THOR: SMPS guru help needed

["Marco Denicolai" <Marco.Denicolai-at-xxxxxxxxxx> (by way of Terry Fritz <twftesla-at-xxxxxxxxxx>)]

Hello all, again.

I need help from you SMPS gurus with the issue mentioned in my previous email:
slower charging profile after the SMPS warms up.

I have been "touching" inside my SMPS looking for components getting hot:
1. the transformer core and windings are cold
2. the transformer output full-bridge diodes (4 strings of 5 BYM36G diodes
connected as a HV bridge) get rapidly hot. This bridge rectifies the
transformer
output pulses to charge the primary capacitor with same polarity pulses.
See the
schematics at "http://www.saunalahti.fi/dncmrc/transf.pdf".
3. the driver IGBT heatsink is only slightly warm.

I can imagine two possible causes:

1. the voltage drop or the reverse recovery time on the diodes grows with
temperature: the manufacturer has no data about this (or I didn't find any)
2. the transformer + resonance capacitors resonance frequency drifts: then the
IGBT drive frequency is no more optimal and this slows down the charging cycle.
As the transformer core is cold the only possible reason is the resonance
capacitor bank (C1 to C12) getting hot and its capacitance drifting. The
bank is
made by 8 capacitors in parallel type PHILIPS 367 KP/MMKP 0.0047uF 2000V. Terry
tested those (3.20.2000) and found

PHILIPS 367 KP/MMKP 0.0047uF 2000V 1 amp 375kHz 9.4 Deg. C

Sadly I hadn't had a chance to "touch" the capacitor bank but I'll do it the
next time.

Anybody has got suggestions about the about problem?

Regards