If the original data are not regularly sampled, that could account for the distortion. (An easy way tto check this is to calculate the standard deviation of the differences in the time vetor, if is greater than about 
or so times the mean of the differences, I do not consider the sampling intervals to be equal, or the sampling frequency to be constant.) If the data are not regularly sampled, first calculate the mean sampling freequency for the data that exist (I usually define the mean sampling frequency as ‘1/mean(diff(t))’ where ‘t’ is the time vector), and then initially resample the data to that sampling frequency usiing resample. After that, resample it to your new sampling frequency.
or so times the mean of the differences, I do not consider the sampling intervals to be equal, or the sampling frequency to be constant.) If the data are not regularly sampled, first calculate the mean sampling freequency for the data that exist (I usually define the mean sampling frequency as ‘1/mean(diff(t))’ where ‘t’ is the time vector), and then initially resample the data to that sampling frequency usiing resample. After that, resample it to your new sampling frequency. .
