I mentioned this over at Ampage a while ago and it still bugs me how to do it. The ripple problems in envelope followers are well documented (by Mark Hammer for instance), so better perhaps would be an envelope generator, the start voltage of which when trigered by a guitar note, is proportional to the strength of the note, and its decay is independant of any control derived from the decaying guitar note - result is ripple free and a nice smooth sweep. Not easy methinks - but food for thought this year.

It sounds as though this would yield some very interesting results. Maybe we could borrow ideas from our synth friends?

http://www.uni-bonn.de/~uzs159/adsr.html

Harry Bissell's envelope follower design, posted at Electronic Design Notes, is fabulous.  When he and his wife were up from Detroit in the fall of '02, he brought his homebrew guitar synth that uses a hex pickup and these envelope followers, and let me tell you, they were THE most responsive envelope followers I think I've ever used with guitar.  A very true representation of the actual signal envelope, with realistic decay and very fast attack time, and inaudible ripple.  If you can't find the design on-line, I can send you either the file or a scan of what I printed out.  More complex than the usual FWR beast but definitely worth it.  Harry is picky about such things, and I'm glad he is!

http://www.reed-electronics.com/ednmag/index.asp?layout=article&articleid=CA265499&rid=0&rme=0&cfd=1

But an envelope generator does sound interesting.

The "ModFX philtre" comes with a envelope generator, the "trigger" mode.  I've played a lot with it, and it's not as responsive as when your effect is envelope driven by your effect signal... it seems as though there's less correspondance.  You can't consistantly get the same sounding effect out of the 2.

-Colin

It's encouraging to hear I'm not the only one who has thought of such a thing as an envelope generator.

I was thinking of using the OTA trick and a feedback control loop so that the guitar signal actually tracks with the generated envelope in RMS power.

This idea I had, uses an envelope follower on the output (so it doesn't do away with the ripple), but with a critically damped control loop, the ripple would be a minimum.  

For clarity, here's the idea:

Input:
buffer, envelope trigger (like a synth) with threshold control

Stage 2:

Variable gain stage adjusted by control output.

output of VGA (likely employing the use of an OTA) is tapped by an envelope follower.

Control:

An actual envelope generator is constructed and fed into a comparison amplifier setup.

the other comparison is the envelope follower signal.

The output of the comparison amplifier is the control voltage for the VGA.

If someone inputs a guitar signal, first a synth-like envelope is triggered, then the signal passes through the VGA and the envelope is detected and compared to the generated envelope.  

In controls terms, it's the error signal of the feedback control loop that sets the gain on the VGA about a bias point.

If this system is well designed, the guitar signal envelope will follow the synthesized envelope quite well and the result will be a more synth-like effect.

Simply, this is like having a compressor with a hard limiting region that is modulated by an envelope generator....

Does anyone have a layout for Harry B's solution with the 3 peak detect circuits?  Just wondering...

Thanks Mark for pointing out this circuit. I managed to pick up a couple of 4584 chips yesterday so I'll give it a go this weekend.

Stephen

You can try Harry himself (harrybissell@prodigy.net) and see if he has and is willing to part with a layout.  He had the boards professionally made (they're real and they're spectacular), but I have no idea how modular they are, or if it is in a form he is willing to share.  He may have a layout for a hex unit but not mono unit, or may have a layout for an entire voice channel (EF, VCF, VCA, etc.) from which the EF can't be extricated, etc.  I caught a quick glimpse under the hood, but I honestly couldn't tell you very much based on that.

I'm guessing that the output of the circuit is still AC, so under most circumstances needs rectification?