Tesaro's problem reminded me of a problem I had last week that
took hours to solve.

Using a 16C71, I could get accurate readings from one of the analog
inputs if I read it on its own, but when I tried to read all 4 inputs
in succession, the reading depended on the adjacent ones.

As this was my first PIC project, I spent lots of time playing with
software timing and everything else I could think of. I wonder how
many times I can erase and reprogram this chip? ;)

Measuring the input with a DVM or scope always showed the correct
voltage on the pin, but the value determined from the A/D varied.

It turned out, after pulling of much hair, that there was a cold
solder joint between the sensor and the chip. It looked just fine
to the eye, measured correct voltages to a high impedance meter,
yet had a resistance of 10K when I finally got around to measuring
the resistance from the 16C71 pin to the sensor it was supposed to
be connected to.

I reheated the joint, and the problems just went away. If I had
been more experienced with PICs it probably wouldn't have taken
so long to find, but it was a good reminder that sometimes you
have to go back to the basics and think about what is going wrong.

That additional 10K increased the time constant on the sampling
capacitor to the point that the readings became dependent on the
previous samples. If the previous sample was close to the correct
voltage, everything appeared to be working fine. Otherwise, the
results only tended in the right direction.

> Tesaro Sandu <tsandu@ST.NEPEAN.UWS.EDU.AU> noticed:
>
> the problem was the flux we use contains some sort of salts...
> when we wash the boards.. these salts wash off.. but they still remain
> in the washing liquid... Then when we rince them .. still a little of
> the solution remained in the liquid..  when the boards dried.. this very
> very very very small amount of salts.. deposited round the crystal/caps
> region.. (beacause thats on the bottom of the board)  anyway .. from the
> salts it made the PCB conduct !!! Not much.. say 500k to 1 MEg within 2mm
> BUT THATS WHAT MADE THE WHOLE PROBLEM!!  Arghh took me a good arm and a
> leg to work that out.. would of never suspected that ..
>
> Just thought i'd let yous know incase any have the same problem

--                                 peterj@netcom.com