Measuring Screen Grid Dissipation

[Gaz]

I've found the best way to drop screen voltage/current and keep the tone is to add series resistance with the choke, or replace with up to 2K resistor altogether. The large choke resistance is still bypassed, so not as compressed, and you can keep the individual resistors smaller this way. In an AC30, for example. 2K 'choke' resistor with individual 100R screen resistors sounds punchier than with choke and individual 1K screen resistors, and has lower screen dissipation to boot. RMS power is about the same, around 25W before clipping. I actually found raising the plate voltage is possible when you do this too (PT permitting), and I actually upped the plate voltage to 340vdc wrc from 300vdc and cooled down the bias a little to compensate. I get 30Wrms clean this way.

As you've intuited, it's the screens that are fragile, and tubes will last a long time the way I suggested. The can handle extra plate voltage as long as bias is conservative. JM2C.

[JazzGuitarGimp]

I've found the best way to drop screen voltage/current and keep the tone is to add series resistance with the choke, or replace with up to 2K resistor altogether. The large choke resistance is still bypassed, so not as compressed, and you can keep the individual resistors smaller this way. In an AC30, for example. 2K 'choke' resistor with individual 100R screen resistors sounds punchier than with choke and individual 1K screen resistors, and has lower screen dissipation to boot. RMS power is about the same, around 25W before clipping. I actually found raising the plate voltage is possible when you do this too (PT permitting), and I actually upped the plate voltage to 340vdc wrc from 300vdc and cooled down the bias a little to compensate. I get 30Wrms clean this way.

As you've intuited, it's the screens that are fragile, and tubes will last a long time the way I suggested. The can handle extra plate voltage as long as bias is conservative. JM2C.

Thanks Gaz. I have a nice assortment of 5W resistors coming in - either today or Monday. I think I'll go back to the individual 470R screen resistors and give a few different values of 5W resistors in series between the choke and the 470R screen resistors a try. This will be a quick mod that can easily be done while I await the VVR parts order. Does it make sense to add another filter cap after the 5W series resistor or will the 22uF cap at the output of the choke be sufficient?

Thanks again....

[Gaz]

Nah, in series with the choke before the screen filter cap will make the filtering even better. You can even try bypassing the resistor with a switch to see how audible the difference. I think it will be subtle enough to live with if you really care about those poor screen grids

[Littlewyan]

Just be careful what value you put there. Don't forget the Express has a 25W 1K Choke Resistor and its 25W for a reason. I measured that resistor under load and it was dissipating at least 10W. Its too hot to touch after the amp has been flat out for 15-20minutes.

The higher the value, the lower the dissipation. Fender used 10K there for a few amps and I think the resistor was only rated for 2W or something along those lines.

Edit: However the lower the value the lower the voltage drop so the dissipation drops again. I guess it evens out in a way. If you go above a certain value then you don't have to worry about dissipating as much heat, same goes if you go below a certain value.

[pdf64]

Good point but bear in mind that those tweed Fenders with high value droppers used beam tetrodes, which draw a lower proportion of plate current than the pentodes generally used in Trainwrecks.

[JazzGuitarGimp]

Progress! So, today I went back to the stock Fender screen grid stopper value of 470R, then I inserted a 2.7K 5W resistor between the screen node (output of choke) and the feed to the four screen grid resistors. And I have some pretty amazing results. Now, with a 285mV RMS signal applied at the EFX return (which is pretty far into PA clipping), I have screen dissipation of 2.08W. When I up the input waveform to 1V RMS, which is way into clipping, I am only dissipating 2.39W. But, here's the kicker: the clean RMS output power actually went up! Before the 2.7K feeder resistor, I was measuring 34W RMS at the onset of clipping. With the 2.7K feeder resistor, I am measuring 42W RMS. I did not expect this. As for sound, I can't say that it sounds any different with the feeder resistor. At reasonable volume levels, I was clip-leading the resistor in and out of the circuit, and could not hear any difference. I realize at high volume there will be some compression due to the new resistor, but I've been playing it well into PA clipping, and I am loving the way it sounds. I almost wish I hadn't ordered the VVR parts. But that will be an interesting extension of this test. I will report back with findings once I've built, installed and tested the VVR on the screens.

[Littlewyan]

Thats interesting. My valve theory isn't amazing but I did notice that on datasheets when the anode voltage goes above a certain point the screen voltage starts to get lower rather than be the same as the anode. Wonder why this increases output power.

[pdf64]

That seems weird, as the screen grid voltage should be lower, which would reduce plate current and so output power.
I suggest checking the findings, dummy load value, meter battery etc.

[JazzGuitarGimp]

That seems weird, as the screen grid voltage should be lower, which would reduce plate current and so output power.
I suggest checking the findings, dummy load value, meter battery etc.

Yes, the screen voltage node dropped by 10V at idle: was 369V, is now 359V. But, and I don't know if this would make a difference, the 6V6's are now idling a bit cooler. I had them at 70% (9.8W) before adding the 2.7K resistor. Now they are idling at 60% (8.4W).

[JazzGuitarGimp]

By the way, thank you, Gaz, for the od in the right direction.

[pdf64]

Yes, the screen grid voltage is very significant in regard of plate current, maybe more significant than plate voltage.
With such a big resistance to the screen grid supply node, I expect that the voltage will drop much more at high signal levels?
Unless you're aiming for a particular effect, I suggest that you reduce that resistance value, keeping it as low as feasible, just enough to keep the screen grid dissipation under control when heavily overdriven.
Maybe something in the range 500 - 1k would be sufficient?

[pdf64]

Double post

[JazzGuitarGimp]

Yes, the screen grid voltage is very significant in regard of plate current, maybe more significant than plate voltage.
With such a big resistance to the screen grid supply node, I expect that the voltage will drop much more at high signal levels?
Unless you're aiming for a particular effect, I suggest that you reduce that resistance value, keeping it as low as feasible, just enough to keep the screen grid dissipation under control when heavily overdriven.
Maybe something in the range 500 - 1k would be sufficient?

You are correct, when heavily overdriven, the voltage drop across the 2.7K resistor approaches 70V. At that point, the screen grids are dissipating 2.39W each. 6V6's are rated for a max of 2.2W, so I am already a bit over the max spec. Would it safe to push then harder?

[tubeswell]

Yes, the screen grid voltage is very significant in regard of plate current, maybe more significant than plate voltage.
With such a big resistance to the screen grid supply node, I expect that the voltage will drop much more at high signal levels?
Unless you're aiming for a particular effect, I suggest that you reduce that resistance value, keeping it as low as feasible, just enough to keep the screen grid dissipation under control when heavily overdriven.
Maybe something in the range 500 - 1k would be sufficient?

You are correct, when heavily overdriven, the voltage drop across the 2.7K resistor approaches 70V. At that point, the screen grids are dissipating 2.39W each. 6V6's are rated for a max of 2.2W, so I am already a bit over the max spec. Would it safe to push then harder?

Did you subtract the 70V from the screen node HT voltage when calculating that screen dissipation?

(And remember that the peak dissipation is not the average dissipation under maximum signal conditions)

[Littlewyan]

I just want to point out that I corrected my earlier post about the resistor value. Don't want to leave incorrect info in a thread.