sample & hold caps/ timing.

[Jim Patchell]

Roel Das wrote:

> Hello again,
>
> I was wondering.
> How do I calculate the value of the sample and hold caps for my midi CV?

    The size of the cap will depend on your refresh time, and the leakage in
the circuit, and how much ripple you can tolerate.  You can aproximate the
ripple by using VoltsPerSecond = CurrentInAmps / CapacitanceInFarads.  So if
your op-amp has a 1 amp leakage current, and you are using a 1 Farad hold
capacitor, the voltage will change at 1 volt per second.

>
> And the time the MUX should be enabled to load them?

    You will basically have an RC time constant here.  If you have a CMOS
switch that has 1000 ohms of resistance, and the capacitor is .001uF, the time
constant is 1uSec.  My rule of thumb is that to settle to 16 bits requires
about 10 time constant times, so you would want to gate for 10uS in this case.

>
> Which caps are best for this? From what I've read (Harry's primer on caps)
> polystyrene film caps would be best. Or polycarbonate film. If I can find
> these in time...

    Harry is 100% correct.  Polystyrene caps are availiable from Mouser
(http://www.mouser.com)

>
> And are there nice replacements (pincompatible and better) for the tl082
> buffers?
>

    Don't know of any off hand myself, they they do exist.  Burr Brown or
Analog devices I am sure have some.  (You can get Burr Brown parts from
Digikey http://www.digikey.com)

    For sample and hold buffers, the input bias current is most important.
Also, if you get a buffer that has a 100 fempto amp bias current and use it
with an analog switch that has 1 nanoamp leakage, you are wasting your money
on the buffer.  You should be able to calculate how much leakage you can
tolerate by estimating what kind of refresh rate you can sustain with your
controller.

>
> I need calculations. They make me look good for the teachers...
> Teachers are mean.
>
> Thanks
> Roel

    Just my humble opinions.

    -Jim