give it a go !

Time to suck it up & TALLY-HO !


(Thanks for taking the time to write & sincere apologies, Transmogrifox : guess I  really needed to hear it from a more "knowledgeable tinkerer" than I (you engineer & program, I gather)... so far, my efforts amounted to *ziltch*... call it what you will : getting old, loosing my touch, trying to keep a brave face but growing tired of failing, and failing, and failing again... oh, well. Not having a great run, as of late. Sorry.)

[effexfreek]

a-HA !


I know what : I'll buy a coupl'a those THAT 4316 chips (I do think I have adapter boards laying 'round here... somewhere). Mouser has them at $4.55CAN ; well, not too bad (but the shipping, 'though : $20CAN ? A-Outch... I'll order more stuff to make it worthwhile).

   THAT way (geddit ?.. couldn't resist), I'll have a (hopefully working) Known Baseline to work with, and THEN, build on from there...

...and, if I still can't achieve an A-mA-Zing! Envelope Follower Unit, I'll just make some-o-those "One-Knob Squeezer" EBK's so fond of... see if I fall in love with'em !



I think THAT'll work out, in my case...

...doesn't sound like "throwing money at the problem", now... does it ? 


P.S. : EBK : did you use those (weird!?) 33178 opamps ?

[EBK]

P.S. : EBK : did you use those (weird!?) 33178 opamps ?

Nope.  Used non-weird TL074s.  There was some discussion involving the designer of the squeezer circuit (Wayne Kirkwood) about those weird op amps, but the ultimate conclusion was that they would only be needed (to prevent oscillation) if the VCA was a traditional Blackmer cell, which the 4316 apparently is not.  https://www.diystompboxes.com/smfforum/index.php?topic=53097.0 

[effexfreek]

Oh yeah... I remember stumbling upon that thread now, while researching Envelope Detection !


Cooool, great to hear about the TL074 : thanks-a-bunch !

[Transmogrifox]

I think THAT'll work out, in my case...

  Good one.  I'm a sucker for puns.

trying to keep a brave face but growing tired of failing, and failing, and failing again...

I'm familiar with this feeling.  Keeping the brave face gets you through.  With a brave face any small success revives your confidence.

At least you don't have managers constantly asking you "when" and marketing guys selling your vaporware...