Originally posted by Ceej
... Any chance this can be salvaged or should I just deep 6 the entire set? I guess I could throw the three remaining ones ON Ebay as a matched pair with 1 backup??...

The tube with mechanical damage isn't even any good as a paperweight (it keeps rolling off the papers). The remaining three should probably get tucked away as backup for your new set.

The more I study tubes, the less convinced I am of the necessity of matching tubes beyond getting the same idle current at the same Plate and Grid voltages. My Ampeg VT-40, with the original pair of Magnavox 7027-A tubes, sounds every bit as good as my Ampeg VT-22, with a fresh quad of Groove Tubes 7027's in it.

While I was playing around with the VT-22 this weekend, (I just had to look into this bias thing), I pulled the 7027's and stuffed in a set of Sovtek 5881's. The difference was dramatic! Power dropped from 125 Watts @ 5% distortion, to just under 75 Watts. And the sound wasn't anywhere near as crisp or punchy. (The VT-22 is a good platform for this kind of experiment, because it has a 5 Ohm resistor on each Plate, which permits easy, though dangerous, Plate Current measurement). I spent some time tweaking the bias, and bashing away with my guitar, to find optimal settings. The 5881's seemed to be happiest with an idle current of about 45 ma with 586 Volts on the Plate. That implies about 26 Watts of Plate Dissipation. My data sheets list 23 Watts as the maximum acceptable value, so they might not have lasted very long in my amp. Ampegs apply voltage levels meant for seriously robust tubes (7027), so 'lesser' types tend to take a beating. For comparison, when I re-installed the 7027's, I biased them at 65 ma of idle current, with 586 Volts on the Plates. That's 38 Watts of Plate Dissipation, which would fry 6L6's, or EL34's, but I ran the amp at 70 Watts rms at 165 Hz, (E at the second fret of the D string), into a resistive load bank for an hour, and the plates didn't even change colour, much less glow red and threaten to melt! Balls aplenty!

So, back the Marshall's EL34 bias. The data sheets tend to agree that 25 Watts of Plate Dissipation is all they should have to deal with. Push-pull circuits allow you to use a large portion of this for the biased idle condition, because one set of tube(s) only conducts the audio signal for half of the time. So, let's assume 45 ma of Plate Current, and 500 Volts on the Plate. That implies 22.5 Watts of Plate dissipation at zero-signal (idle), with a little room left for when its actually doing something. If the Plate voltage is less than 500 Volts, the Plate Dissipation will be lower as well.