t0 is time stamp and defines the time of the first point in the waveform. For the rest of the waveform, dt is used to figure out the time relative to t0. So if I had data taken at 4pm, I could just replace t0 with a timestamp at 6pm, and then the data says it was taken at 6pm. You can check if your waveform already has non-default a value for t0, in which case you wouldn't have to do anything.
Cluster to array works well when everything in the cluster is the same data type, otherwise it has to coerce things be the array datatype. I'm not too familiar with TDMS, but if I was writing to a plain text file, I would probably unbundle each piece of data, convert it to a string (format into string is great for this), then build an array out of that data, before feeding it into the "write file" VI.
