On Mon, Sep 13, 2010 at 5:57 AM, Malcolm Parker-Lisberg
<
mparkerlisberg@...> wrote:
> Alessio
>
> The high impedance present due to electroless tin manifested itsself on a sample and hold 14 bit A to D circuit and associated amplifier. Six boards had excessive errors whilst two worked perfectly. The two working boards were not electroless tin plated. The effect of the tin caused the sample and hold capaicitor DC value to droop during the A-D conversion. It could not be measured with a multimeter. The associatd DC amplifier had incorret gain, The leakage across the gain setting resistor was sufficient to cause errors. All the boards had been cleaned in Arklone to remove flux residues.
>
> Malcolm
>
> I don't suffer from insanity I enjoy it!
To sum up, unless you are using circuits that are going to be affected
by very minute amounts of resistive coupling (probably 10's or 100's
of megohms) you're going to be fine. As mentioned above, there was a
small capacitor that was being discharged by the slight conductive
layer of tin over the entire board, and in this type of high-accuracy
circuit even a few micro-amps of leakage is too much. Probably not an
issue unless you're working with very high-impedance circuits with
MOSFET-input op-amps, instrumentation amplifiers, and medical
equipment meant to measure extremely small changes in voltage and
current (As well as needing to be almost perfectly isolated).
High-voltage circuits may also be affected (maybe 1KV+) since leakage
current corresponds directly to voltage. I would be curious to know
if there would be a certain voltage where the conductive layer would
be vaporized between the traces at the highest potential.
There might be one last problem triggered where you could have traces
running over or around the tin-contaminated area at microwave
frequencies, which might produce odd problems with parasitic
capacitance, particularly with amplifiers (Or amplifiers suddenly
becoming oscillators!) or oscillators themselves undergoing unwanted
frequency shifts.
Regards,
Larry