I'm trying to design an EOG system based on how the JINS MEME glasses did it. 3 electrodes, two on either side of the nose and one of the nose bridge. This will be battery powered. The expected outputs are shown in the image (what JINS MEME academic pack displays).

outputs of EOG system

From this image I understand that the nose bridge is the reference electrode and there are two channels. Almost all EOG systems are bipolar and use some kind of differential instrumental amplifier followed by filters and more amps to increase the gain. From the image am I right to assume that this should use single ended amplifiers instead and do the additive and subtractive operations on the software end?

Also to remove common mode noise like EMG and ECG, I understand that line noise may not be that big of a deal and I'll be using passive band pass filter for the range 0.5 to 40Hz. If I go for a DRL circuit, I would need another electrode for ground then. That might mean this design probably went for a DRL-less circuit, am I right to assume that?

Essentially I'm curious what the analog front end of this 3-point EOG electrode system would be like, I've looked into DRL based bipolar circuits and none of them seem to be applicable to this system.


1 Answer 1


I've take hundreds of EOGs, and built dozens of EOG circuits over the years.

How you go about it will depend very much on what you're trying to accomplish.

First, the method is predicated on the retina (the nerve layer on the back of the eye) is negative with respect to the front of the eye. Therefore, to measure horizontal eye movements, you build an amplifier that can measure an electrode on the left and subtract that from the signal from an electrode on the right.

If you're trying to measure the movement of the left and right eyes independently, then, you should have two electrodes on each eye. If you don't care about each eye, then you can use an electrode on the left temple and one on the right temple (you will lose information concerning vergence). Your signal will be twice as big as using only one eye.

For all cases, you will need some sort of reference electrode. You don't want this to be where it's subject to motion artifact. I tend to use the skin right over the mastoid bone behind the ear. It was recommended to me by people who have been recording eye movements much longer than I have.

The three electrode setup you're talking about is really subject to movement artifact. I would recommend using a reference electrode behind the ear, and an electrode right at the lateral edge of each orbit, if you just want to know where the subject is looking. If you need to know where each eye is looking, I recommend an electrode on each lateral temple, and an electrode on each side of the bridge of the nose (as close to the eye as you can get), along with a reference electrode on the mastoid. If you need one eye, one trode on the lateral orbit, one on the bridge of the nose as close to the eye as you can get, and a mastoid ref electrode.

EOG tends to drift, so many people put a high pass filter, maybe in the 1Hz or lower region,, even though clinical EOG often goes to DC. At the high end, it depends on what you're interested in. If you need to know the shapes of the fast eye movements (saccades), it needs to be above 100 Hz minimum. If you just need to know where the eye is pointing, 40Hz should be fine.

The signal will be in the 10's of microvolts range (which is on reason why doubling your signal size by capturing the signal across two eyes is easier). Plan your gain accordingly. You'll likely want a modest gain preamp (often an instrumentation amp), some way to remove offset (filtering or an offset pot), followed by more gain.

DRL used to be a whole lot more important before readily available amps, but might still buy you 10-20 dB if you do it right. Management of electrode wires (braiding and shielding, for example) is probably more important than DRL.

For the electrodes, Ag/AgCl with some electrode jelly is how to go. I often use disposable pediatric ECG electrodes, cutting the adhesive ring so I can get then right at the edge of the orbit.

  • \$\begingroup\$ is EOG a DC or AC signal? There's no clear answer on the internet \$\endgroup\$
    – roaibrain
    Jul 7, 2023 at 2:43
  • \$\begingroup\$ @Rohan - I was an eye movement scientist for many years, and EOG is effectively a DC signal. Picture just looking off to the left and holding gaze there for a few minutes. An AC filter will make it look like eye position is drifting back to center. \$\endgroup\$ Jul 7, 2023 at 15:43
  • \$\begingroup\$ Of course, for many purposes, you don't need to know that, and if you don't, it's much easier to deal with offsets with a high-pass filter than by any other method. \$\endgroup\$ Jul 7, 2023 at 15:44
  • \$\begingroup\$ I see, any idea why a lot of people AC couple EOG to remove the offset. Since it's a DC signal it seems all the more sensible to just use a high precision ADC to sample the signal with the offset. \$\endgroup\$
    – roaibrain
    Jul 7, 2023 at 15:45
  • \$\begingroup\$ @rohan, I can think of a variety of reasons. We're talking tiny signal, with offsets often larger by several orders of magnitude larger than the signal. In a sense, I suppose you can extend your question to say "why amplify at all?", but if you do feel compelled to amplify, care and multiple stages are required to prevent these offsets from saturating an amp. If, for example, you're just trying to identify REM, it would be silly to nots imply high pass filter \$\endgroup\$ Jul 7, 2023 at 16:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.