In the last article I made a PCB that allowed me to easily inject signals from my signal generator into the EOG cables. This allows me to more easily simulate signals without wiring myself up, but also allows me to isolate the effects of the cables. The first thing I wanted to try was to shield the cables.
To start with I used the unshielded cables and connected them to to the signal generator and generated a 2Hz signal. I probed the output of the second amplifier.
Here we can see that we are actually observing a relatively low amount of noise which is in contrast to the noisy EOG traces we were seeing before. The reason for this is that the output impedance of my signal generator is only 50Ω where as the output impedance of the human body and probe is much higher.
Increasing the resistance between the signal generator and the EOG leads to about 10kΩ resulted in a much larger 60Hz noise component which looks much more like the EOG traces we have previously captured.
From here I progressively added grounded shielding. First I made a set of cables where each of the three leads was a coaxial cable with the outer conductor grounded.
This made a significant improvement but there was still some noise due to to the unshielded signal generator adapter.
To remedy this I 3D printed a housing for the signal generator adapter and coated it in copper tape which I connected to ground.
The noise levels are captured in the following three images. You can see the 60Hz noise decreases as shielding increases.
There is some remaining 60Hz noise. Firstly, the EOG PCB is not shielded so there is a short region where the signal is exposed. Secondly I believe that some of the remaining signal is coming from the signal generator itself seeing as it is powered from 60Hz mains and the remaining signal is a highly distorted signal with a 60Hz components.
Obviously, all of this shielding is useless if it doesn’t improve the signal when taking a real EOG.
Here is an EOG taken with the unshielded cables (in red), and also the 60Hz notch-filtered signal (in blue)…
… and here is an EOG taken with the shielded cables (in red), and further filtered with the 60Hz notch-filter (in blue).
We can see there is a significant improvement between the red traces. Arguably there is not a large improvement between using a notch-filter and using shielded cables. However there are problems associated with the notch filter:
- Not all countries use 60Hz, so you would need to also accommodate a 50Hz notch-filter too.
- There is potentially some 60Hz component in the EOG signal that would be removed by this filter resulting in a distorted signal.
- The induced noise is not a perfect sine wave resulting in harmonics at 120Hz, 180Hz etc which would not be removed by the notch-filter.
- The mains noise consumes some dynamic range of our analog circuitry meaning we have to either increase our voltage or decrease our gain to provide headroom for this noise.
Overall I am pleased with the amount of noise that this has removed. There are also other contributors to noise that I need to evaluate such as power supply noise.