Processing sensor channels

Post Reply
Posts: 2
Joined: Thu Oct 03, 2019 6:40 pm

Processing sensor channels

Post by cll008 » Thu Oct 03, 2019 7:07 pm


I have a few questions in interpreting aux sensor channels from a dataset with 16 aux channels, any advice is appreciated:

1) Is this still the correct way to process GSR signals ... nsor#p2450
2) After scaling resp and pleth, are the units simply uV (no conversion to, for example, mm displacement in resp belt)
3) I'm having a hard time arriving at reasonable temperature values, is this an accurate translation of what happens in ActiView?
signal1 = double(Non_Inverting_Temp - Inverting_Temp) + 2*2^30
a = -1302781; b = 157168911; c = -3342225101;
signal2 = sqrt(4*a*signal1 + b^2 - 4*a*c)
signal3 = (signal2-b)/(2*a)
signal4 = int(signal3*1000)*256
signal5 = double(signal4)/8192*32/1000 (conversion from uV to degC)


Site Admin
Posts: 973
Joined: Fri Mar 26, 2004 7:00 pm
Location: Amsterdam, Netherlands

Re: Processing sensor channels

Post by Coen » Tue Oct 08, 2019 2:49 pm

1) Yes, GSR processing has not changed
2) Yes, RESP and PLETH are in uV, but not calibrated.
3) First step is a little different:

Signal1 = double(Non_inverting_Temp+2^30) - (Inverting_Temp-2^30)

If I remember correctly (it's 20 years ago) the signal (Non_Inverting_Temp - Inverting_Temp) could overflow integer32 values and therefore the adding/subtracting of 2^30 was added before the differential-to-single-ended step.

Best regards, Coen (BioSemi)

Posts: 2
Joined: Thu Oct 03, 2019 6:40 pm

Re: Processing sensor channels

Post by cll008 » Tue Oct 08, 2019 5:34 pm

Thanks Coen! I found the data I was working with scaled the aux sensor channels the same way as EEG channels, so I had to first recover the 32bit words. Everything works now.

Post Reply