Adding Feedback & A Presence Control To A Gibson GA-8T (Cathodyne Phase Inverter model)

[One10]

Removed, wrong catagory. Was attempting to move over to 'Mods" and something went awry, trying to recover what I wrote, And thanks Loudthud for the info.

[loudthud]

This might be a case of trying to do too much with too little. First step is to measure the gain (power amp input to speaker) with all feedback and bypass caps on the 2.2K disconnected. Next, measure the gain again with the 2.2K fully bypassed with something like 22uF. This will tell you how much boost you might expect with caps across the 2.2K. Lastly, connect feedback to the 100 Ohm resistor with the Presence control disconnected. The gain difference between the first measurement and this one will tell you how much boost the Presence control can give. Select the feedback resistor for numbers like 10X gain difference or 20dB are high, but not uncommon. That much gain reduction might be too much and cause the amp to oscillate.

[Helmholtz]

It is often overlooked that adding global NFB lowers the power amp sensitivity.
So if the NFB lowers gain by a factor 5, the preamp needs to provide 5 times the input signal for full output.
If the preamp limits the signal at a lower level it might no longer be possible to achieve full clean output.

Could I see the schematic, please?

[Chuck H]

...

[One10]

This Diagram of a negative feedback with presence control is what I was planning/hope to do, as a similar GA-8T had a 47k feedback resistor on the O.T. + speaker lead & originally was wired directly to pin 8 of the Cathode (V1B). I would also like a presence control, but it cannot work due the cathode bypass (larger) capacitors.

The reason I am attempting this is I do not like the amount of distortion this amp makes on 4-5 vol., or even only 2-3, I can hear it. I like clear. yes, that's right, I want a cleaner sounding amp. boo..hoo.. The presence control is to brighten the top end, as it is too dark (I believe it is due to the cannabis rex? speaker).

(Gibson wired V1A & ViB the opposite of what they show in their schematic, among other anomolies - that is why I hadn't included it and made a crude drawing instead.)

(I could not recover what I wrote previously.)

[Chuck H]

This Diagram of a negative feedback with presence control is what I was planning/hope to do, as a similar GA-8T had a 47k feedback resistor on the O.T. + speaker lead & originally was wired directly to pin 8 of the Cathode (V1B). I would also like a presence control, but it cannot work due the cathode bypass (larger) capacitors.

The reason I am attempting this is I do not like the amount of distortion this amp makes on 4-5 vol., or even only 2-3, I can hear it. I like clear. yes, that's right, I want a cleaner sounding amp. boo..hoo.. The presence control is to brighten the top end, as it is too dark (I believe it is due to the cannabis rex? speaker).

(Gibson wired V1A & ViB the opposite of what they show in their schematic, among other anomolies - that is why I hadn't included it and made a crude drawing instead.)

(I could not recover what I wrote previously.)

This is tonally insightful. I did like the Cannabis Rex right out of the box. As it broke in it lost too much HF in critical chimey frequencies. It never sounded boxy or bland (like some Eminence speakers tend to after break in) but just not quite there. I did like it a lot new though. FWIW my Asian made Celestion V30 (which is well broken in now) sounds similar to the CR as it came out of the box. The Brit made V30's sound entirely different. So maybe venture into the Asian made V30's and suffer the year long break in period. That's how long it took for mine used in a 2xel84 amp. So certainly slower than if it was used in a higher wattage circumstance. There are "methods" for breaking in speakers that might be beneficial for loading a 70W speaker in a lower wattage amp and speed things up a bit. I haven't tried any of them. I don't play out anymore so I have the time

[Chuck H]

I did some CAD sims and found that an application of 100/1 feedback at V1B has a profound affect on gain. As to cap value implementing a presence control I don't understand the problem. Almost any capacitor value is availavle. So maybe you need a higher value than used in a typical presence circuit? So what? Do it anyway! This is your amp and it should sound the way YOU want it to. (<period)

Elevate V1B on a 500R pot wired as a variable resistor.

Send that node feedback from the output jack via a 47k resistor.

Based on your info so far I think a 470n cap wired across the pot as a presence circuit might give you what you want. If there is too much high mid there for you go to a 220n cap.

JM2C

[One10]

Thanks Hemholtz, I was looking into that also.
Chuck H, interesting that your speaker also seemed to lack high end. Maybe I will change the speaker in the future. I think I have a V30 stuck in another small amp, I could try. Yes, adding a presence control turned out to be much different than a Fender amp, as you see the Gibson is a little different. I will look for a 500Ω pot in awhile, still looking to where the topology for adding these parts to this style amp so I can know fully what is going on.
Thanks. I am also waiting on a cheap oscilloscope to arrive, my cathode tubed oscilloscope gave up the ghost (screen died).

[Chuck H]

Stacking the 500R pot under the cathode circuit of V1B does reduce gain somewhat on it's own. The addition of feedback reduces gain further. But as you've note there is enough gain here to clip the power tubes even with lower output pickups. In my simulations I interpret there is still ample gain to drive the power tubes to full power with the mods proposed.

[Chuck H]

Note that the "presence" circuit has the cap between the CT of the pot and the NFB node. This is opposite of the "Marshall" circuit that has the cap between CT and 0V. I can't say why but this arrangement doesn't make for a scratchy presence pot that the classic Marshall circuit is known for. It may be because the capacitor maintains voltage at both ends before the pot CT is fully grounded? YMMV.

[One10]

Hello Chuck H, Thanks for doing the simulations. That is insightful and interesting. I like the simplicity of what you laid out on the schematic and as I found out earlier, I like the stock 47k feedback resistor when I added it to my (w/o any presence circuit) amp. The amp was extremely loud with vol. turned up to 6.

1.) From spec sheets, I see that the 6bm8 should have more, or a lot more gain than the 6v6's? - Is how one could state it?

A.) 6BM8: Mutual conductance = 6400-7500 µmhos, a 73% to 100% more Mc than the 6V6GT.
6V6GT: Mutual conductance = 3700-4100 µmhos
and doesn't this affect, or determine how much feedback you can insert/use , or what is available, anyone?

2.) I am only familiar with the JTM45 Marshall classic circuit (close fender copy?), are you talking about the circuit with a cathodyne style or the LTP inverter?
I tried to adapt the feedback/presence control meant for a LTP inverter (online and they didn't state what type of specific circuit it was for) and not a Cathodyne inverter and it didn't work.

3.) I found out my other circuit (for LTP inverter style) only worked inserting it before the .02uf in parallel with the 2.2k resistor, on pin 8 of V1B.
I used RobRob's paralleling of a 4.7k & 5k-L (mine = 10k-L) pot w/a .1uf cap. wired to the wiper & ground. But since I am wanting to use larger capacitors (for a thicker sound), I believe the negative feedback, and/or presence control may be bleeding to ground with the 5uf cap in parallel w/the 2.2k resistor.

4.) I can try out your adapted schematic soon, going online to get a 500R Pot. and maybe some other value capacitors. For now I only have a 45.4 nF and when paired w/a 500Ω pot. - it should be good for about a 700Hz Fc. cut-off?
I may end up liking the presence control (if I can get one working) to be a little closer to a 1kHz cut-off.

Maybe I have this all wrong (from what I am trying to relearn so far), but isn't what really matters most, is that the ratio between the 'feedback' resistors be kept the same and not so much the actual values, except for the voltage divider resistor soon to be at the end of V1B, so as to not increase resistance much more with the 2.2k resistor and how it influences the behavior of the 12AXA7 and circuit as whole?




Thanks!

[Chuck H]

Yes the mu of the power tube matters WRT the feedback ratio because it does qualify as "loop gain". But since you're not shooting for high feedback levels found in, say, hifi gear to eliminate as much harmonic distortion as possible there is more leeway. Less is more for a guitar amp I think and too much is a bigger problem than not enough.

As to the specific cut off frequency for the presence circuit, just use your ears. Your 45n cap should be fine and still not as low as the knee frequency of the classic Marshall circuit. If you did want to get a lower knee and end up with an electrolytic cap I wouldn't worry about the signal bleed you mentioned. It should be so little as to not matter.

And regarding the ratio vs. the actual value for the NFB divider. For more familiar circuit you're correct. But since we're inserting the NFB at the cathode of V1B operating as a typical gain stage the value of the shunt resistor has a notable effect on gain. So smaller is better and then the NFB series resistor is calculated from there. Since we need a panel mount control pot we're sort of stuck with 500 ohms since that's about as low a value that is commonly available. And since some gain reduction was also part of the goal that should be fine. If you did want to bring the gain back up a little you could use something like a 4.7k series resistor and a 47 ohm shunt and then put the presence cap on a parallel 500r pot. But now with these values you'd need something like a 4.7uf electrolytic cap. Also this circuit has the side effect of not being able to turn the presence all the way off. Similar to the later Marshall presence circuit that parallels the 25k pot with a 4.7k NFB shunt resistor. So...

To keep from using an electrolytic in the circuit, as well as reduce gain a little AND keeping the circuit as eloquent as possible I came up with the circuit I posted.

Note that feeding a lead from the amps output to the preamp creates the potential for lead dress cross talk causing instability. So locate the 47k NFB series resistor as close to the OT connection as practical and route the NFB lead such that it's no closer to preamp components than it has to be.

[One10]

Thanks Chuck H, for the full explanation and detailed answer, and I can understand it!

Note that feeding a lead from the amps output to the preamp creates the potential for lead dress cross talk causing instability. So locate the 47k NFB series resistor as close to the OT connection as practical and route the NFB lead such that it's no closer to preamp components than it has to be.

Thanks, for saving me from a potential problem, I see now how it could easily be attached to a term. strip in use already( next to the O.T.)

Lastly, It looks like this will be the longest wire in the amp, (6 or 7+ inches) I thought about using RG174, but could only find RG316 (doesn't bend/curve well)? Not sure it will make enough of a difference here, although once I used some RG174 in a fender amp's V1 input area and it seemed to stop the local radio station from playing through it.

[Helmholtz]

I think for most effective NFB and presence boost the 2.2k cathode resistor should be fully bypassed (e.g. using a 22µ cap)..
A small cap will only fully pass high frequency NFB signal to the cathode and global NFB and cathode degeneration will fight each other.

[Chuck H]

I think for most effective NFB and presence boost the 2.2k cathode resistor should be fully bypassed (e.g. using a 22µ cap)..
A small cap will only fully pass high frequency NFB signal to the cathode and global NFB and cathode degeneration will fight each other.

Good catch. I couldn't really read that bypass cap value on the schem I posted and thought maybe it said "C2". But now I see in the proposed mod image that the actual value of .02 is indicated. And I agree that it chould be full bypass on the v1B cathode

EDIT: I just ran a simulation with a .02 cap for the V1B bypass and Helmholtz is correct that it takes all the punch out of the presence control. If the overall tone becomes too bassy with full bypass then a 1uf bypass cap seems sufficient to get more gratification from the presence control while still trimming some LF gain.