Screen Resitor Question

[plumcrazyfx]

So, on many boards I hear endless arguments about screen resistor values and how a 470ohm is better than a 1k and better than a 5.6k, etc. for sound.  I was looking at the single ended amps and they are using 10K (5F1, 5F2A), 25K (5C1).  Am I missing something?  Are they a completely different animal because they are SE?

I have a Univox U1226 that I need to get up and running.  I've heard that it runs the plates above 600v while running the screens in the 350v area.  Are there benefits to running the screens close to the plate voltage and is this only a benefit for PP?  Just trying to understand.

[VMS]

That 10k resistor is not a screen resistor. Screen resistor is the additional resistor you see for example in this ax84 P1 design:


http://ax84.com/static/p1/AX84_P1_101004.pdf

[plumcrazyfx]

I guess I'm wondering the point.  In your example the only thing coming off node B+2 is the screen and it is seeing the 10K + 1K from the B+1.  Why do it with two resistors?  I installed a screen resistor in my Champ build but if it's the only think coming from that node and the drop is significant (10K) from the plate node what the reason for the extra resistor?  And how does the significant voltage drop from plate to screen translate into performance?

[sluckey]

There's one additional thing coming off the B+2 node... the filter cap. The resistors in the examples you listed are B+ node dropping resistors. None of those amps actually have a "screen" resistor.

[plumcrazyfx]

So is an actual "screen resistor" only there to isolate the screen from the filter cap rather than to add additional dropping?

[VMS]

Little bit of info on bottom of this page:


http://www.valvewizard.co.uk/gridstopper.html

[plumcrazyfx]

Okay, so I can see the benefit.  So now how do the higher screen resistors impact sound?

[Colas LeGrippa]

Screen resistors are resistors inserted between the B+ and the screens (usually pin 4 ) of the power tubes to limit the current in order to preserve tubes life.  THE CONTROL SCREEN GRIDS are the SIGNAL GRIDS ( PIN 5 of an El34 , 6V6, 6L6, 6550 and so on ). Both grids need resistors. We call resistors for the screen grids''screen resistors'', and the ones for the control grids: ''grid stoppers'', and the last are intended to stop parasitic oscillation. A screen grid resistor of 500ohm or 1 K won't change sound of your amp. Even if they'd drop the voltage by 50Volts you prob wouldn't see any difference.

Colas

[plumcrazyfx]

I use grid stoppers (1.5k usually) but I was wondering what the point was of them in amps that already had the screen significantly lower than the plates.  I was looking at the 6G3 schematic and it was showing 365V on the plates from the output transformer and also showing 365V on the screens after a 1K dropping resistor.  In this case I can definitely see why you want them.  In a 5C1 Champ where the screens are running 20V below the plates due to the dropping resistor I wasn't seeing the benefit of adding a small amount of additional resistance that only separated the filter cap from the screen.  If that is the point to separate the filter from the screen during overdrive then I can see it.  Or if like the grid resistor it is because you want it really close to the tube to provide the most benefit then I see it.  I just didn't want to add it because everyone says so, I wanted to understand the why. 

[sluckey]

The screen resistor provides a bit of protection to the tube by limiting current in high current situations. The 5C1 isn't likely to get up into a high current situation, or so the original designers thought. Another reason you don't see a screen resistor on any of those amps you mentioned is because they are all the cheapest amps Fender ever built. That missing resistor meant more profit. These amps were built for the students, not the professionals.

I just built a cheap ass amp that doesn't have a screen resistor or a dropping resistor. I bought a pair of new guitar picks with the savings! 

[plumcrazyfx]

So add them because they are cheap insurance.  I can see that - I'm working on a build based on the 5C1 with an EL34 and was questioning whether I needed them.  It gets a little hair when all the way up but that's fine because I want it as a pedal amp and want to drive it with fuzzes and boosters.  I'll add the screen resistors and post a schematic when I am done with the tweaking for everyone to tear apart.

[sluckey]

I'm working on a build based on the 5C1 with an EL34

That would have been nice to know right up front. EL34s usually have a considerable higher screen current than those lowly 6V6 Fender amps. I would absolutely recommend a screen resistor (probably 1K/5W).

[plumcrazyfx]

I was working on that but it was more of a general question about their purpose.

[HotBluePlates]

I was working on that but it was more of a general question about their purpose.

You're asking a lot of questions that touch on a lot of different issues, though you may not have realized it.

The screen resistor provides a bit of protection to the tube by limiting current in high current situations. ...

The general purpose of a series screen resistor is to keep the screen from melting should the tube draw a whole bunch of screen current.

You usually don't have to worry about a "whole bunch of screen current" at idle; it happens when you push a lot of output power.

So is an actual "screen resistor" only there ... add additional dropping?

If there is a series screen resistor fitted, it will drop voltage based on its resistance & screen current drawn through it.  (Ohm's Law)

If the screen draws a whole bunch of extra current, the screen dissipates more power (Power = Voltage * Current)

If screen voltage is high, a little more screen current might exceed the screen's dissipation rating: maybe use a bigger resistor to drop the voltage.
If screen voltage is low, the anticipated screen current rise might not bump into the dissipation rating: maybe it's okay to use no screen resistor.
If the tube used has a screen current that greatly increases from idle to max power, maybe more voltage drop is needed: bigger screen resistor.

So whether a resistor is used and its value depends on the tube and supply voltage used, and anticipated worst-case conditions.

Okay, so I can see the benefit.  So now how do the higher screen resistors impact sound?

Normally, they are not picked to impact the sound.

Screen (G2) voltage impacts plate voltage just like control grid (G1) voltage, though it takes bigger changes of G2 voltage to change plate current.  See the top graph on page 6 of this data sheet.

If you have a very large drop of screen voltage from idle to max power, the possible plate current is constrained as you push towards that max power.  That's compression.  So some very large value of screen resistance constricts possible max power output.  For this reason, designers often use the smallest value of screen resistance that will adequately protect the screen, thereby avoiding any added compression or restriction of max output power.

OR...  they might add a large value of screen resistance intentionally to impart compression at some power level.

... I'm working on a build based on the 5C1 with an EL34 and was questioning whether I needed them.  ...

EL34's don't normally used aligned-grids and typically have a larger rise of screen current from idle to max output power than beam power tubes like the 6V6 or 6L6.  Therefore they usually have larger value screen resistors fitted.

[HotBluePlates]

... I have a Univox U1226 that I need to get up and running.  I've heard that it runs the plates above 600v while running the screens in the 350v area.  Are there benefits to running the screens close to the plate voltage and is this only a benefit for PP?  ...

Univox U1226 Schematic

This is a whole other situation.

The U1226 is a push-pull amp, and the screen voltage is ~1/2 of the plate voltage.  This is a special case where the designer tried to get a whole lot of output power from a pair of 6L6GC's by running a very high plate voltage (so large plate voltage swing with signal), and a high load impedance at the output transformer primary.  A convenient source of ~1/2 voltage was found in the voltage-doubler power supply, and the screens were tacked to it.  The ~330v for the screens is enough to allow the peak plate current needed at max power output.

Univox went so far as to tell you idle & full power voltage & average current for the screen & plate nodes.  Screen current rises from 20mA at idle (for both tubes) to only 26mA average current at full power (13mA per tube), while screen voltage falls to 320v.  320v * 0.013A = 4.16 watts, which is less than the 5w rating for this tube.  So they did not use screen resistors.

If you're worried about your 6L6 tubes (or that your modern tube might not meet the same specs as the G.E.'s in the sheet), you might add 100Ω or a bit more to each screen.

Screen resistors ... limit the current in order to preserve tubes life.  ... the ones for the control grids: ''grid stoppers'' ... are intended to stop parasitic oscillation. ...

Sometimes you will run into tubes that need a screen resistor to prevent parasitic oscillation.  These are "high slope" or high transconductance (Gm) tubes, though I've run into problems with some KT66's needing a screen resistor to prevent oscillation (though they have lower Gm than EL34's and some other tubes).

[jjasilli]

+1.


"Normally, they are not picked to impact the sound.Screen (G2) voltage impacts plate voltage just like control grid (G1) voltage, though it takes bigger changes of G2 voltage to change plate current.  See the top graph on page 6 of this data sheet.If you have a very large drop of screen voltage from idle to max power, the possible plate current is constrained as you push towards that max power.  That's compression.  So some very large value of screen resistance constricts possible max power output." 

KOC recommends large, like 1.5K - 2.2K or more screen resistors to "improve" the overdrive tone of EL84's.  I thought this sounded horrible and reversed the mod.  To me the resultant tone was overly compressed and restricted.

[92Volts]

The screen resistor doesn't just limit voltage/current during high average current, it limits peaks through the audio cycle. Every audio cycle, not just tube failures or other "extreme" events... though it has a bigger effect during high volume or overdrive.

In push-pull (class B) you may see 5ma screen current (per tube) at idle and 25ma (per tube) at full power (from an EL34 datasheet).

But you expect close to 0ma half the time as the tube passes less current than idle (potentially zero) as it gets a negative signal. To reach 25ma average it's pulling 50ma during the positive half of the signal-- maybe more at peaks!

So a 1k screen resistor burns off 50+ volts during peak instantaneous current. If you had a 1k resistor followed by a smoothing capacitor, it would burn off 25v due to 25ma average current, not seeing peak currents due to the cap.

That's why the protective effect is so significant, though a low enough supply voltage or a supply that sags enough under the load of both tubes will eventually eliminate the need for this protection.

Regarding running the screens closer to the plate voltage... it might increase power and could change the sound (for better or worse). Not at 600v though. 600v on the screens of any normal audio tube will kill it in a hurry. In fact, if I have the opportunity to run screens lower thanks to a voltage doubler or supply with an available tap, I like to run them around 350v max to prolong tube life.

[HotBluePlates]

The screen resistor doesn't just limit voltage/current during high average current, it limits peaks through the audio cycle. Every audio cycle, not just tube failures or other "extreme" events... though it has a bigger effect during high volume or overdrive. ...

Thanks for pointing this out!

Yes, for those unaware, the plate current peaks during half of each cycle and screen current can peak right along with it.  So large screen resistors can restrict the output tubes' ability to reach peak plate currents, as falling screen voltage restrains the tubes' capability somewhat.

That said, the role as protection from over-dissipation applies to the average screen current, and why I kept mentioning it.  The screen can have brief peaks well in excess of its rated dissipation, as long as the average current or average dissipation is within limits.

For that Univox schematic, 6L6GC screen current actually rises very much higher during peaks than you'd think from the 26mA indicated.  But those spikes are brief, and Univox measured/marked) the average current at max power (they had to do that to know whether screen resistors would be necessary).

[plumcrazyfx]

Thanks.  This really helps.

[wdelaney72]

I'd like to resurrect this thread and add in a scenario I'm stuck on at the moment.
Marshall 2203 100W build. I've done numerous incarnations of this amp and plexis, so I comfortable with the circuit. Amp is done. It works. It's dead quiet at idle and stable. Nothing wrong with it, per se, but I'm struggling to get it biased correctly. I do believe tube burn in will help or resolve this issue, but it did get me interested in considering increasing the 1K 5W screen grid resistors to 2.2K 5W.

For this amp I dug up an old 1203-80 PT from Metropolous that had the original 10xxx taps which produce 500vdc on the plates. I always had the best luck with =C= EL34 as they can handle anything. We all know, that's not really on option any longer. These days I've settled on EH 6CA7. I'm concerned I'll be eating through those tubes pretty regularly with the high high screen voltage from this PT.

I know I need to get the tubes and can caps burned in and settled before this is a real issue, but I cannot get the bias set properly. I have room to go on the pot adjustment but at a current of 17mA (or if you prefer voltage drop of -8vdc against OT resistance of 21.1 ohms) there's room to go. Where I'm now, the orange glow on the plates is pretty strong and it's hot enough to cook a fried egg.


Long winded question.....if I increase the screen grid resistors to 2.2K/5W, I know that will lower the screen voltage a bit and help preserve the power tubes....but will it hold the plate voltage in place and give me a better voltage drop reading?

[SILVERGUN]

I know I need to get the tubes and can caps burned in and settled before this is a real issue, but I cannot get the bias set properly. I have room to go on the pot adjustment but at a current of 17mA (or if you prefer voltage drop of -8vdc against OT resistance of 21.1 ohms) there's room to go. Where I'm now, the orange glow on the plates is pretty strong and it's hot enough to cook a fried egg.

No, this is a real issue now. If your plates have an orange glow you are biased too hot and damaging your tubes. You have to fix this before going any further.
Do you have 1 ohm resistors from cathodes to ground for measuring purposes?...if no, you should, it just makes life easier

Long winded question.....if I increase the screen grid resistors to 2.2K/5W, I know that will lower the screen voltage a bit and help preserve the power tubes....but will it hold the plate voltage in place and give me a better voltage drop reading?

No, the screen resistors are there to limit screen current which happens at higher output. Since bias is a measurement of current at IDLE you'll need to get that sorted out separately. Do that first. Don't pass GO. Don't collect $200. Do whatever it takes to stop burning those tubes up. It has nothing to do with screen resistors.

[pdf64]

I suggest that you fit a 1ohm cathode current sensing resistor per power valve cathode; it make measurements a lot safer, less prone to error, and eliminates the arithmetic.
8/21.1 = 379mA, so an average of 190mA anode current per power valve, if I understood things correctly?

[HotBluePlates]

... a current of 17mA (or if you prefer voltage drop of -8vdc against OT resistance of 21.1 ohms) there's room to go. ...

+1 to pdf64's comment.

500v x 0.017A = 8.5 watts, not enough to get the plate glowing.

I'll take your word you measured the resistance of the OT correctly, and the voltage drop across the OT (not the power transformer or choke wires).

Note that some see the glowing filament at the end of the cathode & think it's "red plating" or "a glowing tube."  I trust you're not making that mistake, so that leaves us with a math error somewhere.

Below is an example of severely red plating output tubes:

[wdelaney72]

Thank you all for the reply. So more detail.
I do have the 1 ohm resistors installed....that's where I measured .17mA


The voltage drop was measured with a Fluke 115. Combine that with the fact that all other measurements are with the meter are accurate in terms of plate voltage. I also used a a different meter for the voltage drop and have similar measurements. I'm measuring voltage drop with the OT Primary center tap (black) and the red probe on Pin 3. I've also measured the center tap OT and the plates directly to ground and subtracted and get the same voltage drop amount.


Here is a pic.....it's def more orange in mine than the near white light in the pic. I still think it's too hot. Interesting, though....the bias holds stable...if it were going to red plate, I would expect it to drift upward...it holds. That's not to say, though, that the tubes are going to last. Despite the readings....It seems too hot based on look and feel.


I've also tried a different set of tubes....same result.

[sluckey]

I do have the 1 ohm resistors installed....that's where I measured .17mA

I doubt you are measuring .17mA. More likely .17A from the look of the tubes in that pic. The life of those tubes can be measured in minutes, not hours, or days, or years.

You need to pull those tubes and verify the voltage on pin 5 of those tubes. Don't put any tubes back in until you have a proper voltage on pin 5.

[pdf64]

Regarding the measurement across the 1ohm cathode resistors, is your meter set to mA or mV?

[wdelaney72]

2 problems on my end. 1) Yes, I had the incorrect mA setting on the meter. 2) I'm messing up the Aiken math.


I don't always trust the mA reading on cheap-o meters, but I think it's good for now.


Here is what I'm measuring with the bias adjusted...yes, the plate glowing is gone and all looks healthy.
mA = highest 30.5mA
mv = highest 142.5 mV
OT Center = 445vdc
Left Plate Pin 3 = 439 vdc
Right Plate Pin 3 = 438 vdc

Resistance OT Primary Left = 20.6 ohms, Right 19.1 ohms (amp was cold...measured this first)
Pin 5 bias voltage can range from -10vdc through -35vdc...keep in mind this is a dual bias setup for 100W with (4) 6CA7.


Now....my burning question is....how does this truly tie out referencing Aiken's Math on Fixed bias Plate / Cathode Method?

[pdf64]

The point of the 1ohm cathode current sensing resistors is that they turn mA into mV.
To read mV, the meter must be set to mV.
An mA reading would be nonsense.

[glass54]

wdelaney72
2 problems for me:

Here is a pic.....it's def more orange in mine than the near white light in the pic. I still think it's too hot. Interesting, though....the bias holds stable...if it were going to red plate, I would expect it to drift upward...it holds. That's not to say, though, that the tubes are going to last. Despite the readings....It seems too hot based on look and feel.

When I have had random Redplating occur (usually Marshalls  ) the Bias supply rapidly heads for 0V or a much lower negative voltage.
And the second, when I check the 3rd page Marshall service manual on the JCM800/2203 100W Master Vol Amp (April 1981) the Bias voltage is indicated as typically -42V for EL34 (6CA7) at 470V HT. I'm concerned that you have a poor range in your Bias supply, I would like to see your range extend from say -20V to -50V. Anything smaller than -20V indicates poor Bias supply design or a problem in this 2203 Amp.
As indicated by pdf and Sluckey, we have the benefit of the 1R resistors and can make better judgements of what going on in the amp  .
Your Fluke 115 should give you reliable results if well treated.
Kind regards
Mirek

[HotBluePlates]

I do have the 1 ohm resistors installed....that's where I measured .17mA

You've got a single 1Ω resistor for 2x tubes?  0.17 probably = 0.17 volts = 170 millivolts = 170mA through the 1Ω resistor.

170mA / 2 tubes = 85mA per tube

500v x 0.085A = 42.5 watts per tube.  Yeah, you're destroying those tubes.


Whatever series resistor you have between the power transformer & diode Neds to be much smaller.  Like Sluckey says, you don't need to install tubes until you have at least -35v to -50v available at pin 5 of each output tube.

[glass54]

I'm a little more skeptical than that  .
Having fixed a few Marshall in my time (and pulled out a bit of hair!) I would use one 1R per tube!!
You get a better understanding and its also good to have 2 multimeters (one per phase of Push Pull if only one side redplates).
....sorry repeating, pull the output tubes and monitor you bias supply for stability.
Load it if in doubt, put a 220k resistor across it to ground, at the feed point to the two phases (junction of 2 x 220k and 56k resistors) will draw less than 0.25mA and make sure it is stable.
Kind regards
Mirek

[wdelaney72]

I do have the 1 ohm resistors installed....that's where I measured .17mA

You've got a single 1Ω resistor for 2x tubes?  0.17 probably = 0.17 volts = 170 millivolts = 170mA through the 1Ω resistor.

170mA / 2 tubes = 85mA per tube

500v x 0.085A = 42.5 watts per tube.  Yeah, you're destroying those tubes.


Whatever series resistor you have between the power transformer & diode Neds to be much smaller.  Like Sluckey says, you don't need to install tubes until you have at least -35v to -50v available at pin 5 of each output tube.

(4) EL34 tubes. (4) 1 ohm resistors...one for each tube. My one meter measures mA across the 1 ohm resistor at 30.1mA. If I use my Fluke for mV it shows .174....this is wheat I'm not following.


My bias supply is (2) parallel, separate circuits. Allows me to use different matched pairs. So yes, my -vdc is lower than what you'd expect, it's only feeding 2 tubes and there's an identical -vdc in parallel feeding the other 2 tubes.

[sluckey]

My one meter measures mA across the 1 ohm resistor at 30.1mA. If I use my Fluke for mV it shows .174....this is wheat I'm not following.

You CANNOT read current across the 1Ω resistor. You must set the meter to read voltage across the resistor. I would trust the Fluke. It's telling you that the tube is burning up! This is confirmed by the pic that SHOWS that the tubes are burning up. The clock is ticking. 

[wdelaney72]

I appreciate everyone chiming in. My experiment was to understand the impact of slightly dropping the screen grid resistor, as well as understand any variance between setting bias from the cathode resistor v. plate current as described by Aiken. My math (decimal) error took my down a bad rabbit hole. No tubes or children or animals were harmed in this experiment.


I first biased by mV across the 1 ohm resistors to 32.5. From there, I measured the voltage drop on each plate. According to the math with the left side resistance measuring 20.6 ohms....I would need to see a voltage drop of -1.3 to be at 32mA. The meter showed the drop was -1.06 vdc...so a little cold, but Id say within a reasonable margin. If I adjust bias to get the voltage drop to -1.3, the mv reading is 42mV.


There's no red plating....no burning smell any longer. All is right. Just gonna burn this in for a bit as it sits now an give it a test drive with volume after my wife is done working. Thank you , again for the input. I'm going to keep the screen grid stock at 1k 5W, since the bias is in no need of help and my math is now working as it should 

[sluckey]

The plate current will always be a bit less than the cathode current. Cathode current equals plate current PLUS screen current.