Today I made a first proper attempt at a QRP class E PA design for 136kHz. Although the values in the output circuit seemed to tie up well with theory I wasn't able to realise the expected high efficiencies, possibly because of the gate drive design. In fact I managed to blow up my last IRF510 because it got very hot and this meant aborting a WSPR test (at a sensible power level). One resource recommended to me is http://www.alan.melia.btinternet.co.uk/classepa.htm which has a E class design spreadsheet.
I also discovered today that 16mm diameter 3C90 toroids do NOT make good LF PA output bifilar transformers: they get very hot. The larger diameter 3C90s appear to be fine.
Overall, a frustrating day!