On 5/12/07, michael brown <n5qmg@earthlink.net> wrote:
> Crist=F3v=E3o Dalla Costa wrote:
>
> > Yes, a problem with the PIC's hardware. Maybe it's just the
> > particular PIC I'm using. But I don't think that it's a problem with
> > my hardware since I plugged in a 16 bit AD from Analog Devices and it
> > has just about 1 lsb noise, and the PIC's 10 bit AD has at least two.
>
>
> I have used a 16F88 on a solderless breadboard to measure and log temps
> from an LM34.  I didn't sleep during the conversion and I still had no
> trouble obtaining consistant 10 bit readings.  I used an external Vref
> that was simply a variable resistor and a filter cap.  1.024V well
> outside the minimum allowed by the datasheet ;-), but hey I wanted .1
> degree resolution so I needed 1mV resolution in the ADC.  Seemed to work
> just fine.
>
> I don't think you should see that kind of noise in your readings, unless
> you have it in your power supply as well.

And we are exactly in the point where we've been. A ripple in the supply is
equivaled with a "noise" on the ADC input which has non zero value
after averaging.
Imagine a simple situation when a linear power supply has a lower voltage o=
n the
input than is required for the minimum value measured at the bottom of
the input ripple. Like a 7805 which hasn't 2...4V at the bottom of the
input ripple (around 4...7V measured with a RMS voltmeter) The output
ripple will be visible at 10 or 20mS (on 50Hz systems) as a reversed
small peak on the Vdd (with lower value than Vdd).

If your supply is clean indeed, a variable output impedance of the
measured signal may cause the same problem. Or a higher input
impedance (like 47K-100K) than datasheet recomandations. Or you have a
software problem and you don't let those 20microseconds TACQ.

Vasile

-- =

http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist