peter green wrote:
>> If your actual input impedence is 910K || 10K (call it 10K), then  a
>> 1ua bias current results in an error of 10K*1ua = 0.01V
>
> Yeah but relatively fixed errors can be calibrated out, it is random
> noise that is a big problem.

But the pin bias currents aren't fixed.  They vary significantly by
temperature, age, what other circuitry happens to be on in the PIC, etc.
You can't reliably calibrate them out by nulling them once at manufacturing
time.

A low pass filter only eliminates some parts of random noise.  It's usually
better to reduce the picked up noise than try to eliminate some of it later.
This is where good grounding, a clean reference supply, and the like are
important.  After all that, if you've got the processor cycles it's better
to oversample then filter in software.  That will filter out some of the
quantization noise, which no amount of analog filtering can do.

Otherwise, a heavy capacitance on the A/D pin can also make things worse by
acting like a memory of recent events.  This means charge pump noise, for
example, can be carried between successive readings.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000.
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist