I want to make an attack control but already have a high source z

[midwayfair]

I've got a transistor-based rectifier. It's the one in the Bearhug, but I don't have a JPG schematic, so here's the description (and here's a link to a PDF with the schematic: http://1776effects.com/wp-content/uploads/2015/03/Bear-Hug-V2-1.pdf):

The output impedance of the transistor stage is about 2.2K (the collector resistor). Followed by peak-to-peak silicon diodes, and the decay cap is 4.7uF with a 120K in parallel on the long decay setting and a 22K in the short setting. (The next connection is the gate of a FET, which should be a high enough resistance to ignore, right?). For reference, the time constant for the attack right now is ~10mS.

If this was an op amp, I could stick a 5K-10K pot in there  after the diodes and make a high pass filter, because the source Z would be trivial and the attack control would have an extremely high range. But an extra 5-10K isn't going to do a whole lot here -- except it will drop the signal a lot on the short mode.

**My main concern here is maintaining the signal strength of the envelope.**

I've got a couple possibilities. #3 I'm really not sure about:

1) I suppose I could make the hold cap 470nF, and then make the parallel resistors 220K (min) and 1.2M (max), then I could use a 50K or 100K pot for the attack. This is the simplest change, but I'm trying to figure out if there's something I'm not taking into account. I do know that the LOWEST attack setting would end up being faster here as well, which is neat, I guess. The time constants with a 100K pot would be 1mS on the fastest setting and 48mS on the longest setting. That's still not a huge range. A 250K pot get the attack up to 100mS, but that sacrifices signal strength.

2) I could use an op amp envelope detector, but that's sort of counter to the idea behind the design in the first place and changes a lot of behavior. This isn't so much my general avoidance of op amps but rather that I have the effect dialed in with the current envelope and I don't want to lose that.

3) Let's say I stick a big cap in series with a pot from the collector to 9V, so it forms a low-pass shelf filter for, say, frequencies above 200Hz. Does this process of removing the high frequencies result in an altered attack time? Will it affect the decay? Or will it just remove signal strength? I don't see how it could affect the time constant too much since the cutoff frequency in the rectifier doesn't change, which it's my current understanding is what determines the attack time.

Any advice?

[PRR]

The JFET gate impedance is nearly infinite. Scale all that side way up, say 50X.

[midwayfair]

Cool, thanks for the advice, PRR!

[Transmogrifox]

Another approach you probably have seen me toying with in other threads is the following:


R7 can be a pot to control attack time.  Another pot could be placed across C7 for release.  R9,R10 can be adjusted to tweak attack release ranges.

This setup can be used to control extremely fast attack times, and can go to reasonably slow attack times. 

Another interesting part is that the higher the signal amplitude the faster the attack.  You can add a limiting function by putting an LED in parallel with R7.

Notice threshold is controlled through R5.  A BJT or darlington could be substituted for Q2 if a harder knee is required.  The MOSFET makes for an extremely soft-knee compression.

I have some scope images linked in another thread.  If nothing else it was really fun to play with this in simulation as well as on the bench making measurements of the compression ratio vs input signal level, attack/release times vs input signal level, etc.

Maybe this provides fodder for experimentation with different approaches.  The QuadraPuss envelope detector adds a bit of current-mode discharge so as I mentioned its main feature is an input level dependence on attack time.  This turns the function into a compressor + limiter to attack fast high-amplitude overshoots quickly instead of clipping percussive strumming.