Checking the indications of dead time length can take a variety of forms. In the various monitoring schemes described above, each switching event results in an indication of whether the dead time was too long or short (corresponding to a “1” logical value, for example) or not (corresponding to a “0” logical value, for example). Multiple switching events will thus result in multiple indications of these dead time values, which may also be thought of as multiple indications of whether zero voltage switching was achieved. The controller can determine whether the dead time should be changed based on any suitable analysis of this multiple set of indications. For example, the controller could adjust the dead time responsive to a predetermined number of consecutive indications that the dead time was too short. Additionally or alternatively, when analyzing “n” preceding events, the controller could adjust the dead time if a majority (n/2+1) of the preceding events so indicate or if a supermajority (e.g., ? or ?) of the preceding events so indicate. In some cases, the controller could adjust the dead time only if all the “n” preceding events so indicate. Thus, rather than control the dead time responsive to single switching events, the dead time may be adaptively controlled responsive to multiple preceding switching events. It is to be understood that this may result in some sub-optimal switching events; however, the net result can still result in improved control as the controller adapts to loading and other conditions to allow for dead times that achieve zero-voltage switching “most” of the time.