Constant current envelope cap drain?

[Fancy Lime]

Hi all,

I am trying to figure out a simple way to drain the holding cap of an envelope following side chain with a constant current that can be adjusted. I think that this may be beneficial for some types of compressor designs for reasons I will explain when and if it turns out to actually work and be beneficial. I wonder why I have never seen this done. Well, it may be a bad idea, I have to try. My currently most feasible idea is to use an enhancement mode MOSFET like this:



I did not manage to spice this in KiCad and gave up. Can anyone tell my why this circuit is a bad idea? Of course it will only drain the cap with a constant current until the drain-source voltage drops below the saturation region and will behave like a boring old resistor after that point. But if we only request a low current, that extends down further. By choosing C2 appropriately, we should get away with well under 1mA and the datasheet says that in that case we should expect the transition from saturation to ohmic region well below 0.5V D-S, which is probably good enough for the purpose.

So, would this work and why not?

Cheers,
Andy


EDIT: I guess one important question is: What is Id @ Vgs=0, or Id(off) in other words, the minimum possible current setting of this setup. The 2K7000G datasheet (Onsemi) says max 1μA, the BS170 sheet says max 0.5μA. That sounds like it should be OK, no?

[PRR]

It would be more designable and predictable with a negative power rail; is that forbidden?

At least not trying to pull-down to the rectifier reference.

The MOSFET is absolutely not specified or controlled for low voltage low current operation. Every one will be different. And different at any temperature. It can be trimmed now and drift away from happiness as it pleases.

Go ahead and do it that way for experimentation.

[PRR]

Not perfect but maybe perfect-enough for you:

BJTs tend to track if Vbe is forced equal. Vce can approach zero (<100mV). While I simmed a current source, it could just as well be a pot+resistor off the 9V rail. The 2mV/degC of Vbe is small compared to 8.4V forcing voltage.


SPICE figures identical transistors track real close, duh. With two jellybeans from a bag there will be a ~~2:1 offset which is no huge deal for a knob. You can still buy several types of 5% match dual-transistors.

[R.G.]

I'm with Paul. Use a two-transistor current mirror. It's a remarkably versatile circuit component, especially since the control side is not coupled to the output side.

[Fancy Lime]

Paul, R.G., thanks for the hint!

Funny how the brain works, or sometimes doesn't. I did consider a current mirror for driving the vactrol of the optocomp but never thought about using it for discharging the holding cap. Makes sense that a BJT circuit would be more predictable here.

Cheers,
Andy