It was with great interest I read Xin Qi's article in the July 2012 issue of Analog Innovation. We have been prototyping a MPT charger, based on the LT3652. It works well on the bench and in field trials, but has its limitations. First the max current of 2A and second the relatively low efficiency of the non-synchronous design. We have measured 84,5% which is about the same as what is mentioned in the 3652 datasheet.
Would a charger based on LT3796 fare better? Has anyone measured efficiency data for the sample charger on page 8? I guess a 3796-based design has no limit on max current, other than the external components? Ie to push 4A, you reduce Rsns1 by a factor 4 and adjust M1, L1, C1 accordingly?