std_ersp还是newtimef

ref

for computing ERSP at the single dataset level, you should use the newtimef function. The std_ersp.m function is for STUDY. As outlined in the function documentation, by default the function processes ICA components. To make it process data channels, you have to provide them as input.

std_ersp(EEG, 'channels', {EEG.chanlocs(1).labels}, 'type', 'ersp', 'freqs', [2 20], 'timelimits', [0 1800], 'cycles',  [3 0.5], 'trialindices', [1:EEG.trials]);

This will save a datafile that has the same name as the EEG dataset with the extension .datersp. You may load the file below

dat = load('-mat', fullfile(EEG.filepath, [ EEG.filename(1:end-4) '.datersp');

Then you may replot the data

figure; 
tftopo(dat.chan1_ersp, dat.times, dat.freqs);

A lot of efforts. Better use the newtimef function unless you want to compute time/frequency decomposition on all data channels or components.