The "huge cathode cap method" on 18watters... does it work?

[Silvio55]

Hi everyone! Iīve been tried to solve the cross over distortion problem on my 18watt lite build I finished three weeks ago. I first did the "Paul Ruby mod" and it helped a lot, along with changing the cathode resistor on the EL84's from 150 to 120 ohm (biasing hotter). There still some buzz and I searched on the web some other method that could help with it. In the 18watt site, I read about the "huge cathode cap" method, consisting in changing the bypass cap of the cathode resitor on the power tubes for a big value one, like 1000uf or 2200uf, this theoretically reduces the voltage swing across the cathose resistor, so the bias wonīt cool that much, reducing the cross over distortion.
Today I tried that method, I changed back the cathode resitor to 150, so the bias won't get that hot, and put a 1000uf cap, I meassured the voltage swing and it was larger than before, with the 120R and the 47uf cap it was from 10V at idle to 15V at full signal with the guitar (hitting the strings real hard!), now it was from 11V to 18V, and the crossover distortion buzz was worse than before, so I tried a 2200uf cap and got the very same results. I cheked everything for some error like bad solder joints, reversed polarity in the caps, etc. but it was all OK.
Has somebody else here tried this method? Or does it really don't work at all?

[MWJB]

The voltage swing rose because you biased cooler, that's also why the crossover buzz was worse. After a point the size of the cathode bypass cap makes little difference...usually after a hundred, or two uf.

What plate voltage & idle dissipation do you have?

Are you 100% sure that you are suffering from crossover distortion specifically, rather than any other artefact?

[Gaz]

From the man himself:

AX84.com - The Cooperative Tube Guitar Amp Project

[Silvio55]

The voltage swing rose because you biased cooler, that's also why the crossover buzz was worse. After a point the size of the cathode bypass cap makes little difference...usually after a hundred, or two uf.

What plate voltage & idle dissipation do you have?

Are you 100% sure that you are suffering from crossover distortion specifically, rather than any other artefact?

Yes, I knew that with a cooler bias the cross over distortion is worse, that's why I switched the cathode R to 120R first, but with the big cap it was supossed to reduce the voltage swing, so it'll keep the bias hotter when pushing the tube hard, so if I would keep the 120R resistor, the bias would become too hot, thatīs why the larger resistor, and thatīs what people recomend at the 18watt site.
My plate voltage is 344V, 10V at the cathode and 14W dissipation at idle with the 120R resistor.
I think itīs crossover distortion, but could be something else, I have another amp a Vox Nigh Train and the Paul Ruby mod helped a lot, almost no buzz at all, but this amp still sounds a bit buzzy or fizzy with the PRmod, it sounds similar to the crossover dist. buzz but less notorious, I donīt know maybe itīs the OT?
The thing that drives me nuts is the fact that I have read all over the web that a big cap reduces the voltage swing, everybody tell you that, even the Avatar 18Watt amp uses the huge cap, but I did it and itīs not true, unless I did something wrong, and I doubt it.

[frus]

but the big cap doesn't reduce the voltage swing, it only slows it down
like increasing the "attack" time on your compressor
no cap can resist the voltage change forever, you would have to have an infinite capacity. and it would only reduce the circuit to fixed bias, which can be achieved using much simpler means

btw, the original 470uF cap is also quite unusually big, most amps have like 22 to 47uF

[Silvio55]

but the big cap doesn't reduce the voltage swing, it only slows it down
like increasing the "attack" time on your compressor
no cap can resist the voltage change forever, you would have to have an infinite capacity. and it would only reduce the circuit to fixed bias, which can be achieved using much simpler means

btw, the original 470uF cap is also quite unusually big, most amps have like 22 to 47uF

Well, reducing the voltage swing was what people said it would do, but of course it isnīt what it really does. I know the zener across the cathode resistor method, but now I donīt have 5W zeners to try it.
The original cap I used was 47uf, not 470uf.

[frus]

I was referring to the original Marshall value (470uF), which can already be considered a "big cap", compared to most of the other amps

[es345]

This the way I have solved it in an 18W amp (see annex)

- step 1: 2 separate cathode resistors- that protects the second pentode if the first is dying
- step 2: "switchable" caps:
closed: normal mode as known, I have used it playing clean without overdrive
open: creates a smooth clipping avoiding crossover distortion.
You loose 3-4W - from my experience negligible

[Chuck H]

I don't know why removing the bypass cap stops crossover distortion. Is it because crossover distortion is more analogous between the tubes and therefor cancels when not bypassed???

I'm wondering if Silvio55 is actually hearing diode clipping atrifacts, as suggested above. The values for the PR mod are amp specific. If we had a schematic with voltages...?!?

[es345]

lets run through different operating modes with clean switch open. In this mode the system works partially as if you would have one single cathode resistor with a cap and partially without a cap.

operation mode 1:
small signal, working in class A mode : still both tubes contribute, but in push pull. Interesting to see that the situation is exactly the same as with a bypass cap. Both tubes are conducting but changing in the opposite direction. So the Voltage at the cathodes is constant - as if you would have a cap to ground.

operation mode 2):
Now let us assume a large signal (A and B op mode). Assume a sinus as a signal, starting from zero (crossover). In the first phase we are in mode 1)
With the increasing voltage we cross the point when one tube switches off. Now there is no "opposite signal"at the cathode of the Tube switched off. As a result you have now no longer a bypassed cathode resistor but both cathode resistors parallel between the cathode of the driving tube and ground (from AC signal point of view). This is basicall a negative feedback which reduces the amplification. As a result you have a very smooth an soft "clipping".

as an illustration there 4 scope pictures attached showing
1 clean signal before clipping starts , clean switch closed
2 the same input signal as in 1 , clean switch open. that results in what I call "smooth clipping" (I like that sound when playing blues)
3 Hard overdrive with crossover distortion, clean switch closed
4 the same input signal as in 3 , clean switch open,. We still have an area of "smooth clipping", then hard clipping, but without crossover distortion

[Silvio55]

Chuck, my voltages with the 120R resitor are: 10V across the cathode resistor, 344V at the plates, I tried for the PR mod zeners from 15V down to 11V and I always got that buzz, less notorious than without the PR mod, but it sounds the same as the cross over distortion buzz to me. I did the PR mod on my Night Train and it worked like a charm, no buzz at all, the NT has almost the same voltages and a 120R cathode resistor too, and I ended up with 12V zeners there.

[Chuck H]

Ok. Let's look at the rest of the circuit and how other things compare. What size are the PI coupling caps, power tube grid bias resistors and grid stoppers on the 18 watt compared to the NT??? Have you tried both amps through the same speaker?

In the article where Paul Ruby discusses his zener mod, his quest starts with his, as yet to be solved, confusion over why one amp has bad sounding crossover distortion and another doesn't. He even tried swapping the OT's!!! And still, one amp sounded good and the other one didn't. The only thing I noticed on the scope shots was that the "good" sounding amp only showed the crossover notch on one side of the waveform into a dummy load. Into a speaker load both amps have the crossover jagg on both sides of the waveform, but the "bad" sounding amp has a much greater dip and spike at the clipped ends. I'll surmise that the "good" sounding amp had some imbalance between the PI outputs. And that imbalance minimized the crossover notch on one side and/or reduced some NFB that, because of phase anamolies, caused the dip and overshoot in the "bad" sounding amp. What exactly is happening I can't say because of my limited understanding and a lack of specific information on this theory. But my point is that some imbalance in the PI may contribute to better sounding overdrive. Remember that we only call it "crossover distortion" when it sounds bad. Otherwise we either don't even notice a "problem" or we call it "swirl". Just for the heck of it, examine the differences between the NT PI circuit and your 18 watter. Then swap the PI tube in either amp and see if there is any difference. In past experiments swapping tubes I've noticed that, microphonics not withstanding, the PI tube seems to make as much, or more difference the OD quality of an amp as the first preamp stage! These were early tube swapping efforts and I didn't take any measurements. But I'll bet that the PI tubes that were less balanced were the ones that sounded best.

[Chuck H]

That's great. Kind of funny looking wave form, but there sure isn't any crossover distortion. It looks similar to PI or cathode follower clipping. I can't quite hear it looking at the wave form. But I'd be interested in how it sounds.

[Silvio55]

Ok. Let's look at the rest of the circuit and how other things compare. What size are the PI coupling caps, power tube grid bias resistors and grid stoppers on the 18 watt compared to the NT??? Have you tried both amps through the same speaker?

In the article where Paul Ruby discusses his zener mod, his quest starts with his, as yet to be solved, confusion over why one amp has bad sounding crossover distortion and another doesn't. He even tried swapping the OT's!!! And still, one amp sounded good and he other one didn't. The only thing I noticed on the scope shots was that on the "good" sounding amp only showed the crossover notch on one side of the waveform into a dummy load. Into a speaker load both amps have the crossover jagg on both sides of the waveform, but the "bad" sounding amp has a much greater dip and spike at the clipped ends. I'll surmise that the "good" sounding amp had some imbalance between the PI outputs. And that imbalance minimized the crossover notch on one side and/or reduced some NFB that, because of phase anamolies, caused the dip and overshoot in the "bad" sounding amp. What exactly is happening I can't say because of my limited understanding and a lack of specific information on this theory. But my point is that some imbalance in the PI may contribute to better sounding overdrive. Remember that we only call it "crossover distortion" when it sounds bad. Otherwise we either don't even notice a "problem" or we call it "swirl". Just for the heck of it, examine the differences between the NT PI circuit and your 18 watter. Then swap the PI tube in either amp and see if there is any difference. In past experiments swapping tubes I've noticed that, microphonics not withstanding, the PI tube seems to make as much, or more difference the OD quality of an amp as the first preamp stage! These were early tube swapping efforts and I didn't take any measurements. But I'll bet that the PI tubes that were less balanced were the ones that sounded best.

Thatīs the schem I used for the 18W, all the same values execpt for C1, I made a mistake and put a .22uf instead a .022uf cap there, today I tried with a .0022uf but nothing changed, the same buzz (and the same low end), so I didnīt even bored in putting the correct .022uf cap.

http://www.prowessamplifiers.com/sch...s/Vox_NT15.pdf

Thatīs the NT schem, I changed C10 to .0022, EL84 gridstoppers are 100K now, and added 10K gridstopper on the third gain stage and a 68k gridstopper on the PI.
The NT has a cathodine PI, so I canīt compare them, I tried changing tubes on the 18W PI, I tried a JJ ECC83s and a Mullard 12AX7 long plate reissue with the same results on the buzz, maybe I should try the Sovtek 12AX7W I have on the NT, but I donīt think that will make a difference. The 18W has an imbalaced PI, both plate resistors are 100K.
I always used both amps with the same cab, the NT matching cab loaded with a 25W Greenback.

[Chuck H]

Well, if we just examine the differences in the grid circuits, since the PI's are different anyway, the NT has a 200k load followed by a 100k resistance. The 18 watt has a 470k load followed by a 8.2k resistance. If I were a voltage trying to get to a power tube grid I'd take the 18 watt circuit any time. It's worth noting that the 18 watt PI is probably putting out more swing than the NT PI. Still, if you want to reduce the time constant, and therefor how fast the grid circuit can discharge, you could try 220k grid loads and 100k grid stops on the 18 watt. It will change the tone. A little darker and less hairy sounding. But this would make the 18 watt grid circuit similar to the NT grid circuit. Then you'll be closer to apples and apples for an evaluation.

Also, the NT schem shows a pair of .1 caps with a 470k load between them as each output coupling. Not sure why they did that. But the 18 watt schem shows the PI has .01 (ten times smaller) coupling caps. Is this accurate to both your amps?