Screen Current in EL34 Trainwrecks

[Alexo]

Hi Folks,

I must confess that I still haven't built a wreck. But I am finishing up a build using EL34's at 425 volts (actually cathode biased, 450 on the plates minus 25 or so on the cat) hitting a 6K6 load with a choke+1K filter for the screen supply, and individual 1K screen R's.

According to the Mullard datasheet, this scheme should be good to go, and it seems to coincide exactly with what's in all the wreck schematics around here. But those old datasheet circuits tend to fall short in guitar circuits, as they didn't anticipate anyone blasting them into distortion, so the screen voltage or the load is often too high. I've witnessed the screens arc and flash in a couple builds as I wound them up and started to measure screen dissipatioon under heavy signal. usually the dissipation exceeds the rated limits and things go kaboom, then I scale back the screen voltage to a safer setting or halve the OT load via a different tap, etc..

Running the plate curves for EL34's at these voltages and loads, it sure as heck looks to me like I'm about to do the same thing. The 6K6 load line intersects the "knee" at a much lower point than I'm comfortable with, and I can't imagine any designer finding this rating acceptable. Here's a load line with the screens all the way down at 360 and it still looks dangerous, things will be even worse when they're running closer to 400. Er, actually I guess being cathode bias, they'll be close to 370 after all, considering the voltage drop over the screen R and subtracting cathode voltage. But it still looks dangerous.

And yet you trainwreck-ophiles seem to run your amps comfortably at these settings for decades.

End rant. Bottom line, my question is: How do the screens in EL34's hold up in your trainwrecks?

[Cliff Schecht]

From what you have drawn, everything looks fine. Yes, part of the load line runs above the max dissipation curve but your actual bias point is below this. No issues there. Your load line runs well below the knee of the curve but this is only in the class A portion of the swing. When it goes into class "B", you hit right at the knee of the curve which is ideal for our application. Again, no issues here. As far as the voltages, 425V ain't shit for an EL34 and 400V is well below the max screen voltage, even with modern tubes.

In class AB mode it's not uncommon to go past the actual rating of the tube. It's only when the tube is set to idle above the recommended dissipation that you start getting redplating and arcing/flashing. Your problems with arcing may not be due to bad design but bad tubes. The general consensus is that newer tubes just can't take a beating like the olds ones did and thus are no good for some higher power designs.

[Alexo]

Hey Cliff, come to think of it, much of that arcing was in a pair of China KT66's that would subtley redplate when idling at a meager 17 watts.

But I think my poorly labeled load lines may have confused you. The blue line is the combined A+B load line of the 6K6 curve, the red one is a 3K3 class B curve thrown in for comparison. 6K6/4 = 1650, 425/1625 means the class B portion would hit 250ma if the plate could hit zero volts, which hits pretty far below the knee. But the 3K3 load line hits the knee just right during class B, imo. But 3K3 looks a little heavy on the class B side of things for my liking to really want to use.

...unless I misread your post. your comments are reassuring either way though.

[Alexo]

Well I guess this is not the hot topic I imagined it would be, but finally having gotten around to this, I thought, for posterity's sake, I'd post my results:

Generally, under heavy signal, (i.e. starting to clip) the screen dissipation maxes out at around 7.5 watts (guess those Mullard engineers knew their stuff after all) but cranked into early square wave territory, there is a narrow range where they can peak at up to about 10 watts. But beyond that, the screen R's take over and they actually cool off as you increase gain.

Seems good enough for me - they also did not glow, which is more than I can say for some Marshalls. But this was all done with a 5U4GB in cathode bias, so you wreckers with diodes may be pushing your screens a little harder.

[Structo]

Well Alexo I find it interesting.

I hang in the Dumble section more even though I have built a 6V6 Rocket amp.

In my dabbling with tube amps the cathode biased verses fixed bias seem to be two different animals and can act very differently when the power amps are pushed to their limits.

There is also a mentality difference among guitar players.
I find the Marshall guys always want to turn all the dials to the max while guys that play more refined amps tend to be more conservative with their settings.
Especially if they are running NOS tubes or valuable sets of transformers.

I'm still learning and the plate voltage/ screen voltage and current demands of both are still kind of vague to me.
For example in my 50 watt Dumble clone, it seems that the 6L6 plate and screen voltage is very close to each other.
Especially after I installed a choke verses the power resistor it originally had.
So it doesn't drop as much voltage to the screens as before.
I have been tempted to put a resistor there in addition to the choke to see the effect of that.

[rooster]

alexo - For the experimenter in us all, have you ever tried to bring the screen voltage close to the plate voltage in an EL34? Turns out I did this in error here recently. This was a J/J E34L, BTW. Running a Class AB pair, I had only a choke with an ohmage value of 470 in place to reduce the screen voltages. From everything I have ever read regarding EL34 type tubes, this should have cooked the tubes. This was with a plate voltage of 440, BTW.

Well, tubes did fine, no problem, but it took out the flame proof 3 watt resistor serving as the choke in about 3 hours of hard play time.

....All of which leads me to sudddenly consider that, in some cases, a good amp design is more a consideration of how the parts supplying current might survive - and not so much the tube. Food for thought.

[Alexo]

Yikes! Well I would much rather have to replace a resistor than a set of EL34's. In my earlier days of designing amps, I arc'ed my share of power tubes because of the screen voltage, I was just guessing at what it should be, and since so many cool amps used a choke, I would too, much to my tube's consternation - some got lucky and the screens just glowed each time I hit a chord before I noticed and dropped their voltage down.

You can certainly do the choke thing if you design for it - you just have to choose the right load so as not to slam your screens and pull tens of ma through them. Higher screen voltages want smaller OT impedances and vice versa.

I've put the choke in series with a voltage dropping resistor, and it works really well, you still get the clarity and definition from the choke.

Honestly, I think screen current was a big part of the natural selection that led to the evolution of rock and roll guitar amps. If you crank up an Ampeg or some such amp that basically has a hifi output section, you will kill it.

You could run the screens as high as you wanted if you never cranked your amp into overdrive, and many example circuits from the datasheets are set up this way. Optimum specs for clean output, but the screens are sitting ducks for a meltdown.

[Alexo]

...one more thing: pushing your screens too hard can sound really good, you get some really nice singing compression that way. I suspect this was part of the classic JTM45 character, and why there are so few survivng KT66's around today.