Quick question about screen resistors

[yurka]

What is the appropriate power rating for 6V6 screen resistors? I have been seeing anything from 1W to 5W in schematics. Will 1W be sufficient? Also, someone had suggested bumping up the 470R resistors to 1K in a blackface style build for better reliability, thoughts?

[tubeswell]

1W will do for most 470R Rg2 for 6V6s. 2W if you want to be 'safe'. 6V6 screens only see a couple of mA at idle, and maybe a few 10s of mA instantaneously at peak times under signal conditions. That's my theory anyhow.

[vibratoking]

AFAIR, the 6V6 is rated at about 5mA to 7mA of screen current max. That it about 74mW of power disspation for a 1.5k resistor. A 1W resistor is more than 10x above the max for the tube.

[David Root]

Regarding 1K vs. 470R for BF screens, yes it's better protection. Higher values will also lower the screen voltage vs. the plate voltage which is a Good Thing.

I generally use 470R but in a 3W rating, in 50 or 100W builds.

If you check out a tweed circuit that has no screen resistors you'll usually find the screen runs about 2V higher than the plate. Of course new Tung-Sol 5881s weren't $330 a pair in the '50s so no-one worried too much about it until the narrow panel tweeds came out with 470R screen resistors to reduce failures.

EL34 needs higher screen resistors than 6L6GC because its screen is flimsier.

[martin manning]

I think the folks using 5W are doing so in the hope that the resistor will survive a screen grid short. That means they are also hoping that a fuse will blow before something more expensive than a resistor is taken out.

[pdf64]

Yes, I advocate 1 watt resistors for beam tetrode screen grids. A likely failure mode is a screen grid short, which can either blow a 1 watt resistor or, depending on various factors, a transformer winding.
Agreed that a resistor isn't a fuse; neither is a transformer winding, yet both can fuse, and so in the absence of an amp filled with a fuse for every conceivable failure mode, the use of sensibly specified resistors seems reasonable mitigation.
The criteria for 'sensible' being that the resistor wattage rating is specified such that it will be greatly exceeded by a likely fault current but that it won't blow with normal guitar amp abuse.

If you check out a tweed circuit that has no screen resistors you'll usually find the screen runs about 2V higher than the plate

There's a wide range of tweed designs with almost as many screen supply arrangements; some of the schemes that don't have individual screen grid resistors are the 5E3 and 5E5A.
The former has a 5k resistor feeding its screen supply node, very unlikely to cause screen grid overdissipation etc.
The latter feeds its plates and screens from the same node, which is very likely to result in screen grid overdissipation when the amp's overdriven.

My take is that the key to a 'kind' screen supply is that it has sufficient resistance between plate node and the screen grid terminal, 'sufficient' being around 1k for 6L6GC with VB+>450 and a 4k p-p per pair load.
For El34 that should increase to 1k5 minimum.
A problem with BF / SF Fenders is that the chokes used tend to be lowish resistance, eg <200 ohms; when overdriven, there can be excessive screen grid current if 470 ohm individual resistors are used. However, rather than using 1k individual screen grid resistors, better to use a higher resistance choke, eg ~500 ohms.
This is because individual screen grid resistors act as a source of local NFB, reducing gain but in a signal level dependent manner (because screen grid current tends to be very low at low signal levels then increases sharply at high signal levels. The NFB is due to the screen grid voltage being in opposite polarity to the signal at the control grid, and the bigger the screen grid resistor, the greater this will be.
That brings up the other aspect of the screen grid; it also acts as a control grid, and so good practice is to fit suitable grid stoppers at the tube base for all such grids; even if the choke is say 1k resistance, individual screen grid resistors should be fitted, maybe only 100 ohm, in order to mitigate for oscillation.
Pete

[jdw3]

Would the same apply for el84 types? I'm trying to decide what size screen resistors to use; it has a single 470ohm feeding both tubes.

Is the 470, combined with 2.2k at each tube, too much resistance?

[Firestorm]

Choosing the "right" screen grid resistance is a balancing act. The screen is "supposed" to lie at a fixed potential. With signal, the plate voltage will swing over a huge range; a positive going signal can pull the plate near 0 (actually 40 or 50 volts). This is part of why triodes sound how they do and why they may be biased differently. In tetrodes and pentodes, the screen insulates grid 1 and the cathode from that changing plate voltage so that plate current is no longer affected much by plate voltage. By connecting the screen right to the center tap as early tweeds did, the screen sits at pretty much the same voltage throughout the cycle and the tube can develop its maximum output. However, the screen, though quite small in surface area, does pick up some electrons so current flows in the screen. Too much current, not good. Screen resistors protect the screen from over current, but also cause the screen voltage to vary with signal (a little), which is opposite to its intended purpose. You want enough screen resistance to protect the screen, but no so much that you make the tube start to sound like a triode (unless that was your intent). Experimentation is likely called for. You can easily see if the screen is over dissipating; it will glow brightly with strong signal. So if you put a 470R on and the screen is still lightling up, try 590R or 680R. I think the screen supply dropping resistor in the B+ rail will have less effect on the screens' behavior (it's decoupled and has its own cap to supply current), so fiddling both the values of the dropping R and the screen R may get you where you want to go.

[pdf64]

See Aiken's findings on screen grid resistor values for EL84
Test number 1: Output power and screen dissipation
http://www.aikenamps.com/JJ_EH_tube_tests.html
Pete

[jdw3]

Ok, I installed screen resistors in my '70 Traynor BassMate, which burned up tubes and popped fuses.

I used 2.2K@2w. These are after the stock 470ohm resistor on the board.

The amp sounds great! A little quieter but not really any tonal change. I cranked it for a minute and the tubes didn't even quiver.

My amp was missing one of the two caps in the bias circuit. What could that have caused? I installed it. The amp basically sounds the same.

[C Moore]

FWIW.....
On an AC15 type amp I just built, I used (out of necessity) 2k7 1 Watt screens with JJ 84's.
best

[jdw3]

FWIW.....
On an AC15 type amp I just built, I used (out of necessity) 2k7 1 Watt screens with JJ 84's.
best

How did you come to 2.7K? Just trying wrap my brain around the math.

[C Moore]

FWIW.....
On an AC15 type amp I just built, I used (out of necessity) 2k7 1 Watt screens with JJ 84's.
best

How did you come to 2.7K? Just trying wrap my brain around the math.

I did not use any Math.
2.2 just barley got the screen voltage below the plates. The next value in my bins was 2.7.

[tubeswell]

Would the same apply for el84 types? I'm trying to decide what size screen resistors to use; it has a single 470ohm feeding both tubes.

Is the 470, combined with 2.2k at each tube, too much resistance?

There is no 'too much'. The bigger the screen grid resistor (Rg2), the less variation there is in screen grid current under signal conditions. If you use massive Rg2 (i.e.; like 1M), the pentode or tetrode will act like a triode (because the screen will no longer be acting as an effective electron accelerator). And in the 'somewhere in between' size of Rg2, the Rg2 acts to provide screen current feedback (which is kinda like cathode current feedback, but for screens), whereby at peak screen current times, the tube current is actually reduced and you get screen compression, which can sound quite nice, but it is obviously at the expense of gain (because you don't have maximum tube current).

The concept of having a screen in a tetrode or pentode is to ensure that there is a constant tube current. When the plate pulls more current, the screen pulls less, and when the plate pulls less current the screen pulls more. The Rg2 is there to stop the screen pulling too much current during the part of the signal cycle that is peak screen current. (Otherwise the screen would get into over-dissipation and fry up).

If you don't have an Rg2, then the screen is better off being set at a lower voltage than the plate under idle conditions. If the screen voltage is at the same (or higher) voltage than the plate, then it is prudent to have something in the way of a screen grid resistor for the reasons explained above.

[pdf64]

I used 2.2K@2w. These are after the stock 470ohm resistor on the board

My approach would be to increase the 470 ohm Screen B+ node supply resistor to 1k or 1k5, and use smallish individual screen grid stoppers, mounted on the tube bases, eg 220 or 470 ohms.
That minimises the local negative feedback, gain / power compression.

If you use massive Rg2 (i.e.; like 1M), the pentode or tetrode will act like a triode (because the screen will no longer be acting as an effective electron accelerator)

That doesn't sound quite right to me; my understanding is that pentodes can be used as triodes when their screen grids are strapped to their plate.
Rather a massive (unbypassed) screen grid resistor will tend to hamstring the tube, it won't be able to operate as intended due to all the NFB created.
As Vsg tends to zero, so does plate current, ie the tube gets turned off.
It would just act as a really weak tube, rather than a triode equivilant.
Pete