... was thinking of using 4 6550s of kt88s for a marshall build go get as close to 100 watts as i can. So i was sitting down and down a little math to kinda figure the tubes output biased at 70% just to get an ball park idea of what i can kinda expect to see. Here's what i got
6550/kt88 42w x .7 = 29.4w x 4 = 117.6w
I know that because of variations in you circuit, tubes, and power supply you won't actually be 117.6w. ...
Idle plate dissipation does not equal power output, especially in class AB amps.
Also, you should be able to get
200w of output from a quartet of KT88's. See the Marshall Major. The Ampeg SVT originally used a sextet of 6146's, but later switched to a sextet of 6550's for
300w of output (and often measuring more than that on the bench).
Best bet, skip the math and copy data sheet conditions, or copy a known guitar/bass amp. At least then, the heavy lifting of power output stage design is done for you.
If you really, really want to design it yourself, we'll need to start from (sort of) scratch and figure out all the details for the output stage.
Course, with KT77's, you could copy a 100w Marshall amp, use a quad and be all set.
... Plus i also read at the tone-lizard.com that you can put a 20mfd. 500v in parallel on pin 4 and increase your grid resistor to 5k 5w for a really cool sound. He said you'd loose a few watts but the sound was really good. I believe he called that running true pentode or something like that. ...
I have no experience actually using KT77's. Something I read recently that makes sense is that they are sonically between the 6L6 and EL34, having faster breakup than 6L6's and less compression than EL34's. This makes sense, because the "KT" series tubes are "kinkless tetrodes" which was the euro-name for beam power tubes (6L6's and 6V6's are beam power tubes, EL84's and EL34's are true pentodes).
I haven't read that part of the Tone Lizard site (is it back up and running now? was offline for a while), but I can only imagine he's applying the name to
suggest the resulting sound is different than normal operation of beam power tubes.
So the only critical part from what you report is the use of "too-big" screen resistors. A high-value screen resistor causes the screen voltage to drop when the tube draws more screen current, which happens during peak currents. The drop in screen voltage then limits plate current somewhat, and the net effect is a compressed sound.
This is something I suggested a few years back on this site, and I know Tubenit has experimented with using 3kΩ or so screen resistors. The added compression would typically only happen at high volume settings unless you get really crazy with increasing the resistance. 2kΩ, 5kΩ, whatever gives you the sound you want without limiting power output beyond what you find acceptable will work.