<< Checking the loop calculations using a formula from Helmut Bensch, DL4KCJ. pa0nhc 20160418 (C) The use, publication, copy and modification of all info on this site is only permitted for non-commercial purposes, and thereby explicitly mentioning my radio amateur call sign "PA0NHC", as the original writer / designer / photographer /publisher. (C)
 At the tuning capacitor, G0CWT measured the feed point impedance of a 10m circumference loop : on 7.1  MHz    22.2 Ohms. on 3.65MHz      2.4 Ohms.

According to DL4KCJ, the loaded Q of a loop antenna cannot be measured via its feed line (which transforms impedances).

 The loaded Q of a loop only can be accurately calculated using the formula : Ql = Xc / Zload

7.1  MHz : Ql = Xc(63pF)  / Zl(7MHz)  = 356 / 22.2 = 16.
3.65 MHz : Ql = Xc(240pF) / ZL(3.5MHz)= 182 /  2.4 = 76.

The voltage over the tuning capacitor at 100W power now is :
Urms = (SQRT(P x Zl)) x Ql.
7.1  MHz : (SQRT(100 x 22.2))  x 16 =  754 Vrms => 1063 Vp.
3.65 MHz : (SQRT(100 x   2.4)) x 76 = 1177 Vrms => 1660 Vp.

 At 66pacific (small loop calculations) these voltages both are calculated to abt. 4kVp. THIS PAGE GIVES NO CORRECT CALCULATIONS FOR 1/4 WAVE CIRCUMFERENCE LOOPS.

Check :
7MHz: I = SQRT(P / Zl) = SQRT(100 / 22.2) = 2.12 A   (flowing through the tuning capacitor).
Uc = I x Xc = 2.12 x 556 = 756Vrms => 1065 Vp. (not 4kVp).

3.65 MHz : I = SQRT(100 / 2.4) = 6.45 A
Uc = 6.45 x 182 = 1174 Vrms => 1655 Vp. (not 4kVp!).

Using 100W, a plate distance of 2mm is sufficient. Using 800W on 7,1 MHz, the capacitor voltage will be abt. 3kVp
Tested in humid weather, open cabinet, with 100W power NO arcing was noticed.
The 5W heather prevents condensation in a closed cabinet.