data spikes

[_Bryan]

Hello,

I have attached an image of some current temperature data and the spikes I am getting.  I am logging data at 1 sample a second and there is about 8 hours of data in the image.  The spikes can go to -160,000°C, usually for 1 or 2 samples then return.  Any idea on the best way to get rid of them?

[Matthew_Williams]

Hi _Bryan.  First guess is that you have an intermittent connection that is opening up briefly.

What hardware are you using to make the measurement?

As far as salvaging your existing data, can you just drop anything below an arbitrary threshold?

Another approach would be to just discard any readings outside of 3 or so standard deviations of
the previous n readings.  Or just discard any that differ from n-2 or n+2 by more than x degrees.
Lots of ways to identify readings this out of whack.

Matt

[_Bryan]

Hello Matt,

  I am using a pcie-6259 and scb-68, all new equipment on a new machine.  I just looked at the data again and only the thermocouple data gets the spikes.  In the screenshot just above 0 is a pink line which is a voltage signal and has no spikes.  And around 25 there is a brown line which is a temperature reading for the office which has no spikes.  That temperature is a short run, but the others run about 30 feet outside into a pit.  And it is the pit temps that are having issues.

  I am currently working on a filtering method to clean things up, but if I could just fix the problem it would be better.