Yes, pushing a 1969 to 75 pep is REALLY pushing it. It's not a mod I'd push on to customers, it's a mod that I did (as well as quite a few others) back in the early part of the decade for the keydown scene. They do NOT last forever at that level... HOWEVER, it is a selling point when you can actually watch a 5Kw slug wriggle with JUST a radio 
The invent of the RFX-75, MOSFET conversions, etc. pretty much pushed these types of mods to the back burner, but some people still question how they where done, or if they where actually EVER done.
The 25 watt level of the 1969 is pretty much pointless for class C high level modulation.... With 'plate mod', you can pretty much figure on 3-4 times the diss level of the transistor as PEP value of output. The 1969 in a 148 or other SSB chassis is run class AB.... In the 29 and associated 'xformer modulated' radios, it's run class C.
Your mathematics is correct for 100 percent modulated radios, that modulate symmetrically. With ANY negative clipping, compression, etc. that throws it out the window.... And you need TWICE the voltage to get 100 percent positive peaks... To get the 'textbook' modulation percentages from the 29, you need 26 volts of complete swing.... Twice the B+ to negative baseline... Or, optimally, to about 1 volt above baseline (which is what the circuit I presented actually does, prevents the carrier from EVER reaching 0, or the applied voltage, actually.... You ONLY prevent the modulator from doing it's job, really.... Hitting 0 volts on negative excursions)
Hope that helps explain it a bit better. NO, we are not getting linear output, BUT, we are 'plate modulating' a final amplifier, which means it HAS to run Class C. Class C is EASY to get 3 times Cdiss or Pdiss, 4 times even... 3 x 6146s, a 30 or so watt tube, will net 400 watts easily of plate modulated AM..
And let's not get started on the "controlled carrier' type radios... Like the DX60 and a few others.... Which is what NPC REALLY is..
--Toll_Free
