Hi,
second solution - adding 10k resistor providing positive voltage to Q5 drain works as well, I got back to 1nF C33.
My power output was still low (below 20Vpp) so I removed all of the final BS170, checked the drive - nice, square 5Vpp, it was a bit distorted with the finals in place.
Decided to mount a few fresh 2N7000s (bought 50pcs some time ago) instead of BS170s. With just one in place, I got about same power output (~18Vpp). Mounted another one, keyed down, a flat line on the oscilloscope - and 0.8A of current consumption! Even after unkeying. 7805 got very hot, turns out 74ACT00 is now about 2,5 Ohms Vcc-Gnd and gets warm after connecting power. The second 2N7000 I mounted was apparently dead (shorted gate-source).
So, few days off until new 74ACT00 arrives and I get the necessary utilities to unsolder the old one, I hope the uC and Si5351 survived without damage. Also, 2SC(...) PNPs I bought do not work well in pulse shaping circuit - pulses begin very sharply and end too slowly (>10ms).
I think I spent more time than I should building it ;)
73 Olgierd SQ3SWF
