You do not convert Eapp data to dwell time data. These are orthogonal and come from the raw data of your measurement.
First you identify how many "states" (different Eapp values from HMM search on your raw data). Now you collect the times spent in each "state" into dwell-time histograms.
Make sure you identify the beginning of the trajectory per a given molecule and do not count the first dwell time in the first state (because you do not know how long did it really stay in that state). Make sure you identify the end of a trajectory due to bleaching and do not count the last dwell time in the last state of a trajectory (you don't know how long it would have stayed in that state if bleaching wouldn't have occurred).
Specifically regarding FRET, try to collect dwell times of states for only those cases in which the sum of the donor and the acceptor fluorescence channels stays constant - if not, you might count quenching/blinking occurrences and not FRET changes.