Silver Freezer - inspired by Orange Squeezer et. al.

[earthtonesaudio]

UNVERIFIED (6/35/11) but I plan on prototyping tonight.  It's a compressor built around a JFET voltage-controlled resistance like the Orange Squeezer.  But it incorporates some ideas from Mark Hammer's Tangerine Peeler (sidechain with feedforward/feedback options), the Fast Audio Peak Limiter from here (full wave rectification and alternate drain/gate feedback strategy), and Free_Electron's suggested use of "SWTC technology" from this thread. Pot1 and Pot5 are probably slightly redundant, but I'm not sure yet.

The unique EarthTonesAudio contribution is simply the use of an LM317 for the JFET source bias.  I still have to verify if the output range of the '317 in this configuration will work with a few different types of JFET.

[arawn]

looks interesting

[garcho]

How has this panned out? It's hard to have too many compressors (unless you're a top-40 producer  )

[earthtonesaudio]

Oops... I don't think I ever breadboarded it!  At least I don't remember doing so.  Looking at it now after having slept a few times it seems a bit unnecessarily complicated.  The LM317 for biasing the JFET only makes sense if the Vref is also regulated, unless making the source voltage user-adjustable gives some useful feature.

I might revisit the main ideas here in the near future.  In the meantime, no one has verified this yet that I know of.  It looks like it should mostly work though.

[PRR]

> The LM317 for biasing the JFET only makes sense if the Vref is also regulated

No. The source of JFET should be at Vref+Vto+th. For stompbox precision, a sturdy voltage divider will do for supply from 8V to 12V.

Since there is zero DC load on the LM317, your "390" might have to be as low as 120 for a worst-case '317.

Also (since '317 can't pull-down) C2 ought to be as big as C4.

C8 R15 can possibly be omitted; the DC error will not be significant compared to 1N4148 threshold.

C10 C11 R17 seem to only add negligible CV ripple.... what am I missing?

No attack/release time-constant options?

EDIT.... the DC voltages are wrong somewhere. IC1C IC1D are biased to +4.5V. C3 will charge to +4V...+/4..4.4V idle, rising with signal. Therefore source of JFET must be above 4.5V just to get the gate to stop conduction, above 5V for the channel to turn-off. However POT1 is all at 4.5V. so at the very least C7 polarity is wrong. And with source up at 5V, and max swing of opamp about 1V-2V short of supply rail, there's not a lot of rectifier headroom.

[earthtonesaudio]

Paul, I think you meant to type C4... and you're right, the polarity is wrong!  Thanks.

What I meant by the "317/regulatedVref" comment was more like "the LM317 is overkill for this application".

C10/C11/R17 are plagiarized from this page: http://sound.westhost.com/project67.htm for JFET distortion cancellation.



I only see a problem with rectifier headroom in the specific case where POT3 is set mostly CCW and the rectifier gain is high or the input signal is large.  Then an input could be large enough to forward-bias the JFET gate, eek!
If POT3 is set CW this can not happen (except perhaps for transients) because the JFET will limit the signal available to the rectifier.

Anyway I'm not really happy with this design and I think it needs an overhaul.  4 knobs with no attack/release is a bit silly.