Gibson GA-40, GA-20 input pentodes variables, screen bypass, etc Help!

mountainhick

Tele-Meister
Joined
May 2, 2021
Posts
324
Age
64
Location
Rocky Mountains
Looking for help interpreting what's going on with the differences between these three iterations of screen voltage supplies.

The main things I am uncertain about:

1- I am not seeing how to calculate voltages for the screens in the second and third diagrams below. Looks like a two stage voltage divider, but if I calculate it that way, I don't come up with an appropriate voltage.

2- All three have a cap which look to me like AC blocking cap off the voltage supply to the screens.. Is this it's function? Does it also act in any way as a bypass?

3- I don't understand the relationship of plate and screen voltages in these circuits as they relate to gain and headroom. Any articles on this? or simple explanation?

4- Circuits 2 and 3 below had tremolo, which I removed from the schematics. They also had a 5 stage very steep cutoff CR network on the output to filter the LFO effect of the tremolo, I am not sure how this equates in terms of sizing a coupling cap. In series, the caps would total .001uf, but does that change with the resistors?

LFO-Filter.jpg


5- Any significant difference with the 411K vs 320K input impedances? The lower impedance seems to go with lower plate/screen voltages.


First diagram, "microphone" input on a Gibson GA-40:

Inputs 470K/ (3.3M+470K in parallel =411K) or .47 of the input voltage, 411K input impedance

220K plate and 1M screen resistors, (this I calculate as 47V at the screen) plus .05 bypass (blocking?) cap on screen,

1.5K Cathode resistor, and 20uf cap bypass

Ga_40_55_mic-pre-1.jpg




Second diagram, "instrument" input on a Gibson GA-40. This had a tremolo that connected to cathode and plate. This schematic has the tremolo and LFO filter removed.

both inputs same, upper input, 100K/320K, .76 of the voltage input. 320K impedance.
510K plate, 1M screen resistors, plus .25 bypass (blocking?) cap on screen
1.5K with 20uf cathode bypass

Ga_40_55_inst-pre-1.jpg



Case three: GA-20, also had same kind of tremolo removed from schematic.

inputs same 100K/320K, 0.76 of the voltage input, 320K input impedance.

A little bigger cathode resistor, 2.2K

ga20t-inst.jpg
 
Last edited:

andrewRneumann

Tele-Afflicted
Joined
Mar 22, 2020
Posts
1,570
Location
Cincinnati, OH, USA
Lots of good questions here.

1- I am not seeing how to calculate voltages for the screens in the second and third diagrams below. Looks like a two stage voltage divider, but if I calculate it that way, I don't come up with an appropriate voltage.

You've forgotten that the screen itself is drawing current. For the purposes of making calculations, I would estimate the screen is like a 250K resistor. Coupled with the 1M dropping resistor, you have a voltage divider that drops voltage by 80%.

2- All three have a cap which look to me like AC blocking cap off the voltage supply to the screens.. Is this it's function? Does it also act in any way as a bypass?

Blocking cap no. Bypass cap yes. The current the screen draws is not constant when you apply an AC signal. This screen current varies up and down more or less in sync with the AC signal. This will cause the voltage on the screen to fluctuate. It is exactly like a cathode resistor--it's out of phase with the input signal and is local negative feedback. To keep screen voltage relatively constant, we need to shunt this AC signal to ground. Another way to think of it is that the cap will supply the current to keep voltage constant, instead of forcing the current to come through the 1M dropping resistor which will cause voltage to fluctuate.

3- I don't understand the relationship of plate and screen voltages in these circuits as they relate to gain and headroom. Any articles on this? or simple explanation?

Without getting too far into the weeds here are a couple generalities:
1. Gain is mainly controlled by the choice of tube and the plate load resistor chosen. This is the same as a triode.
2. Input headroom (input swing before clipping) is mainly a function of screen voltage--the higher, the more input headroom is available assuming center bias conditions.
3. Output headroom available (maximum clean output swing) is mainly a function of plate voltage, but you have to screen voltage set for the plate load resistor in use. There is a lot of interplay here.

4- Circuits 2 and 3 below had tremolo, which I removed from the schematics. They also had a 5 stage very steep cutoff CR network on the output to filter the LFO effect of the tremolo, I am not sure how this equates in terms of sizing a coupling cap. In series, the caps would total .001uf, but does that change with the resistors?

I'm not sure... if I were you I'd just use a single 4.7n and hope for the best. If you are in the mood for trial and error, start with 1n and adjust up to taste. Someone could determine where the -3dB point is for 5 CR stages and then calculate a single cap with the same -3dB point. Possibly you?

5- Any significant difference with the 411K vs 320K input impedances? The lower impedance seems to go with lower plate/screen voltages.

Probably not. Pentodes are relatively immune from the Miller Effect. That's fancy talk meaning that the input capacitance is low. So you can have a relatively high amount of series resistance in the input without losing the high end of the signal.

I think those series resistors on the input are more about making sure you can have two instruments plugged in at the same time and not lose all of one instrument signal through the other instrument. Basically, these are mixing resistors and aren't really there to lower the input impedance, although they have that effect. That's my take on it--could be wrong. Most every input wants to have the highest impedance possible while still respecting the maximum resistance between grid and cathode (Rgk max).
 

mountainhick

Tele-Meister
Joined
May 2, 2021
Posts
324
Age
64
Location
Rocky Mountains
Lots of good questions here.



You've forgotten that the screen itself is drawing current. For the purposes of making calculations, I would estimate the screen is like a 250K resistor. Coupled with the 1M dropping resistor, you have a voltage divider that drops voltage by 80%.



Blocking cap no. Bypass cap yes. The current the screen draws is not constant when you apply an AC signal. This screen current varies up and down more or less in sync with the AC signal. This will cause the voltage on the screen to fluctuate. It is exactly like a cathode resistor--it's out of phase with the input signal and is local negative feedback. To keep screen voltage relatively constant, we need to shunt this AC signal to ground. Another way to think of it is that the cap will supply the current to keep voltage constant, instead of forcing the current to come through the 1M dropping resistor which will cause voltage to fluctuate.



Without getting too far into the weeds here are a couple generalities:
1. Gain is mainly controlled by the choice of tube and the plate load resistor chosen. This is the same as a triode.
2. Input headroom (input swing before clipping) is mainly a function of screen voltage--the higher, the more input headroom is available assuming center bias conditions.
3. Output headroom available (maximum clean output swing) is mainly a function of plate voltage, but you have to screen voltage set for the plate load resistor in use. There is a lot of interplay here.



I'm not sure... if I were you I'd just use a single 4.7n and hope for the best. If you are in the mood for trial and error, start with 1n and adjust up to taste. Someone could determine where the -3dB point is for 5 CR stages and then calculate a single cap with the same -3dB point. Possibly you?



Probably not. Pentodes are relatively immune from the Miller Effect. That's fancy talk meaning that the input capacitance is low. So you can have a relatively high amount of series resistance in the input without losing the high end of the signal.

I think those series resistors on the input are more about making sure you can have two instruments plugged in at the same time and not lose all of one instrument signal through the other instrument. Basically, these are mixing resistors and aren't really there to lower the input impedance, although they have that effect. That's my take on it--could be wrong. Most every input wants to have the highest impedance possible while still respecting the maximum resistance between grid and cathode (Rgk max).


Awesome reply thanks Andy!

Re: the screen and plate voltages, Am I hearing then that higher plate voltage means more headroom, and screen voltage is typically adjusted upward with the plate voltage?

Re: The coupling cap, I don't necessarily feel a great need to approximate the cutoff 5 stage filter, as I won't have the LFO in my version. I was thinking of starting with the same value as the top example of 0.1uf , then tweaking from there. 4.7N seems pretty small. I don;t think that a really steep cutoff frequency of 49 would have effect at 82hz, but there's rumor that the tremolo channels sound pretty different from the mic channels. So the exploration is comparing input voltage/impedance, plate and screen operating voltages, and whether that filter produces a different result than typical coupling cap value.

So another question then, I removed the Tremolo part of the circuit from these schematics, i hope my thinking is somewhat on target: the tremolo pulses voltage/current based on the oscillation of a 6SQ7, but i am assuming that the tremolo off condition would result in a neutral effect on the 5879. Here's one of these with the whole tremolo circuit. Is my assumption on, or just bunk?

Gibson_ga_40_55-trem-channel.jpg
 
Last edited:

andrewRneumann

Tele-Afflicted
Joined
Mar 22, 2020
Posts
1,570
Location
Cincinnati, OH, USA
Awesome reply thanks Andy!

Re: the screen and plate voltages, Am I hearing then that higher plate voltage means more headroom, and screen voltage is typically adjusted upward with the plate voltage?

Re: The coupling cap, I don't necessarily feel a great need to approximate the cutoff 5 stage filter, as I won't have the LFO in my version. I was thinking of starting with the same value as the top example of 0.1uf , then tweaking from there. 4.7N seems pretty small. I don;t think that a really steep cutoff frequency of 49 would have effect at 82hz, but there's rumor that the tremolo channels sound pretty different from the mic channels. So the exploration is comparing input voltage/impedance, plate and screen operating voltages, and whether that filter produces a different result than typical coupling cap value.

So another question then, I removed the Tremolo part of the circuit from these schematics, i hope my thinking is somewhat on target: the tremolo pulses voltage/current based on the oscillation of a 6SQ7, but i am assuming that the tremolo off condition would result in a neutral effect on the 5879. Here's one of these with the whole tremolo circuit. Is my assumption on, or just bunk?

View attachment 943511

Input headroom on an input stage doesn't need to be big, so raising screen voltage doesn't really get you much. If you want more output swing (ie gain), you'll need a larger plate load resistor. But then again, you're not going to be getting close to overdriving the input stage right? I guess I'm confused by what you are going for here. Are you familiar with plotting load lines? A lot of this becomes clear when visualized on a graph. We're bobbling 4 variables.

Those 5 cascading filters would really cut out the low end and even some mids. I modelled it in Falstad and got -3dB around 400Hz. That would sound very different compared to a single cap of the typical size. (About 400pF / 1M is the 400Hz -3dB filter of the typical RC type.)

I suck at tremolos. Never designed or built one myself. It's on my "someday" list to learn how those work and build one. But looking at the schematic, if the 6SQ7 wasn't oscillating, it would still be sending DC current through the cathode bias resistor. If you removed the 6SQ7, you would be changing the bias point of the 5879. So probably need to change that cathode resistor to keep everything the same voltage-wise. You'd also end up with more signal at the right end because you are removing the "mixing" resistors between the two stages. Probably no problem because you have a volume knob there--just turn it down a little to compensate. Or you could figure it all out and use a 500K volume pot or whatever to make it exactly the same.
 

mountainhick

Tele-Meister
Joined
May 2, 2021
Posts
324
Age
64
Location
Rocky Mountains
Input headroom on an input stage doesn't need to be big, so raising screen voltage doesn't really get you much. If you want more output swing (ie gain), you'll need a larger plate load resistor. But then again, you're not going to be getting close to overdriving the input stage right? I guess I'm confused by what you are going for here. Are you familiar with plotting load lines? A lot of this becomes clear when visualized on a graph. We're bobbling 4 variables.

Those 5 cascading filters would really cut out the low end and even some mids. I modelled it in Falstad and got -3dB around 400Hz. That would sound very different compared to a single cap of the typical size. (About 400pF / 1M is the 400Hz -3dB filter of the typical RC type.)

I suck at tremolos. Never designed or built one myself. It's on my "someday" list to learn how those work and build one. But looking at the schematic, if the 6SQ7 wasn't oscillating, it would still be sending DC current through the cathode bias resistor. If you removed the 6SQ7, you would be changing the bias point of the 5879. So probably need to change that cathode resistor to keep everything the same voltage-wise. You'd also end up with more signal at the right end because you are removing the "mixing" resistors between the two stages. Probably no problem because you have a volume knob there--just turn it down a little to compensate. Or you could figure it all out and use a 500K volume pot or whatever to make it exactly the same.

These Gibson amps use this as input/gain stage, then straight to the PI, and they do significantly overdrive. I am just trying to sort out channel design and chronological version changes related to their OD character just to figure a starting point. I'll use the 5879 as one channel, and close to 5E8 twin style 12ay7 for another channel. The Gibson GA-20 has this as well, though 12ay7 channel a little different to the Fender version.

I may use another triode or dual triode between these input stages for a better tone stack and to drive an EFX loop.

Here's another later version, nicely annotated by Rob! https://robrobinette.com/images/Guitar/HowAmpsWork/Gibson_GA-40_V3_Annotated_Schematic.jpg




Gibson_GA-40_V3_Annotated_Schematic.jpg
 
Last edited:

andrewRneumann

Tele-Afflicted
Joined
Mar 22, 2020
Posts
1,570
Location
Cincinnati, OH, USA
These Gibson amps use this as input/gain stage, then straight to the PI, and they do significantly overdrive. I am just trying to sort out channel design and chronological version changes related to their OD character just to figure a starting point. I'll use the 5879 as one channel, and close to 5E8 twin style 12ay7 for another channel. The Gibson GA-20 has this as well, though 12ay7 channel a little different to the Fender version.

I may use another triode or dual triode between these input stages for a better tone stack and to drive an EFX loop.

Here's another later version, nicely annotated by Rob! https://robrobinette.com/images/Guitar/HowAmpsWork/Gibson_GA-40_V3_Annotated_Schematic.jpg




Gibson_GA-40_V3_Annotated_Schematic.jpg

Ok I think I see what you are going for. I would assume that the input stage pentode does not overdrive, but that it supplies enough swing to overdrive the PI and/or the PI overdrives the power tubes. So all you need to do is make sure you aren't losing gain by removing the tremolo circuit. Am I right?

If anything, by removing the tremolo circuit and that cascading filter, you will be increasing gain of V1 and it's coupling circuitry. So no problem overdriving in my estimation.

Tone-wise--removing that filter is probably going to be the biggest effect on tone. If you like the tone of the tremolo channel, why not just leave the filter in? Find a tone stack simulator (I use https://www.falstad.com/afilter/ for quick checks like this) and take a look at the difference between different configurations.

I hope I've provided something of value here. If not, I am probably misunderstanding your questions.
 

mountainhick

Tele-Meister
Joined
May 2, 2021
Posts
324
Age
64
Location
Rocky Mountains
Ok I think I see what you are going for. I would assume that the input stage pentode does not overdrive, but that it supplies enough swing to overdrive the PI and/or the PI overdrives the power tubes. So all you need to do is make sure you aren't losing gain by removing the tremolo circuit. Am I right?

If anything, by removing the tremolo circuit and that cascading filter, you will be increasing gain of V1 and it's coupling circuitry. So no problem overdriving in my estimation.

Tone-wise--removing that filter is probably going to be the biggest effect on tone. If you like the tone of the tremolo channel, why not just leave the filter in? Find a tone stack simulator (I use https://www.falstad.com/afilter/ for quick checks like this) and take a look at the difference between different configurations.

I hope I've provided something of value here. If not, I am probably misunderstanding your questions.
We're on pretty similar wavelengths. The issue of gain is essentially the question (Also tonal differences between MIC and instrument/tremolo channels). So I am OK dropping the tone issue and focusing on gain for starters. I do want overdrive, but also want enough clear range to capitalize on the clear-breakup edge with that channel. The other 12ay7 channel is intended to be configured as the "clear channel" with some headroom.

The GA-40s apparently became more gainy as they evolved. Just want to determine a starting point to work from. The first circuit diagram in my original post, and the last full schematic show the microphone channels sans tremolo. I'll stick with this topography for now, and not muck it up with the tremolo considerations, then experiment from there. If it is great without the trem filter, I'll leave it, if not, I'll try that filter.

I can pretty much flip a coin in terms of the values between the first diagram, and Rob's schematic. Just wanted to explore the theoretical possibilities.
 




Top