Picture
A question I have for the future...there's a huge range of low pass filter frequencies different groups use, but what really is the optimal cutoff? What parameters influence this?

2019-12-14: I designed my low pass filter cutoff to be 10 Hz and it worked, but why? And how would I find the most optimal cutoff?

This question has been bothering me for a while...the typical EOG signal is 0.1-20 Hz according to Google. Using 10 Hz worked the first time, and I never bothered to test other values. I searched up EOG low pass filter frequencies online and found a huge range from 5 to 200. Why is that?

Interestingly, I recently found a paper that is trying to remove "contamination signals" from their EEG measurements. I could use what they learn about filtering EOG signals to help me! Here's the link: https://doi.org/10.1155/2018/4853741

​Their conclusions lean toward filtering out lower frequencies than I expected: on the range of 6-8 Hz. I'll update later on my thoughts as to why.



​Electrodes on left
  • L - left temple
  • R - right temple
  • G - grounded to forehead
    • Stabilizes signal so there's less drift
    • Gets rid of some 60 Hz interference


Instrumentation Amp
  • Generates voltage difference
  • Changing gain of amp
    • G = 1+ (50 kOhm/Rg)
    • Rg is connected on the amp's pins 1 and 8
    • Ex: adjust to gain of 500 by using Rg = 100 ohms
  • Instrumentation amp will output 500x the amplified voltage difference


1st order RC low pass filter
  • Resister R_filter
  • Capacitor C_filter
  • Prevents anti-aliasing (not a concern here bc by Nyquist, we need to sample at least 20 Hz for an expected 10 Hz bandwidth and the Arduino ADC samples at 10 kHz which is more than enough)
  • Cuts out noise at all frequencies we don't need
  • RC system works capacitors allow high frequencies through easily but obstructs lower frequencies
    • Impedance Z = 1/(2*pi*f)
    • Creates voltage divider w voltage across the capacitor so filter only allows lower frequencies (cutoff for 3dB intensity is governed by:
      • Fc = 1/(2*pi*RC)
  • Adjust R and C to cut off signals higher than 10 Hz bc biological signal for EOGs is expected in that range


Non-inverting Amp
  • Don't want too much gain - adjust to about 1.3 by changing R1 and R2
  • Gain of amp = 1 + R2/R1
  • Don't go over 5V
  • Measure filtered signal with oscilloscope to see range (should be 2-4V)
    • Instrumentation amp gain was 500x for range of 4-8 mV
    • Target is 5V (full range of arduino ADC)