Observation:
In line 68 of _edf.py, the data is padded with zeros using the following line of code: data = np.pad(data, (0, int(pad_width)))
I had a situation where the range of my raw EEG data was way far from zero which caused some problems. I was wondering if instead of padding with zeros, it would be better to pad with the last data point value.
This update hasn’t yet been incorporated into the stable MNE release. Could you try using the development version of MNE to see if it resolves the issue? Alternatively, you could wait for the upcoming MNE 1.8 release, which is expected within the next few weeks.