Dear Raghunathan,

here is your screen dump again:



If one has a careful look at the spikes, then something interesting can be noticed:
Each spike value is wrong of course. But this wrong spike value seems to be the correct value that the curve should have right at the moment when the PRECEDING spike occured. To illustrate what I mean, please have a look at the following snippet:




Take a certain peak value, that seems to be totally wrong, shift it in direction of the black arrow to the left. This yields a certain red point, which seems to be the correct value of curve, right there where the preceding spike occured. This principle works with the most, but not all spikes.

Because of this characteristic scheme of spikes, I don't believe that these spikes result from noise or something injected via stray capacitance: There could not be this strange correlation between the individual spikes...
So, finally, I tend to agree with Donald and Jacob, which believe that there's something wrong with the code. It seems, that the data was sampled correctly, but drawn at wrong locations on screen. This could be the result of several mistakes: A wrong data logging, that does not store the data in the right order of occurence, a wrong transmission to PC, that does not send the data in the right order of occurence, or your VB (visual basic?) program, that does not put the data at the right locations on screen.

There's a test, which could clarify, whether the above assumption is correct: Connect analog input of ADC to analog ground directly at ADC. Then, as all the data sampled must be identical now, the order of occurence does not play any role and the spikes should disappear. The same should be noticed, if signal from your valve presents a constant signal. Then, the spikes should also disappear.

Maybe it's a good idea to have a look at your ISRs (interrupt service routines) and what's happening on the stack...


Good luck,
Kai