Re: electrical problems with Galvanic Skin Response circuit ( GSR SCL )

Hi Steffen

I would firstly make sure that you don't induce 50Hz ripple via your supply
(+5VA_ISO on the schematic), but if the circuit is noise free with a
resistor, then it can't be the problem.  Nevertheless, you should ensure a
clean supply, from a linear regulator such as a LM7805 or similar.

Looking at your scope screenshot, I can see that the period matches your
mains (assuming that the timebase is indeed what you specify), but it
doesn't look like simple hum.  You don't run a phase-controlled dimmer in
close proximity?  Any other electric machine?  You don't grab on to your
soldering iron or some other electric device while measuring?  The SCL
measuring wires don't lie or cross over some current carrying extension
cables or similar?

SCL is a very slowly changing modality, so you could implement a low-pass
filter with a cut-off at a few hertz.  I would stay somewhere under 10Hz.
You could simply do so by increasing the value of capacitor C9 and/or add a
passive RC filter at the output of the instrumentation amplifier.  In the
latter case you would need to consider the input impedance of your ADC. The
R in the passive RC filter should be substantially lower than your input
impedance of your ADC.  At least 10x but ideally closer to 100x.  The R
together with the input-Z will form a voltage divider which will have your
ADC under-read the voltage and it will shift the cut-off frequency
slightly.  This is assuming the input-Z is a constant, in which case you
could compensate for this either in the analog or digital domain.  With
many ADCs the input-Z is NOT constant, due to switched capacitors etc. so
watch out for that.

When filtering, it is always a good idea to do so as early as possible in a
circuit.  So at C9/R24 would be best.

It seems your circuit works until you touch the inputs.  So the problem
seems to lie beyond the circuit, somewhere in your test setup or
environment.  This circuit is working with reasonable large signal
voltages, the input should always be maintained at 500mV automatically, so
noise should not be an issue.

Perhaps you could connect a fixed resistor across the input and THEN
connect first one and later both SCL probes to your finger and see how the
circuit behavior is changed.

Regarding your resistor value query; I only work with SMT components.  A
resistor marked 1003 means 100_plus_3_zeros, i.e. 100 000 ohms.  Similarly
you would have the following values:
2702 = 27K
222   = 2K2
101   = 100
100   =   10         ( ! )
1804 = 1M8

The last digit is always the multiplier.  If only 3 digits are used, it is
likely that the resistor has a highish tolerance, such as 5%, but you will
get 1% resistors that also just use 3 digits.  If 4 digits are used, it
means that it's from one of the finer graded E-ranges which normally are 1%
tolerance.

Regards
Stefan

------------------------------------------------------------------------------
Learn Graph Databases - Download FREE O'Reilly Book
"Graph Databases" is the definitive new guide to graph databases and their
applications. Written by three acclaimed leaders in the field,
this first edition is now available. Download your free book today!
http://p.sf.net/sfu/13534_NeoTech