RE: Analog Input Board Recommendations

You have to add the two numbers.

When you switch channels it takes 6 uS to settle to 1 LSB, then you can do
the conversion
which will have an accurate answer 5 uS later.  Digitizing a changing input
is risky -- usually it'll give a reasonable answer but sometimes it'll spike
with garbage like when 0x01fff is the input but the converter wins the race
with the changing voltage and reads 0x02fff instead.  That's why there is
usually a sample and hold circuit in there somewhere.

In a reasonable board hardware sequences things -- something triggers start
of conversion, the multiplexer channel is changed, after a settling time the
S/H is put to hold and the A/D converts the stable sampled voltage.

Unfortunately there is no shortcut to carefully reading the fine print of
the boards data sheets and manuals.

--wally.
 

-----Original Message-----
From: Tim Holy [mailto:holy_at_pcg.wustl.edu]
Sent: Wednesday, October 02, 2002 1:20 PM
To: Calin A. Culianu
Cc: comedi_at_comedi.org
Subject: Re: Analog Input Board Recommendations


Hi Calin,

On Wednesday 02 October 2002 01:12 pm, Calin A. Culianu wrote:
> Hi Tim,
>
> I was looking at the Measurement Computing specs for the PCI-DAS 6402
> board.  They say their board has an input settling time (to 1 LSB) of 6
> us, and an A/D conversion time of 5us.  Now let's say I'm switching
> channels, do I have to add those two numbers up, or do I take the worst
> case of 1 of those numbers to arrive at my settling time?  What I need to
> figure out is if it takes 11us or 6us to switch channels on that board?
>

I don't know, unfortunately. You might have to call them to find out what
they 
mean by that.

--Tim


_______________________________________________
comedi mailing list
comedi_at_comedi.org
https://cvs.comedi.org/cgi-bin/mailman/listinfo/comedi

Received on 2002-10-02Z17:44:25