EMG recording shows discontinuous jumps in data
I'm using a Cyton+Daisy board to record EEG and EMG data at 500Hz to an SD card while study participants walk on a treadmill.
For the EMG data, I'm using 2 OpenBCI disposable foam gel electrodes on the tibialis muscle (ie. the shin). They are connected to the Daisy board on the N8P pins (bipolar) using OpenBCI snap electrodes. Using the GUI, I turn off BIAS and SRB as instructed here.
For the EEG data, I connected 14 channels which are recording good data. I use one earlobe as SRB for both the Cyton+Daisy board, and I use the other earlobe as BIAS for the Cyton board.
The EMG data is showing massive jumps when the participants walk on the treadmill. '4_trials.png' shows some stationary EMG data collection (which looks good) and then 4 trials (3 mins each) on the treadmill where you can see large jumps in the data. 'ZoomIn.png' is zoomed-in on part of one trial. It looks like useful data is being collected around ~3.5*10^5 and around 0, but for some reason there are jumps between those two levels. (All this data has been scaled to microvolts).
I've tried grounding the EMG wires to AGND. Should I try connecting the EMG to BIAS or SRB? Any ideas what could be causing this?
I've been stuck on this problem for a few weeks, so any help is appreciated!!
4_trials.png: https://imgur.com/w799C7j
ZoomIn.png: https://imgur.com/RrENWaR
Comments
Can you clarify what the scales read? Why is your midline / idle trace at ~ 3.7 * 10^5 ? Both axes seem to have a 10^5 label. Are you pre-filtering to remove the DC bias? This is typically done with a high pass at say .5 Hz, or a bandpass. Have you tried adjusting the channel scale factor on the EMG channel? These number seem way high.
What happens when you follow the tutorial steps?
https://docs.openbci.com/GettingStarted/Biosensing-Setups/EMGSetup/
William
Also, how does the EMG time series look when graphing in real-time with the GUI?
I scaled the data from the SD card according to this page. I used the wrong gain (24 instead of 1) and scaled it to uV, which is why the y-axis is too large.
Here is 2 mins of treadmill-walking data with the correct gain, after running through a high-pass filter at 0.5Hz: https://imgur.com/ebiUxgb
Here is a zoom-in on 10 seconds: https://imgur.com/r3BQW68
It looks like the EMG data is occasionally jumping ~8 volts during heel strikes, but still collecting good data on either end (ie. not saturating). I did try the tutorial set-up on the shin muscle with the knee as AGND, and it collects good data stationary, but shows the same disconnections on the treadmill.
My working theory is that there is a problem with the snap electrode interface during walking. I'm going to try using paste electrodes to collect EMG data and see if the problem remains.
Also, I haven't seen the EMG data in real-time with the GUI because I'm collecting data at 500Hz, which doesn't stream over Bluetooth (only saves to SD card). For this next test I'll switch to 250Hz and stream using the GUI.
@super248 Is everything fine when the Cyton is at-rest and no electrodes attached? This is another crucial scenario to rule out physical damage.