simple sample/hold circuit?

[Magnus Danielson]

From: Andy Sloane <andy at guildsoftware.com>
Subject: simple sample/hold circuit?
Date: Sat, 2 Oct 1999 11:58:10 -0500

> I'm thinking of making a quick, cheap midi2cv converter with multiple outs
> using a PIC16C84 and a single DAC.. Like a PIC version of PAiA's design
> (which IMHO is a little overcomplex).  My question is, what's a good way to
> do sample and hold on the DAC output so that I can scan through all the
> outputs using a single DAC?  I guess it doesn't need to hold for very long,
> but I haven't written the PIC code yet so I don't know how long it needs to
> hold for.
> 
> My electronic devices book suggests a design simply using a MOSFET and a
> cap.  Is this the best way to go?  I've never really experimented with
> mosfets before.  Obviously you'd need an opamp follower after that or
> something... So are there any other s&h tricks I should know?  Can you do
> this with BJTs somehow?  or opamps?

I'd go with a 4066 and a cap. Don't forget a high impedance buffer.
With the 4066 (or similar, there are many diffrent analog switches to match
your needs, 1-8 and 1-16 in the CMOS familly) you can ship in normal digital
levels for control.

The high impedance buffer could possibly best be some JFET input op-amp
in a source follower or implification configuration.

The 4051 will give you 1 to 8 with decoder (3 address + 1 inhibit) in a
16 pin DIL.

If you can keep a steady pace (constant update frequency), this could work
pretty well. Keeping to caps with low leakage migth be nice. For not that
much money you can get pretty decent caps, so it's really not a big issue.

With a curcuit like this, I would make a interupt routine of some sort, so
that I update the caps at a steady regular phase from some memory positions.
Whenever the other code is finished, it updates these memory position.
This way will the memory positions and not the leaky caps be the point of
syncronisation. Small tricks like this makes the analog output more stable
while you get larger freedom in what you do internally with the other stuff.

Also, I would let the interupt update one cap, the next interupt update the
next cap etc. This will increase the interupt frequency, but on the other
hand you would not have to wait for a DAC conversion to complete in order to
step to another while being inside the interupt routine, while you wait you
simply schedule in your normal buissness.

Cheers,
Magnus