Screen Voltages 5E3/6G3 Build

[syscokid]

Just put together a rough attempt at a more simplified 6G3. Omitted the tremolo circuit. Added grid stoppers and screen resistors. Cathode bias for the power instead of the 6G3’s fixed bias. I started with the 5E3’s 250R cathode resistor, but power tube dissipation was 140%. A 400R resistor changed that to 100%, with 30v dropped across the cathode. Plate voltage is 407v, but plate to cathode is 377v. Screen voltage is what has encouraged me to post this thread… It’s at 410v. Using a pair of JJ6V6S’s which spec at 500v/plates and 450/screens.

The following schematic is a copy of the 6G3 but modded with my preferences described earlier. The noted voltages are stock to the original schematic:

Fender Deluxe 6G3 MODDED Schem.jpeg

Amp sounds pretty impressive. Very strong lows and mids, breaks up easily and will not get harsh even with the tone control dimed. Very quiet and the set of Modulus Amplification trannies remain on the low side of warm.

[maxkracht]

Screen voltage being a bit higher than plate at idle isn't uncommon. Voltage will drop across the screen resistors when you play. JJs can handle some abuse anyway.

[cdemike]

Cool build! I agree with Max both that the JJ’s can stand up to that kind of abuse and that screens running higher than plates isn’t too uncommon. Still could be worth experimenting with some different configurations to see how it sounds with those screen grids running lower. First thing I’d try is using a voltage divider between the screens node and the screens themselves. Andy Le Blanc had a big thread exploring this as a solution to the same problem your amp has, and he suggested using a 1k/220k divider: https://ampgarage.com/forum/viewtopic.php?f=6&t=18855

[Colossal]

Are you sure your screens are being fed by B+2 and not B+1? With B+1 dropped across 1k (and then across your screen resistors), your screen voltage should be much lower. Then, going back to a hotter bias resistor like 250 or 270R, you should get something like 24V on the cathodes.

[syscokid]

Cool build! I agree with Max both that the JJ’s can stand up to that kind of abuse and that screens running higher than plates isn’t too uncommon. Still could be worth experimenting with some different configurations to see how it sounds with those screen grids running lower. First thing I’d try is using a voltage divider between the screens node and the screens themselves. Andy Le Blanc had a big thread exploring this as a solution to the same problem your amp has, and he suggested using a 1k/220k divider: https://ampgarage.com/forum/viewtopic.php?f=6&t=18855

Wow... Cool thread! Definitely will be trying the mod described here by Andy Le Blanc. Andy stated that the 220K bleeder should have a minimum of a .5W rating, but other replies have suggested to increase the minimum rating.

I copied John_P_ WI's screen voltage divider and augmented it with my notes in the green lettering, showing questions for R1 and R2, and the actual voltages just recorded:

IMG_1882.JPEG

[syscokid]

Are you sure your screens are being fed by B+2 and not B+1? With B+1 dropped across 1k (and then across your screen resistors), your screen voltage should be much lower. Then, going back to a hotter bias resistor like 250 or 270R, you should get something like 24V on the cathodes.

Good question and you had me verify! Yes, screens are fed from B+2. Using a variac to take measurements at a consistent 120 VAC, here's a more comprehensive voltage report just recorded:

IMG_1883.JPEG

[Helmholtz]

Just as with plate voltage the screen voltage limit applies between screen and cathode and not to ground.
A tube can't know what its pin voltages are wrt ground.

[maxkracht]

Andy stated that the 220K bleeder should have a minimum of a .5W rating, but other replies have suggested to increase the minimum rating.

If the voltage rating is good enough, .5W would probably be fine but 1W+ would be more reliable. It's not dropping much voltage except when the amp powers off. You can probably get away with exceeding .5w for the time it takes to discharge the caps, but if you have something bigger I would use that.

[cdemike]

IMO

Cool build! I agree with Max both that the JJ’s can stand up to that kind of abuse and that screens running higher than plates isn’t too uncommon. Still could be worth experimenting with some different configurations to see how it sounds with those screen grids running lower. First thing I’d try is using a voltage divider between the screens node and the screens themselves. Andy Le Blanc had a big thread exploring this as a solution to the same problem your amp has, and he suggested using a 1k/220k divider: https://ampgarage.com/forum/viewtopic.php?f=6&t=18855

Wow... Cool thread! Definitely will be trying the mod described here by Andy Le Blanc. Andy stated that the 220K bleeder should have a minimum of a .5W rating, but other replies have suggested to increase the minimum rating.

I copied John_P_ WI's screen voltage divider and augmented it with my notes in the green lettering, showing questions for R1 and R2, and the actual voltages just recorded:

IMG_1882.JPEG

0.5w is not adequate in my opinion. Part of this is bias on my end since power where I live is wildly inconsistent (wall voltage lately has been ~145V where I live for the past few days...), but I think it'd be best to plan on a safety margin of at least 2x. Power dissipation (P) is equal to current (I) x voltage (V); P = I*V. We know B+ at that node, so we know how much voltage we expect those resistors to drop, and we can use Ohm's law to determine I = V/R. Substituting V/R in for I, we get P = (V^2)/R.

So we have B+ = 410V and R= 220,000 (I'm ignoring the 1K; if you use 5% tolerance resistors anyway, the tolerance itself is larger than that resistor).

P = (V^2)/R
P = (410^2)/220000
P = 0.764W

So 1W is sufficient, but barely. I'd go with at least 2W (or two 1W 470k resistors in parallel) to get that safety margin -- less heat, less concern from a reliability perspective, etc. Your 3W 1K resistors should be adequate, so no need to change anything there, and if you have any spares, you could use that as the top leg of the voltage divider.

[syscokid]

Just as with plate voltage the screen voltage limit applies between screen and cathode and not to ground.

Thanks! That makes sense.

Andy stated that the 220K bleeder should have a minimum of a .5W rating, but other replies have suggested to increase the minimum rating.

If the voltage rating is good enough, .5W would probably be fine but 1W+ would be more reliable. It's not dropping much voltage except when the amp powers off. You can probably get away with exceeding .5w for the time it takes to discharge the caps, but if you have something bigger I would use that.

Thanks for this too. Ground them at one of the mounts of each socket?

Fortunately I just found a pair of 220K's in my stash, that I think are at least 2W rated. I can't remember what brand these are. Can anyone identify who makes these in the following resistor porn pic?

IMG_1885.JPG

[syscokid]

IMO

Cool build! I agree with Max both that the JJ’s can stand up to that kind of abuse and that screens running higher than plates isn’t too uncommon. Still could be worth experimenting with some different configurations to see how it sounds with those screen grids running lower. First thing I’d try is using a voltage divider between the screens node and the screens themselves. Andy Le Blanc had a big thread exploring this as a solution to the same problem your amp has, and he suggested using a 1k/220k divider: https://ampgarage.com/forum/viewtopic.php?f=6&t=18855

Wow... Cool thread! Definitely will be trying the mod described here by Andy Le Blanc. Andy stated that the 220K bleeder should have a minimum of a .5W rating, but other replies have suggested to increase the minimum rating.

I copied John_P_ WI's screen voltage divider and augmented it with my notes in the green lettering, showing questions for R1 and R2, and the actual voltages just recorded:
IMG_1882.JPEG

... I'd go with at least 2W (or two 1W 470k resistors in parallel) to get that safety margin -- less heat, less concern from a reliability perspective, etc. Your 3W 1K resistors should be adequate, so no need to change anything there, and if you have any spares, you could use that as the top leg of the voltage divider.

Perfect! Thanks.

[cdemike]

I'd avoid mounting them at the socket since the current flowing through the divider would make it into the cathode resistor, making bias measurements somewhat complicated. If you have room on your board, I'd mount them there and run a lead from the bottom of the divider to the ground lug either for the reservoir cap or the cap filtering the screens node. If you don't have room, a small terminal strip is easy to mount using one of the tube socket screws and keeps the grounding scheme tidy.

[maxkracht]

Fortunately I just found a pair of 220K's in my stash, that I think are at least 2W rated. I can't remember what brand these are. Can anyone identify who makes these in the following resistor porn pic?

I believe your resistors are vishay ccf series 2W


For pedantry sake: My assessment that .5w is probably fine is because i'm pretty sure iv'e seen 1/4w filter cap balancing resistors used in a commercial amp without issue. (was enough of a surprise to stick in my memory) I have definitely seen 1/2w many times. I typically use 2w resistors for this purpose. Considering the excess wattage is so brief, to the point the resistor likely doesn't get more than warm to the touch, a quality resistor should handle it. In this application, it's also very low stakes if the resistor should go high resistance or open. You just get a few more volts on the screen and maybe the amp sounds slightly worse. That said, I would still feel more comfortable with at least 1W.

[syscokid]

I'd avoid mounting them at the socket since the current flowing through the divider would make it into the cathode resistor, making bias measurements somewhat complicated.

I was implying grounding the 220K bleeder to a lug that is grounded at one of the socket's two mounting points and not to the socket's cathode lug.

Sort of like this?

Screenshot 2025-05-29 075831.jpg

[syscokid]

Just got through trying the 1K/220K screen voltage divider mod (Andy mod?), using my illustration from my previous post. At idle, screens are now 5 volts less than the plates, 398v and 403v respectively. All B-nodes dropped slightly less than one percent, except for B+2 which dropped from 410v to 403v.

Very subtle change in sound. Yes, it's a bit smoother and I don't seem to hear the instantaneous mild crunchy fizz when slamming full chords when the amp is cranked anymore. It was pretty loud in this 11 x 11 converted Mancave room. DB meter was displaying 115 dB's... Ouch!

I'll be replacing the mix of resistors for the cathode bias circuit with a proper single 400R resistor. I doubt I'll find a brown Ohmite in 400 ohms. I'll have to get one of those white sugar cube style ones!

IMG_1887.JPEG

Greatly appreciate the help I get here from this forum!