sample & hold caps/ timing.

[Fraser, Colin J]

> -----Original Message-----
> From: Grant Richter [mailto:grichter at execpc.com]
> Sent: Thursday, April 13, 2000 4:14 PM
> To: Roel Das; Synth DIY
> Subject: Re: sample & hold caps/ timing.
> 
> The calculations are based on the RC time
> constants formed by the internal resistance of the switch
> (Ron) and the capacitor. An RC time constant describes
> the amount of time it take for the DC value to go to 63%
> of the final value. The rule of thumb I was taught is
> the value will have settled to 99.9% after 10 RC or 10
> times the time constant.
> 
> Your scanning rate has to update all output channels
> in less than 1 millisecond. So take 1 millisecond
> and divide by the number of output channels. That gives
> you your "dwell" time at each S/H. Now find the Ron
> of the multiplexer switch and chose a capacitor such
> that the dwell time is 10 times the time constant.

An alternative view of this...

In my own cv convertor, all input is handled by interrupts.
The main loop does little more than refresh all the channels as fast as it
can.

If you want to update all the channels to reach a new value within 1ms,
there is no point having a time constant on each channel much less than 1ms.

Think of the demultiplexer as a multi-channel switched capacitor low pass
filter.

Each cap gets the same percentage share of the DACs time.
A small amount of the potential duty cycle will be wasted waiting for the
DAC to settle before opening the switch, but with a suitable DAC this will
be only make a small difference.

Each channel therefore has a RC constant equal to (number of channels)*
switch resistance * C.
Use this to calculate the size of cap needed for the 10RC value of 1ms.

You don't actually need to charge each cap to it's target value in one go.
Just refresh each channel as fast as you can while idle, and the effective
time constant of your switched cap filters will determine the time it takes
to get to the correct value.


Colin f