Eye tracker X,Y dva drift

Hi Jaewon.
I need your help with the following. Today, I noticed that there is a drift in the X,Y, signal acquired from the eye tracker. Please take a look at the attachments.
1-Saccades detected during the task, starting at fixation point (0,0)
2-X and Y traces over time for the same session

Please let me know your advice, how may I correct for the drift?
It is strange that this is not reflected on the MonkeyLogic interface, since the fixation window that I have established in my codes is 2dva diameter and the task was correctly performed throughout this session.
This drift is more than 5 dva on each X, and Y.

Thank you very much for your help. I will appreciate any ideas on how to fix this issue.

It looks like a grounding problem. From what you said, I guess you plotted the signals recorded by your signal processing system, didn't you?

I do not think there is much I can help as long as you use the voltage input. In fact, NIMH ML has nothing to do with this problem, if your AI configuration is correct.
https://monkeylogic.nimh.nih.gov/docs_NIMultifunctionIODevice.html#AIGroundConfiguration

You might not see the effect of the drift, if you turned on [Auto drift correction]. Then, you can just use the eye signals saved in BHV2, since they are corrected against the drift already.
https://monkeylogic.nimh.nih.gov/docs_MainMenu.html#DriftCorrection

You can switch to TCP/IP, which is digital. I personally do not like voltage input because of this grounding problem and the resolution loss.
https://monkeylogic.nimh.nih.gov/docs_TCPIPEyeTracker.html

Hi Jaewon.
Thank you very much for your response. It is very helpful.
Yes, the plots are using the signal I recorded on the acquisition system (Neuralynx).

I will try the options you sent.
It is definitely a good idea to use the signal from the bhv2 files, I would have to figure out the way to process it and sync it with the spikes and LFPs timing...

Thank you.

If your task starts with central fixation, you can also slice Neuralynx signals into each trial, subtract the X and Y values at the time of fixation in each trial and put them back together.

You have the same timestamps in both Neuralynx and BHV2. Either way should be easy.

That's a great idea.
Yes, indeed our task requires to hold fixation throughout the trial, so I can definitely extract the fixation position at the start of each trial.
I will try both methods and will let you know how it goes.
Thank you very much again for your help!