[Mne_analysis] Error using filter data

[Dirk van Moorselaar]

        External Email - Use Caution        

Dear MNE developers,

I am currently using MNE functionality in a time frequency analysis. I am using a fifth order butterworth filter to bandpass filter data with the following syntax:

filter_data(data, 512.0,8, 12, method = 'iir', iir_params = dict(ftype = 'butterworth', order = 5))

, where data is a 3 dimensional array (epochs X channels X time points). The curious thing is that this function works fine with one dataset, but gives me an Value error stating that the length of the input vector x must be at least paddle, which is 1084.

The dimension of the data set that works is 934 by 64 by 1077, while the dataset that crashes is 1084 by 64 by 1077. So both datasets have the same number of electrodes and time points but only differ in the number of trials. 

It is unclear to me why padlen is determined by the number of trials. Am I misunderstanding the use of this function or is there something else that I am doing incorrect.

I really appreciate the help,

Best,

Dirk