diff --git a/klusta/nlx2DAT.m b/klusta/nlx2DAT.m
index 3997b25..d244569 100644
--- a/klusta/nlx2DAT.m
+++ b/klusta/nlx2DAT.m
@@ -1,4 +1,4 @@
-function nlx2DAT(animal, experiments, channels, folder2save)
+function nlx2DAT(animal, experiments, channels, broken_channels, folder2save, common_average_reference)
 
 % by Mattia 8.11.18 based on https://github.com/kwikteam/klusta/issues/48
 
@@ -33,23 +33,18 @@ function nlx2DAT(animal, experiments, channels, folder2save)
 % nints = round((memsize.MaxPossibleArrayBytes / 6) * 0.95);
 
 % find all the experiments for the animal that you want to spike sort
-experiments_animal = structfind(experiments, 'animal_ID', animal); 
+experiments_animal = structfind(experiments, 'animal_ID', animal);
 
 % loop over the experiments
 for experiment = experiments(experiments_animal)
     
-%     disp(strcat('writing_experiment_', experiment.name))
+    %     disp(strcat('writing_experiment_', experiment.name))
     % load first file just to calculate the number of block in which you want
     % to partition your recording
     ExtractMode = 1;
     file_to_load = strcat(experiment.path, experiment.name, filesep, 'CSC', num2str(channels(1)), '.ncs');
     [~, temp, ~] = load_nlx_Modes(file_to_load, ExtractMode, []);
     clength = length(temp);
-%     nintsdata = 2 * clength * numel(channels);
-%     
-%     % Partition size of data
-%     nBlocks = ceil(nintsdata / nints);
-%     blocksize = round(nints / numel(channels));
     
     % Create output file
     fname = experiment.name;
@@ -59,30 +54,20 @@ function nlx2DAT(animal, experiments, channels, folder2save)
     
     %% write data
     
-%     fprintf('Saving block 000/000 -ch00')
-%     for blockNR = 1 : nBlocks
-        
-        % Index for chunk
-%         IDX(1) = (blockNR - 1) * blocksize + 1;
-%         IDX(2) = blockNR * blocksize;
-        
-%         if IDX(2) > clength && blockNR == 1
-%             IDX(2) = clength;
-%         end
-        
-%         samples = zeros(length(channels), IDX(2) - IDX(1) + 1, 'int16');
-        samples = zeros(length(channels), clength, 'int16');
-%         fprintf('\t-ch01');
-        for channel_idx = 1 : length(channels)
-%             fprintf('\b\b%2d', channel_idx);
+    samples = zeros(length(channels), clength, 'int16');
+    for channel_idx = 1 : length(channels)
+        if ~ ismember(channels(channel_idx), broken_channels)
             file_to_load = strcat(experiment.path, experiment.name, filesep, 'CSC', ...
                 num2str(channels(channel_idx)), '.ncs');
             [~, samples(channel_idx, :), ~] = load_nlx_Modes(file_to_load, ExtractMode, []);
         end
-        
-        fwrite(fidout, samples, 'int16');
-        
-%     end
+    end
+    if common_average_reference == 1
+        samples(~ ismember(channels, broken_channels), :) = ...
+            samples(~ ismember(channels, broken_channels), :) - ...
+            int16(mean(samples(~ ismember(channels, broken_channels), :))); % common average reference
+    end   
+    fwrite(fidout, samples, 'int16');
     fclose(fidout);
 end
 end