Talk to me about screen resistors...

[chris_deveau]

Granted, i may just be having a "dumb day" here, but i was just looking at the 18 Watt Lite IIB schematic and comparing it to the Sweetwood layout i built my amp from.

So i notice the two 100 ohm resistors coming off of pin 9 on each EL84. Now, i don't have these in the amp i built.

So the schematic says they are optional, but "for a little extra insurance". Insurance against what?

I am thinking of switching the amp over to 6V6s for giggles, and i understand that with 6V6s i would be wanting to make these two nonexistent resistors into 470 ohm resistors.

So, i have three basic questions.

1. What do these resistors protect against?

2. How would they be connected in the current Sweetwood layout i am using? I mean physically. Like i said i might be having a dumb day, but what does the board look like with these resistors in place? How do they connect to the 1k 3W that is hanging off the board and to pin 9 of one of the EL84s right now? Will i need to modify the board with another eyelet to make this happen, i.e. run the 1k/3w to a new eyelet then hang the two new 470 ohms off the board to the screens(is that pin 4 on a 6V6?)?

3. What wattage do these suckers need to be?

Thanks for anything you've got!

-chris

[Tof]

My guess is 100r is there to make sure Vg2 < Va (Vp) to controll the flow of electrons to the plate.

You can see it, the red wire do the rest to connect both pin 9 together.

[stoo]

A long read...http://www.webace.com.au/~electron/tubes/screens.htm
Stew

[zaphod_phil]

So i notice the two 100 ohm resistors coming off of pin 9 on each EL84. Now, i don't have these in the amp i built. So the schematic says they are optional, but "for a little extra insurance". Insurance against what?

I got those resistors put there in the Lite IIb schematic - they're originally from the Jason's mods - but the description is incorrect. What they do is to help provide some isolation between the two screen grids. Remember, the two EL84s are operating out of phase with each other. Each screen grid also has a signal appearing on it that looks quite similar to the signal on that tubes anode (aka "plate"). So without these resistors, each screen grid is attempting to feed an out of phase signal into the other tube's screen grid.

2W each should be fine. I'm afraid I don't know anything about Sweetwood's layout, but I'm sure you'll figure it out.

[clyde]

Some early Marshalls didn't use any screen resistors and some Traynors and other amps shared a common screen resistor. I've never heard of screens being out of phase and stoppers being used to correct this. Please
explain in a little more detail if you could Phil.

[Manoteal]

When you overdrive a power tube screen current goes WAY up. Without a screen resistor to drop the voltage under these conditions, not only would the screen grid try and burn itself up, causing premature tube failure, but the increase of plate current would also damage the tube. So when the tube is overdriven,(we love that sound!) having a screen resistor lowers the current in BOTH the screen and plate loads. I don't have mine in there for 'sound quality' I have it in there for tube longevity. I see that many amps don't bother with a screen resistor, and I can't say what these designers had in mind when they did that (saving money or selling tubes?). Perhaps they were not thinking of massively overdriving their output tubes like we do. Some have only 470 ohms or other low figures, but most designers in the DIY group use 1K. That seems to be a minimum, and you can go higher. I had a 1K, 2 watt at first and it got too hot, so I upped mine to 2K 5Watt and it is doing better. But it only gets hot when I crank it all the way up and play for an extended period of time. But I like to think that my tubes will last longer since they are being kept within their design parameters (2 watts max for screen grid). (This is for a single ended EL84 amp).
Manoteal

[Mooreamps]

So without these resistors, each screen grid is attempting to feed an out of phase signal into the other tube's screen grid.

and ah, in a push-pull stage, why would this be a problem ??

-g

[zaphod_phil]

Some early Marshalls didn't use any screen resistors and some Traynors and other amps shared a common screen resistor. I've never heard of screens being out of phase and stoppers being used to correct this. Please
explain in a little more detail if you could Phil.

YEs, and the Marshall 18W is a prime example of an amp that has both screens connected together, and the amp still works. However, what I'm saying is absolutely true, and quite well known. In fact there's a form of "self-split" PP power stage, where the grid of the 2nd power tube in the PP pair is driven from the signal on the screen grid of the first power tube. And it works pretty well. Putting some resistance between the two screen grids reduces some of that antiphase signal being applied to opposite screen grids.

in a push-pull stage, why would this be a problem ??

So in a PP amp each power tube is then working *against* the other to some extent. Now how much of a problem this is, I'm not sure. I would love to see a theoretical analysis of it. However, it's worth remembering that the vast majority of PP amps do use separate screen resistors, rather than shared ones. Tone Lizard also shows a similar screen grid arrangement in his Marshall Myths article - http://www.tone-lizard.com/Marshall_Myths.htm Although he has recently edited most of the explanation out which used to be there.

[Manoteal]

]in a push-pull stage, why would this be a problem ??

ZP said- "Now how much of a problem this is, I'm not sure. I would love to see a theoretical analysis of it".

Gary, can you help out here? I know you have a good theoretical answer to this.

Anyone else (amp designers) care to chime in?
M

[hucky]

screen grid operation is very complex in tube amps.
.... it is the place where tubewar is taking place.
even experts are fighting ... finally the screengrid has to save the amp, garant the sound while not burn the amp.

going ontopics while staying on the simple path:

1. the screengrid is turning a triode to a pentode (or tetrode)
2. it works like a second control grid but if it is going to higher
voltage than the anode it ist getting the anode and the
cathode current is flowing true the screen grid.
3. in the basic circuit the screengrid should always stay at lower voltage
than the anode and "burning" max. 2 watts

going to the lite IIb and the "shorted" screengrids:

1. the normally biased pentode (anode positive > screengrid negative)
is like a low impedance ac connection from anode to the screengrid.
2. in a pushpull amp the primiry side of the OT is working like a autotransformer. while the active tube is pulling down one side off the OT
the inactiv tube side is going up (higher than +B). even if the innactiv tube is not taking current the cathode and screengrid is going up and the
second (active) "shorted" screengrid is going positive (in the wrong direction)
3. in reality it is not going positive - but it is not going down as it should go.

jumping up to the 36w:

for same details but ohter reasons there should be 4 screengrid resistors on the 36w instead of only 2

next time we have to talk about the value of the resistor and where is the best place to connect to the powersupply, and the decoupling cap ... and how this is influencing the sound, overdrive distortion and tube life.
.... more subjects to come.

[Mooreamps]

]in a push-pull stage, why would this be a problem ??

ZP said- "Now how much of a problem this is, I'm not sure. I would love to see a theoretical analysis of it".

Gary, can you help out here? I know you have a good theoretical answer to this.

Anyone else (amp designers) care to chime in?
M

I've already posted the mathematics of how to calculate the screen resistor values of varies power tubes on another forum. I've read the article ZP mentioned, quite awhile back actually. The article does mention using a value of 5 K for the screen resistor for the EL34, which is pretty close to the value I calculate for a single-ended PA stage. But, I will take a few minutes to comment on this for a push-pull PA stage, and then I've got to get some work done today on my single-ended 18 watt build.

As far as my analysis of this point is certainly not rocket science, but I'll go in deep enough to comment on my point of view. Also, from my point of view, screen grid operation for an audio application is really pretty straight forward.

The thought of the counter EMF in the screen grid circuit of a push-pull power amplifier stage.

My first impression is, it can't really be an issue since any counter emf on one PA stage generated by the opposing PA stage would only serve to help limit the amount of screen grid current generated by that PA stage. Now, what would be the amount of counter emf generated ?

Let's see, for generally speaking, that screen grid current of a PA stage is approx 1/10th of plate current. If we know what the effective plate resistance is of PA stage ; biased class A ; then we can assume what the effective screen grid resistance is for that stage.

For example, let say a PA tube idleing statically at 40 milliamps with a plate voltage of 300 volts, the effective plate resistance is 7500 ohms. Then, lets say the effective screen grid resistance is 75K ohms. Now, lets apply a signal to the stage at 1024 cycles, and lets assume the effective PA plate and screen grid impedances are the same. The voltage of the signal at the first PA stage shall be the value of screen grid current times the resistance of the screen grid resistor. So, for sake of agruement, screen grid current is 4 milliamperes through 1 K ohm, gives a voltage of 4 volts rms. The driving impedance is approx 75 K ohms for the first PA stage, and the load impedance of the opposite PA stage is also 75 K ohms. This divides the signal voltage in half as seen on the screen grid high voltage supply rail. Now all we need to know is the signal impedance of the screen grid high voltage supply rail.

We used a signal frequency of 1024 cycles, which is on the upper end of the fret board. It's some note on the music scale, I'll think of it in a minute. The signal voltage is 2 volts rms with a combined effective impedance of 37.5 K ohms ; 75K ohms in parallel. Lets say the value of the screen voltage supply cap in the power supply is 22 uF. The capacitive reactance of this cap at 1024 cycles is approx. 7 K ohms. So, without going in too much more deeply, since 7 K ohms is a less impedance than 37.5K ohms much of this signal will be shunted to ground through the 22Uf power supply cap verses being applied to the screen circuit of the opposing PA power tube, with the ratio of signal shunt reduction of approx
1/5th. So, a counter emf of 1/5th of 2 volts will be seen at the screen grid of the opposing screen grid circuit.

So, what would be the effect? Since the counter emf is the signal frequency, and not power supply hum or noise, the effect is neglegible.

-g

[zaphod_phil]

..... any counter emf on one PA stage generated by the opposing PA stage would only serve to help limit the amount of screen grid current generated by that PA stage.

However, you have one screen grid swinging high and another simultanoeusly swinging low. This presents the opportunity for one screen grid to drive current into the other, which probably isn't too good for either.

So, a counter emf of 1/5th of 2 volts will be seen at the screen grid of the opposing screen grid circuit.

I like the approach you've used here for your analysis. However, looking at real world scope shots of an 18W November amp, we see signal swings of the order of 40V to 120V on its screen grids. Note the AC voltage readings at the top right hand corner of the scope screen. In this case there are 1k screen current limiting resistors on each EL84, so the two screen grids will have little effect on each other. But it just shows what can happen.

It should also be born in mind that it's generally recommended as good practice to have screen grid stopper resistors soldered directly to the tube socket pins to help deter parasitic oscillations.

[Mooreamps]

Which leads me to ask another question. If the signal voltages generated from both screen grid circuits are approximatly the same amplitude, yet 180 degress out-of-phase, would they not cancel each other out ? ; assuming the relationship of the impedance of the screen grid circuit in parallel with the ac reactance of the screen filter cap in the power supply.

-g

[Manoteal]

I wish I could comment more on the theoretical, but practically I do find a 40 VAC and a 40VDC as measured across the screen resistor, and that works out to about 40mA. Heck! The plate is only drawing that at static! I must assume that the plate current goes up as well in this overdriven condition, and we must make allowances for it if we are to run our amps this way. Now, Gary, you probably don't run your output tubes full blast since you have a Power Brake to limit current through the tube (as do many others). But persons like me that want the full on sound at reasonable volumes must pad down the output after the output transformer and thus have strange things happening inside that little tube. I notice that the cathode resistor gets hotter the louder I play, and that holds true for the screen resistor as well (which I have now at 2K, 5 Watt), (and isn't current constant in a Class A design?). I think not.
Theory is great for design, but there is something to say about what actually happens in practice. Am I the only one that has a hot screen resistor, or is everyone babying their amps and not pushing them into full overdrive?
Manoteal

[Mooreamps]

Hence, it's one of the reasons why electronic power scaling works so well. Except, with my circuit and with master PA volume controls, one can pretty much dial in the output volume and where the power tubes start to break up. You can play softly, and the tone is still fairly clean, until you strum a string hard and it cruches right through.

-g

[zaphod_phil]

Which leads me to ask another question. If the signal voltages generated from both screen grid circuits are approximatly the same amplitude, yet 180 degress out-of-phase, would they not cancel each other out

Yes, and no. If you sum the screen currents, there would be a cancelling effect, and you will see much less fluctuation with signal swing in the current being drawn from the PSU. However, if you look at the current fluctuations in each individual screen grid, those will be pretty large. In fact you will find that the screens grids are driving current into each other on opposite half cycles, without some form of current limiting resistor between them. The screen grids of most 18W amps get some pretty severe abuse, and it's quite common to see them glowing orange when an amp's being played.

[hucky]

I notice that the cathode resistor gets hotter the louder I play, and that holds true for the screen resistor as well (which I have now at 2K, 5 Watt), (and isn't current constant in a Class A design?). I think not.
Theory is great for design, but there is something to say about what actually happens in practice. Am I the only one that has a hot screen resistor, or is everyone babying their amps and not pushing them into full overdrive?
Manoteal

You just played the test where the theory is unspoken.
Most guitar amps are designed from old tube handbooks from companys like GE, RCA, Philips and so on. ... the orginal tweeds, the copys, tuneups and other boutiques.

Rarely you can find some serious hints and statements about screengrid handling in the old books. mostly they tell you how to go to reduce distortion.
But it will be very hard when not imposible to find something about screengrid current and screnngrid power in overdriven, overloaded guitaramps (single notes over 2 hours).
most current guitaramp designers are feeling happy to limit the screengrid
power to 10% of the anodepower while running a sinewave with no clopping. i am shure you feel that this is not the real story. if this calcul is matching real playersliv(f)e whe could build the screegrid resistor with
a powerresistor of maximum 1.2 watt.

Eddy VH has his own aproach to get the right tone out of his master 5150:
take a fresh GT quartett, pull 1 tube to get the unmatched SE sound and
repeat the same procedure every day. ... the slave amps are running in
standard PP and are not washing out the tubes every day.

My personal idea about screengrid handling is very offside off all the
current hot sounding guitar amps:
we have to control and run the screengrids @ much lover voltage
this will help to control softening the poweramp cliüüomg distortion
and powercompression
while not burning and washing out the poor musicians highpriced tubes
i get water in my eyes when is see musicians running highquality highpriced NOS powertubes in their hotrodet marshall burners ....
even for the high quality NOS the countdown has started when running out of specs all night long.
when we have burned all the NOS - we are obligated to rebias our grids
to get a good sound out of our amps by running new production tubes while not washing out the sound during the gig.
i hope to see some poweramp "damping" knobs on the future amps to control the screegrid fire while compressing the power.
... there is plenty of room for new builds
... thereis even more room for multiple patents for Randall Meza

... but first we have to make the diagrams, cool our 18watts, then Ranall can do the rest of the writing while putting new patent numbers on the catalogs.

long offtopics as normal - sorry for my bad writing
i do better in writing german, french, schematics, pascal, c++ ... than english

[Mooreamps]

Well, I've not yet brought up a push-pull power amp stage as yet. That will be my next build as I am wrapping up my SE 18 watt amp. I'll just have to cobble something up, and take some measurements on it.

The other point is also well made, my tubes don't get the chance to run hot when the power brake is dialed down. But I'll have to take those measurments when the amp is running at full power.

-g

[zaphod_phil]

My personal idea about screengrid handling is very offside of all the current hot sounding guitar amps:
we have to control and run the screengrids @ much lower voltage
this will help to control softening the poweramp cliüüomg distortion
and powercompression while not burning and washing out the poor musicians highpriced tubes

I agree! But screen current also has to be limited, as well as screen voltage. It's not enough just to keep the screen grids under 300V.

i do better in writing german, french, schematics, pascal, c++ ... than english

c++ !

[Manoteal]

In addition the the abuse in pentode-mode, I have my amp running triode and ultralinear at times. But I keep that screen resistor in there no matter what mode I am in, even if in triode you are "supposed" to connect the screen to the plate. My amp does always keep its screen voltage below the plate's. But it would be great if there was a 'rule of thumb' for screen resistor selection based on overdriven output tubes. For maximum power and tube life, there has to be a compromise, but I would err towards tube life at the expense of a little power loss.
Like Hucky says, when all the NOS tubes are gone, they will be gone forever. . .
M