Multiplexing a full wave rectifier!

[Magnus Danielson]

From: "Bjorn Julin" <bnillson at hotmail.com>
Subject: Re: Multiplexing a full wave rectifier!
Date: Wed, 04 Oct 2000 22:16:30 GMT

> >My answer would be like this:
> >
> >Yes and No.
> 
> Me to, first i was sceptical, but!
> 
> >Yes, if you would have to set up hardware just to do the multiplexing
> >I think it will be more effort than just making several precision >full 
> >wave rectifiers.
> 
> I have the oposite feeling, if one spends time to do it right
> in this application i will save a lot of componennts (goal1)and
> setup time, calibration time, and solder time. (goal2)

Well, I don't rule out that you could do with less components, if you
certainly feel that you do, please go ahead!

> >No, first, the rectifier function as such is not a very time >dependent 
> >thing, so you don't build up much of a state as you
> >do in say filters.
> 
> Hmm, takling about filters, is it insane to multiplex a allpass
> among a bunch of channels? I dont need to go above 12Khz!

Yes, multiplexing an allpass filter is insane, in reality you would
get crosstalk between the channels, since the capacitors would keep
the charge. Multiplexing the caps too would possibly solve this, but
then you have a lot of switches, so... here the benefit is less
obvious. The cross-talk would be strongest at the low frequency, and
better at higher frequencies, but it would be so bad that you better
just mix the signals directly, since that is basically the effect you
will get.

> >However, you naturally causes filtering effect due to the sampling
> >effect. Also, a envelope follower would be sligtly receptive to
> >glitches when you go from low-to-high or high-to-low values.
> 
> Sampling effect is the minor problem in this app.
> Hmm, glitches, the rectifyer are in front of the mux?

If you run a switch-rectifier-switch-envelope follower setup, if you
have two distinctly diffrent levels on two channels and the switching
is not well-matched, then the envelope follower for one channel would
sence leakage of the incorrect channel for some short time. Thus, the
detected level would become incorrect. I think that this effect is not
very strong, so I would not care about it most of the time.

> >If it is just to make better use of the rectifier hardware, then I am
> >leaning to the more doubtfull level, where as if you really just want
> >diffrent inputs you surely should go ahead.
> 
> I want o save "lots and lots of components" thats why i want
> to multiplex.

If you do feel that it does (and I having not really made an serious
effort to compare component count) then I see no problem to why it may
very well work while acheiving the goal that you want.

Cheers,
Magnus