Track/hold on an envelope follower

[EBK]

I have a bunch of things already in the prototype queue, so I figured I'd throw this idea out here, so someone can either tell me it has already been done, or someone else can develop the idea before I do and let everyone benefit sooner.

Anyway, here is what I envision:
Two stomp switches, one being ordinary bypass switch, and the other being a momentary type. 
When the effect is turned on, and the momentary switch is held down, the envelope-controlled effect operates as usual.
When you lift your foot from the momentary switch, it maintains the present envelope-controlled voltage, and the controlled whatever (filter, etc.) stops changing until the momentary switch is again pressed. Thoughts?

Hopefully, I've explained that well enough...


I suppose it could be reversed, with the momentary switch effecting the hold mode rather than the track (perhaps more useful).

[GibsonGM]

Going to follow this with some interest...might need a uC, you think?   Even an arduino...

To continuously sample the envelope voltage, then provide that voltage when the switch is engaged...

[EBK]

Nah, just take away the discharge path (well, charge path too) from the envelope cap, I think. 
If the cap voltage is fed to a JFET-input op amp (like TL072), all is well.

[TejfolvonDanone]

Nah, just take away the discharge path (well, charge path too) from the envelope cap, I think. 
If the cap voltage is fed to a JFET-input op amp (like TL072), all is well.

Looks like that's it to me.
EDIT: Maybe a rail-to-rail op-amp would work better if the envelope's output goes to any of the rails.

When you lift your foot from the momentary switch, it maintains the present envelope-controlled voltage, and the controlled whatever (filter, etc.) stops changing until the momentary switch is again pressed. Thoughts?

For me it would make more sense if you'd have to press the button to stop the envelope follower. Maybe you could use transistors as the switches and add another switch to control which mode you use.

[anotherjim]

Sample & hold.
The envelope capacitor can be isolated by switching off it's charge/discharge paths while the cap voltage remains available via a high impedance buffer.
Think 4016/4066 can do the switching well enough.
With an electronic latch circuit, switching gets good debouncing and the footswitch could be tap operated instead of held.

[GibsonGM]

Nah, just take away the discharge path (well, charge path too) from the envelope cap, I think. 
If the cap voltage is fed to a JFET-input op amp (like TL072), all is well.

Nice.  As long as the impedance is very high, it should hold, I see...

[slacker]

You might need to think about capacitor types, sample and holds normally use low leakage film caps whereas envelope followers tend to use electrolytic caps to get large enough cap values. Using an electro you might find the held voltage droops unacceptably as the cap self discharges, I guess it depends how long you want to hold it for. If that's a problem you might need to redesign the envelope follower to use film caps tweaking the rest of it to account for the smaller cap values.
Jim mentioned 4016/4066, I've buit S&H using these in the past and they work great.

[GibsonGM]

That's kind of what I was thinking, that leakage would limit the time you can hold.   You may be holding your 'hold' button, but it is decaying...how fast remains to be seen.    But I'm no expert on envelope followers!  So maybe not.   

I'm waiting to see the 'experiment' and how long a time is obtained...and if it can be kept simple.

[anotherjim]

An add-on approach would be to buffer an existing follower cap (probably already is followed by a suitable buffer), then insert the S&H before whatever is controlled.
Only needs a single switch to disconnect input. RC can be small value & fast. We want the C for the track & hold; not a time delay. C can be mylar/polystyrene or whatever your favourite low-leakage type is. Real world use, your usual film caps might do perfectly well.
R value might need to be largish if you don't want a sudden jump when a hold is cancelled when the actual envelope voltage is very different to what was held.
Buffer S&H cap by some mondo high Z amp. I like TLC272 cmos amps for similar "to ground" swing of the good old LM358.
Would be workable with direct switching via a momentary footswitch, although danger of leakage is increased by offboard wiring.

[StephenGiles]

Didn't Aphex use this track and hold idea?

5.4.1 Silence Gate *
This is a gate which affects only the gain control system. The audio is not being gated. When the input signal falls below the user setable threshold, the leveler gain will be locked at its present value until sound returns at a level above threshold. The background noise will therefore be prevented from swelling up during pauses and silence. The silence gate threshold can be set from 10dB above to 38dB below the input reference level. A 1 second delay is applied to the action of the silence gate to prevent it from interfering with the action of the DVG. The silence gate can be defeated when not needed.

5.4.2 Dynamic Verification Gate DVG *
Like the silence gate, this affects only the gain control system, and audio is not being gated. During program material, the DVG inspects the instantaneous peaks and compares their amplitude with the recent historical average peak value. Only during the brief times when the instantaneous peak exceeds the peak average is the leveler gain allowed to move. In this manner, the slewing
of leveling gain is converted to a series of small adjustments which take place at times in the audio envelope when the corrective gain changes will be less noticeable to the ear. The DVG naturally serves as a short term silence gate because brief pauses in program cause it to gate the leveling. However, after about 1 1/2 seconds of silence, the DVG releases control over gating. By that time the silence gate will have detected silence and the leveler gain will be locked by the silence gate. The DVG can be switched on and off as desired.

[EBK]

Definitely some sort of track and hold there, but that's really complicated sounding (very interesting stuff though, after a second read -- it's given me some ideas for future projects).

I'm thinking essentially of something like this:

But, replace the op amp on the left with a precision full wave rectifier, and probably throw in a series resistor.  Of course, I'm also using it for track and hold instead of sample and hold, but the ideas are similar (continuous control voltage instead of sampling pulses).

[anotherjim]

I think JFET switch would need balanced supply so the gate can go negative of the source - or the C volts will need to be constrained to some positive minimum with single supply.
Relying on the cap to smooth the rectifier means a big cap with some R in series, unless rectifier is one with an integrator included.