[sdiy] CMOS as voltage variable resistor

[Magnus Danielson]

From: Grant Richter <grichter at asapnet.net>
Subject: Re: [sdiy] CMOS as voltage variable resistor
Date: Wed, 01 May 2002 01:37:21 -0500

> I am willing to fight to the death for your right to disagree with me ;^)
> 
> It could be useful, so here goes,
> 
> > 
> > I don't think so.  The control pin is buffered so you will have
> > almost NO control of the "linear" state of the switch...
> 
> Perhaps it is still possible to get A series 4016s rather than buffered B
> series? That would give more linear range. The idea is usually shown with
> matched photoresistors, but other types of voltage variable resistors
> (MOSFET or JFET) may be able to use the same technique.
> > 
> > Its a switch, on or off... no linear range at all. Think of it like a
> > comparator with no feedback... try and make that stay at exactly
> > zero volts... no way!
> 
> I believe the purpose of the B series is to increase the overall voltage
> gain, but there must still be some linear region. Otherwise you could not
> make op-amps out of CMOS inverters, which you can do according to the
> National Databook. I would expect the linear region only to be about a volt
> in width, but our friend the op-amp might manage that.

OK. Here I can't keep shut!

Thing is, in the B-series, there is unbuffered variants of some chips,
for instance, the 4069UB is a hex inverter - in the B-series. Each
inverter is just a classic CMOS inverter, a N-channel MOSFET and a
P-channel MOSFET. It's available and just asking for being bent beyond
recognition.

We've seen a number of different uses of a bent 4069UB, among those a
VCO, a VCF, a VCA... and I've tossed together a concept drawing (never
tested - so it is full of misstakes... I am just about to learn these things
properly) of a VC-ADSR.

Naturally, the goal is to bend the 4069UB beyond all recognition until
all ICs in a synt is 4069UBs. Naturally, it will sound awful, but that
was never the point to start with... ;O)

(Now I do it again, doing odd things for the sake of doing odd things
and not realted to what is practical or even "good" engineering.)

> > OTOH... Grant is correct in that putting one TG in the feedback of
> > an opamp can linearize the ON RESISTANCE... in fact it can cancel
> > both the on resistance and any voltage dependence of on resistance as well.
> > 
> That would be very useful, wouldn't it? Perhaps we should do a little bread
> boarding. Being able to reduce the distortion in a JFET variable resistor
> would also be very useful. Could the idea work for that?

Starting form an CMOS switch chip? No. If you want a linearized JFET
variable resistor setup to be as linear as possible I think you might
want to start outside the CMOS chip series, at least through-wise.

Cheers,
Magnus