Hi MF,

Last night i've been at home while transmitting on the indoor loop
again. The TX power was doubled, i.e. about 2 mW ERP, based on Rik's
convincing calculation/estimation.
The temperature dependency of the antenna current was much more
expressed. I started at 70 % and moved to 100 %, then back to 70 %
whithin less than one WSPR TX period. 100 % is equal to about 20 A
antenna current.

As mentioned, the tuning C consists out of 2.2 nF + 6.8 nF both
WIMA-FKP-1 and a 1 nF variable capacitor.
Yesterday, it was clearly noticable that the 6.8 nF cap becomes warmer
than the 2.2 nF cap, although there were thermally coupled to each other.

The datasheet,
https://www.wima.de/wp-content/uploads/media/e_WIMA_FKP_1.pdf, tells
about the AC voltage derating curves, page 77, the 2000 VDC chart.
But they say "Permissible AC voltage in relation to frequency at 10 °C
internal temperature rise (general guide)."

I interprete this chart combined with the text beside it as follows: "If
you want to avoid a temperture rise of more than 10 °C, due to life-time
of capacity-deviation considerations, then you should not apply more
than x volts at a certain frequency"
and NOT like "In any way you must not apply more than x volts at a
certain frequency because otherwise the capacitor will explode and the
whole building will burn down immediately!"

Furthermore, the thermal connection of the leads to e.g. a ground layer
will have an effect on the temperature at the working point. So, is the
chart based on the worst-case or the best-case?
All in all i see no risk at all to apply the full 700 V rms AC level, at
least for smaller capacity values and if they have a good connection to
a heat sink that pulls the heat away from the internal plates.

Another thing in the resonated loop arrangement is the fact that the
current and voltage will drop as soon as the capacitor is getting
damaged or becomes to warm, so this is a self-regulating circuit.

I've just built the arrangement shown here:
http://www.iup.uni-heidelberg.de/schaefer_vlf/MF/MFcaps.jpg
It also gives 9 nF, but with a better distribution and heat transfer. If
necessary, a fan could be used to keep the temperatures down.

Will be tested tonite on MF. But 20 A at 475.7 kHz and 10 nF already
means 670 V, so i can't do much more here.

73, Stefan