Star grounding problem

[Alan0354]

Merlin's schematic does confine the heavy power supply ripple currents, and they don't get into the audio (if you draw it correctly........).

I've used similar grounding schemes with single ended and push pull amps and reproducibly experience no significant hum.

Douglas Self (who know a bit about analogue audio........), also advocates a similar ground scheme.

Hopefully, Merlin will provide a more authoritative answer than I seem capable of supplying.

I follow his drawing in Fig 3.11, I drew the current path out in post #39 to show how the current of his. To make it clearer, even if SE don't have PI, my theory still holds if you inject the cathode current into one of the pre-drive stage. Just because you don't see a problem don't mean it's the correct way to do it. People can get away with a lot of things that the theory said they are not suppose to do. That doesn't make it ok. You can look at the current path from post #39, it is quite simple. Merlin has to show the advantage of injecting large current into the preamp section to convince me.

Point is you want to start with a clean design. So it is more forgiving later. Fender SF/BF are one of those that you can rip up a lot of the signal path, hook up one tube to another, playing around all over it. If you don't alter the cathode grounding point, don't touch the B+ bypass and the plate resistor connection, you are pretty safe. I remember I ripped out the reverb and tremolo circuit and used them for cascade gain stage on my Twin in 79, I jumper the signal path around to different tubes. I never ran into problems. I then tried to modify the Michell that use one point ground and I ran into humming problems.

Leo Fender is one of the few I admire more the more as I get into music electronics deeper. He had the foresight to do the grounding like a ground plane. That is 30 years before time. Then he built a guitar with a piece of wood and bolt a neck on. He then put pup on a wood block and slide around under the strings to get the right sound. Then in 1984, he was the first one to patent the first active noise cancelling circuit that used in the Elite Strat. Then he had the nice circuit that Marshall copied..........then he changed to BL and SF that still the standard by which amps are judged. He is the father of modern guitar.

Leo is like Bruce Lee in Martial arts, light years ahead of others. They create the trend. Yes people are getting better, but it would be very different without them. These are true genius.

[Wombaticus]

So why do you want to have a dedicate cap just for the screen, then go through a resistor before provide the HV for the PI? Just to put the cathode current into the screen circuit? for what?

Ah. This, I think, is showing up your unfamiliarity with tube guitar amps. Quite often the target voltage for the PI node will be 100V or more lower than the target voltage for the screens. The screens in musical instrument amplifiers especially are often run at much higher voltages (often close to the plate voltage itself) than is typically the case in more sensibly engineered audio amplifiers or other tube circuits. Indeed, not infrequently the screen supply voltage will exceed a sensible number for the HT supply of a typical preamplifer triode. In any case, I think it's safe to say for the last forty years very few guitar amp designers have opted to share the PI and screen filter caps. That this is true even of manufacturers (like Fender) that were notoriously penny-pinching I think says volumes.

[Alan0354]

Ah. This, I think, is showing up your unfamiliarity with tube guitar amps. Quite often the target voltage for the PI node will be 100V or more lower than the target voltage for the screens. The screens in musical instrument amplifiers especially are often run at much higher voltages (often close to the plate voltage itself) than is typically the case in more sensibly engineered audio amplifiers or other tube circuits. Indeed, not infrequently the screen supply voltage will exceed a sensible number for the HT supply of a typical preamplifer triode. In any case, I think it's safe to say for the last forty years very few guitar amp designers have opted to share the PI and screen filter caps. That this is true even of manufacturers (like Fender) that were notoriously penny-pinching I think says volumes.

Read my post again, I think you missing the moon!!! It is not just about the PI. My question is why do you want to pull a high current path from the PA and put in in circuit after the choke? This is not about tube amp, this is about grounding.

Why do you want to put noise to the screen? I thought you use the voltage after the choke and put a big filter cap for the screen to reduce noise. Then why do you want to put noise on the low side of the screen filter cap that will go right to the screen?

[Steve Conner]

Oh come on guys. You're arguing over maybe 1" of wire worth of a difference, in a part of the circuit with low gain to the output. (The grid of a power tube is typically about unity gain to the speaker, and the screen maybe one-tenth.)

Back in the old-days of multi-section caps, the plate and screen filter caps could share a negative terminal and the two topologies would reduce to the same thing.

[loudthud]

Tubes are not sensitive to absolute voltage, they are sensitive to voltage relative to the cathode. Since the cathode is tied to a noisey ground, the screen filter should also be tied to that ground. The choke does a wonderful job of removing supply ripple from the screen node in conjunction with the screen node filter cap. It also allows the screen node to sag down rather quickly when current demand increases ripple on the main B+. Having a separate filter for the PI allows the use of a quiet ground for that cap. The dropping resistor isolates any noise on the screen node from getting to the PI and further down stream as long as a quiet ground is used for those nodes. Remember that filter caps are a low impedance for noise so any noise on the ground side goes right to the B+.

The problem with most star grounds is the stack of spade lugs on a transformer bolt. Change the order of the spade lug stack and you change the noise. If you made a star ground with a ring of eyelets and one in the center, noise would be reduced. Still, it would be better to route the PT center tap to the first filter ground.

[Wombaticus]

Read my post again, I think you missing the moon!!! It is not just about the PI. My question is why do you want to pull a high current path from the PA and put in in circuit after the choke? This is not about tube amp, this is about grounding.

Why do you want to put noise to the screen? I thought you use the voltage after the choke and put a big filter cap for the screen to reduce noise. Then why do you want to put noise on the low side of the screen filter cap that will go right to the screen?

Allen, I don't think I'm missing the point. If you read what I've said, I haven't yet really commented on what I suspect is the main thrust of your posts, and so far I've reserved judgement.

The issue for me, and I suspect for others, is that you often compose posts that talk about many issues at once. That's OK, and I think you make good points. But unfortunately, while your English is generally pretty decent, I'm not infrequently left confused by your manner of expression. So it's often necessary to peel away all the extraneous bits to get at your central thesis.

For instance several of your earlier posts seemingly suggested that having PI and the screen sharing a common filter cap was a major concern of yours. I wrote a post clarifying that, and then subsequently answered a direct question from you about why it might be desirable to have the PI and the screen on separate power supply nodes. At which point you told me that I'm "missing the moon".

Anyway, the bottom line is that here and in other threads, it seems like you and the other posters are talking past one another. In general, it's necessary to clear up all of the confusing side points before we can really be confident that everyone is talking about the same thing. A further case in point is your last drawings of Merlin's ground scheme. The problem is that if you read Merlin's chapter, he pretty clearly advocates *avoiding* exactly the kind of chassis attachment points that you show in your drawing of his scheme. But in fact, his preferred embodiment of a grounding scheme has one-and-only-one chassis attachment point, with that point being preferentially located near the input. Maybe it's pertinent to your central point or maybe not, but let's clear that up to make sure we are all talking about the same thing.

In this thread, perhaps another point of confusion seems to be about what actually constitutes a star point... A lot of people seem to think that a star point necessarily implies that there is a chassis attachment at the same point. But this is not necessarily the case.

[Alan0354]

To clear it up.
1) I disagree with the one point ground at the input or any part of the circuit. I advocate to have a ground plane like Fender that ground along the way. I explained the reason about the why ground current will not interact if you have each tube circuit very close to it's filter cap.
2) At the very first post I had the drawing, I said assuming the PI share the B+ with the screen voltage and share the same filter cap. BUT if there is no PI, the explanation holds as I explained that the cathode current of the power tube will contaminate the ground at any circuit before the PA section.
3) I isolate all the high current path of the PA from the rest of the circuit.
4) This is not a tube amp problem, this is a grounding problem. This is how you fix the ground by looking at how the current moves. Whether I know tubes is irrelevant. This is grounding.

[Alan0354]

Oh come on guys. You're arguing over maybe 1" of wire worth of a difference, in a part of the circuit with low gain to the output. (The grid of a power tube is typically about unity gain to the speaker, and the screen maybe one-tenth.)

Back in the old-days of multi-section caps, the plate and screen filter caps could share a negative terminal and the two topologies would reduce to the same thing.

You can have two star grounding point 1" apart, but if you put the ground in the wrong star, you can have hum. Remember, current choose the path of least resistance even at very low frequency.

[Alan0354]

Tubes are not sensitive to absolute voltage, they are sensitive to voltage relative to the cathode. Since the cathode is tied to a noisey ground, the screen filter should also be tied to that ground. The choke does a wonderful job of removing supply ripple from the screen node in conjunction with the screen node filter cap. It also allows the screen node to sag down rather quickly when current demand increases ripple on the main B+. Having a separate filter for the PI allows the use of a quiet ground for that cap. The dropping resistor isolates any noise on the screen node from getting to the PI and further down stream as long as a quiet ground is used for those nodes. Remember that filter caps are a low impedance for noise so any noise on the ground side goes right to the B+.

The problem with most star grounds is the stack of spade lugs on a transformer bolt. Change the order of the spade lug stack and you change the noise. If you made a star ground with a ring of eyelets and one in the center, noise would be reduced. Still, it would be better to route the PT center tap to the first filter ground.

If you look at my drawing, the cathode of the power tube should have very little noise by the way I use the star ground at that one point. I showed how the current runs and show there is not current running into the ground.

That's the whole point of my design..........isolate and make sure all the high current and ripples have their return within the little island...............away from the main chassis. The stub resistor from point C to ground do not carry any current from the power amp. This is explained in post #39.

[Alan0354]

I have been thinking about the screen ground. The only thing I can even think of is that the screen can draw a lot of current with the signal and this will throw noise to the grounding point of the screen filtering cap. So you want to put the grounding point of the screen filtering cap with the cathode return. This at least can make sense because the screen and cathode form a current loop by itself.

That's a little weak, but at least this can make sense a little. If you can have solid proof, I might be willing to change my mind.

[jpfamps]

Why do you want to inject a large cathode current from the power tube to any circuit that is powered after the filter choke like it is shown in Merlin's Fig 3.11 and other people here?

Because you don't.

See figure 13.7.

[jpfamps]

But in fact, his preferred embodiment of a grounding scheme has one-and-only-one chassis attachment point

I think it is worth noting that both Douglas Self and Kevin O'Connor advocate one chassis ground point as well.

[Alan0354]

Because you don't.

See figure 13.7.

Fig 13.7 don't show anything relate to this. But I came up with a reason in post #55 that can make some sense why the cathode and screen share a current loop. But if that's the case, the screen bypass cap has to move to the same stud of the star to accommodate this in Fig 3.11. Or else by taking care of the screen current, you introduce the major current from the plate to the quiet ground.

[Wombaticus]

To clear it up.
1) I disagree with the one point ground at the input or any part of the circuit. I advocate to have a ground plane like Fender that ground along the way. I explained the reason about the why ground current will not interact if you have each tube circuit very close to it's filter cap.
2) At the very first post I had the drawing, I said assuming the PI share the B+ with the screen voltage and share the same filter cap. BUT if there is no PI, the explanation holds as I explained that the cathode current of the power tube will contaminate the ground at any circuit before the PA section.
3) I isolate all the high current path of the PA from the rest of the circuit.
4) This is not a tube amp problem, this is a grounding problem. This is how you fix the ground by looking at how the current moves. Whether I know tubes is irrelevant. This is grounding.

I kept asking a very simple question and no one yet answer that yet.

Why do you want to inject a large cathode current from the power tube to any circuit that is powered after the filter choke like it is shown in Merlin's Fig 3.11 and other people here?

Well, as you acknowledge here, you didn't just ask a question, you said all kinds of stuff.

On point one above, you disagree, fine, but I'm not sure that you adequately justified WHY you disagree with that strategy so vehemently. In my amps, the one-chassis-connection modified star ground system works for me first time, every time - I think a big selling point for hobbyists is its predictability. On the other hand, with my admittedly limited understanding of these issues, it seems like the scheme you advocate might theoretically have some benefit at RF, but it's entirely unclear to me why it should have any superiority, generally speaking, at audio frequencies. You make a big deal about the low impedance (as opposed to resistance) of the ground plane, etc., which leads me to suspect that you're still really thinking in high frequency terms.

And I guess I have to disagree with you about your subjective assessment of the noise performance of Fender amps. My own admittedly subjective opinion, frankly, is that it's generally just OK at best (even with brand new filter caps, etc.), and often just not very good, really. At any rate it's easy to build an amp with equivalent or greater gain that is much quieter than most blackface or silverface Fenders, although I acknowledge that there is much more in play here than just the grounding scheme.

As to what has emerged as your simple question -- Theoretically, I don't know, and I look forward to others chiming in. What I do know (and with RGs comments firmly in mind) is that in practice, with several new amp designs I've built with an otherwise rationally motivated ground scheme, it has never made a scintilla of difference that I could detect if the power tube cathodes returned to the -Ve of the screen or to the plate supply. And as Steve has pointed out these are bound to be both electrically and physically close.

[jpfamps]

Well, as you acknowledge here, you didn't just ask a question, you said all kinds of stuff.

On point one above, you disagree, fine, but I'm not sure that you adequately justified WHY you disagree with that strategy so vehemently. In my amps, the one-chassis-connection modified star ground system works for me first time, every time - I think a big selling point for hobbyists is its predictability. On the other hand, with my admittedly limited understanding of these issues, it seems like the scheme you advocate might theoretically have some benefit at RF, but it's entirely unclear to me why it should have any superiority, generally speaking, at audio frequencies. You make a big deal about the low impedance (as opposed to resistance) of the ground plane, etc., which leads me to suspect that you're still really thinking in high frequency terms.

And I guess I have to disagree with you about your subjective assessment of the noise performance of Fender amps. My own admittedly subjective opinion, frankly, is that it's generally just OK at best (even with brand new filter caps, etc.), and often just not very good, really. At any rate it's easy to build an amp with equivalent or greater gain that is much quieter than most blackface or silverface Fenders, although I acknowledge that there is much more in play here than just the grounding scheme.

As to what has emerged as your simple question -- Theoretically, I don't know, and I look forward to others chiming in. What I do know (and with RGs comments firmly in mind) is that in practice, with several new amp designs I've built with an otherwise rationally motivated ground scheme, it has never made a scintilla of difference that I could detect if the power tube cathodes returned to the -Ve of the screen or to the plate supply. And as Steve has pointed out these are bound to be both electrically and physically close.

I concur with your judgement.