Screen resistor values for cathode-biased EL84 (1974 variant)

[PharmRock]

18-watt 1974 circuit with an EF86 vs. 12Ax7 in the normal channel.

Various schematics show a shared or separate 100R or 1K screen resistor(s) for the EL84. The original schematic has a shared 100R. https://el34world.com/charts/Schematics/files/Marshall/Marshall_18watt_schem.pdf

According to Merlin:
"The screen-grid of a power pentode should also have a grid-stopper. This protect the screen from over-dissipation when the valve is overdriven, which causes the average screen-grid to draw more current. A screen-grid stopper limits the screen current to a manageable level, hopefully saving the screen grid from melting. This resistor should be at least 1k ohm, 2W or more. Some amps use 470 ohms, but I know from experience that this is not always enough."

In the case of the 1974 18-watt, is this a case of where there is more than one right answer?  My board is laid out for separate screen grid resistors but I can easily make it a shared resistor with a jumper. 

For what its worth, I may (down the road) implement "Geezer's power control" https://el34world.com/Forum/index.php?topic=4401.0
If I go that route, would it be best for each EL84 to have its own screen grid resistor, since I will be splitting the cathode cap and resistor for each tube?

Thanks.

[sluckey]

I used a shared 100Ω on my 18W just because the original used a shared 100Ω. Geezer's power control don't care how you power the screens.

[PharmRock]

Thanks...wasn't sure if I would need each EL84 to be completely "separated" if I went that route.
I'll stick 100R's in there for now. 

[HotBluePlates]

... would it be best for each EL84 to have its own screen grid resistor ...

Thats a question that YOU answer.

You consider what benefit you get from individual resistors vs a shared resistor.  You consider what benefit a large vs small resistance gives.  You assign a relative importance to these different factors, then YOU decide where exists the best compromise among the competing requirements.

According to Merlin:
"The screen-grid of a power pentode should also have a grid-stopper. This protect the screen from over-dissipation when the valve is overdriven ...

Depending on your terminology preferences, Merlin is "mixing metaphors" here.

We might normally think "grid stopper" to describe the resistor that rolls off very high (RF) response by interacting with the tube's input capacitance.

We might think "screen resistor" for the purely protective function that Merlin describes.

Really, these are one & the same resistor, which means we should think about the different functions performed by this single part.  That will guide us towards whether there should be individual- or shared-resistors, whether the resistance must be right-at the socket or can be positioned elsewhere, etc.  Again, you're making a decision among competing tradeoffs here.


FWIW, when Groove Tubes came out with their KT66 tube, I had a problem with a pair that tended to oscillate.  It sounded much like we would expect from a microphonic preamp tube.  The solution wound up being to add a screen resistor (in grid-stopper role) right-at each output tube socket in that amp.  That amp did not have any screen resistors before, and hadn't needed help with any other 6L6-type I had used up to that point.

So in that case, the screen grid's role as "an input that results in a plate-current variation" became relevant, where it hadn't been as important with other output tubes.

[PharmRock]

Thanks HBP!
Yeah the goal over the past couple of years has been for me to figure out more of the "why" behind these changes.  I've been reading up some more and came upon this interesting thread with some posts/replies by tubeswell: https://www.tdpri.com/threads/question-s-about-el-84-screen-resistors-yep-thats-right.953533/

In your last statement about the screen grid/plate current variation, that is maybe where I was thinking, sound-wise, there would be an advantage of going with smaller values (e.g. 100R) vs. the higher values, like 1K.  Perhaps the response/feel of the amp would change.  Again, still trying to take in all this stuff and make sense of it.

A post I read (can't remember where, but I think it was Kevin O'Conner) said that having dedicated (vs. shared) screed grid resistors would limit collateral tube damage if one tube had a failure.  If that is the case, it seems like a good approach to have.

As to the physical location, I have the resistors mounted on the board vs at the tube sockets, similar to how many of Doug's layouts are. 

Once built, changing resistor values and taking voltage readings and trying to discern any audible differences in tone/response will be relatively easy.  Theoretically I might be able to say "100R sounds better than 470R, which is better than 1K", or see clear differences in voltages.  Granted, I probably won't be able to still figure out the "why" completely.  I suspect the answer may lie in the ability to read load lines and such (as pointed out in the linked thread, above), but I've got a long way to go before I am comfortable with that.  Still crawling up the learning curve.

Thanks again.

[PRR]

I doubt we will live long enough to try ALL the combinations of resistor, voltage, and style, to know all the tradeoffs.

We know from TV research that an over-driven power pentode lives longer with some resistor at some supply voltage depending on design balance choices inside the tube. But is 10r enough? Is 10k too much? It seems (for our tubes) to like something inbetween, but that is a wide area. And there may be no "perfect" or "magic" value.

[HotBluePlates]

Once built, changing resistor values and taking voltage readings and trying to discern any audible differences in tone/response will be relatively easy.  Theoretically I might be able to say "100R sounds better than 470R, which is better than 1K", or see clear differences in voltages.  Granted, I probably won't be able to still figure out the "why" completely.  ...

The EL84 data sheet doesn't make it as easy to see as some other sheets...

Scroll to Page C2; this is "screen voltage of 250v."  Look at the "Vg1 = 0v" curve, and see where it crosses a plate voltage of 100v:  140mA.

Now scroll to Page C4; this is "screen voltage of 300v."  Look again at the "Vg1 = 0v" curve; it crosses a plate voltage of 100v at 160mA.

When screen voltage dropped 20% (300v to 250v), plate current dropped 14% (160mA to 140mA).  That would be "compression."

Wil a 470Ω resistor cause much compression at 0v on G1, with the Plate at 100v?  Turn to Page C3.

     We might idle with bias of -10v at 300v on the plate:  G2 current is 2mA.
     We have a peak at 0v bias, 100v on the plate:  G2 current is ~24mA.
     ∆ G2 Current = 22mA, and 470Ω * 22mA = 10.3v.

We might anticipate plate current compression due to screen voltage-drop will be reasonably small.  The "IF" built in here is that I just pulled some example numbers & we would need an actual loadline to know the peak plate current (and the plate voltage at which it occurs).  Those facts would help us use the graph on Page C3 to estimate what screen current will do.

In your last statement about the screen grid/plate current variation, that is maybe where I was thinking, sound-wise, there would be an advantage of going with smaller values (e.g. 100R) vs. the higher values, like 1K.  Perhaps the response/feel of the amp would change.  Again, still trying to take in all this stuff and make sense of it.

Short version of the above:  0Ω is "no induced compression" while "many-kilohms" is "much induced compression."  Historically, manufacturers used the smallest resistor that gave the screen adequate protection so as to avoid more-compression and fewer watts of output at the loudest settings.

Play the amp quietly, and you almost certainly won't hear any difference.

[PharmRock]

This is great info...thanks for taking the time to walk me through the thought process on this.  It certainly helped a lot , especially with connecting the qualitative and quantitative aspects of what is going on.  Your "tutorials" on these subjects are excellent.

I'll continue to look this over and when I get this thing up and running, experimentation with the resistor values (say 100R, 470R, 1K) will happen.  I think I will start with 100R, since this is a tried and true value according to not only the original schematic but also many of the variations circulating around.