I've seen 3 different recommendations for routing filter grounds to reservoir

[Mike Sulzer]

True enough, but I've read about a very high gain guitar amp (think it was a Dumble) that claimed to have an overall gain of near 1 million. That's a gain of 120dB !

EDIT: In high gain amps the signal is limited by clipping at various stages, but the buzz, hiss and hum start at a much lower level than the guitar signal but are then not clipped.

The hiss and hum are limited when the guitar is played because a high gain amp spends most of its time with multiple stages in saturation or cutoff, and there is very low incremental gain. If the guitar is not played, then the full effect of the hum and noise is heard unless circuitry is provided to take care of that.

[Gregg]

In high gain amp when guitar is not played hiss (shhhhhh) is unavoidable but that doesn't apply to hum unless you're running non rectified preamp tube/s filaments.

[R.G.]

When I suggested it I thought it was obvious that it was only practically applicable to the preamp. I guess I should have spelt it out.

Sorry if I wandered off. I'm slogging through a head cold and may have simply missed the nuance. Part of my missing it is that from my point of view, the piddly little several hundred watts (!!) of power to run even the output stage of a tube amp is easy enough to regulate, although it takes some work and thought to do it well. So I flat missed what you were thinking.

Kevin O'Connor did have some interesting suggestions. One of those was using a few-uF plastic film cap and a power-device run regulator/cap multiplier per stage or couple of stages in the preamp section, all running from the single B+ at the first one or two caps.

trobbins
I find it very hard to measure any meaningfull level of rms ripple on a preamp stage rail with my Kiethley 197, which has a 200mVFS range. I haven't yet set up to measure that hum spectrum using a soundcard to get better comparative levels of attenuation as the typical ladder RC filter extends back to the input stage - I'd guess there must be some blogs or threads going in to that. I sort of like the idea of taking a comparative output signal measurement for when a known resistance is added to a particular 0V inter-stage link in config#1, to see how much contribution a parasitic level of wire resistance would likely add.

Good comment and idea. The ripple on the power supply at a preamp stage is almost not measurable on the few amps I've tried measuring it. I love my picoscope, and the next time I've got it connected up to a tube amp I'll scope the B+ leads at various places and dump that into the spectrum analyzer function on it.

Also good idea to add some real resistors to the ground returns and validate the theory.

Gregg
I'm not an expert in theory but from experience and from what I've seen in different amps I can say that many grounding approaches will work, some of them as far from the star ground method as you can get. In PCB amps star ground is almost never used especially in multi channel "multi switching" amplifiers. It's mostly bus ground and sometimes all caps are clustered together at the power supply section of the PCB which is a violation of the theory but it works - the amp is quiet, doesn't oscillate etc.
The same applies for where to ground to chassis. Both at the input jack and first reservoir cap will work. Sometimes the second is quieter.

You're right. I was alluding to that when I inserted the caveats that grounding is perverse. Star grounding is definitely not the only way to get a properly grounded amp. It's just the only way that can be proven to provide that before you tinker with what goes where, how big the wires are, exactly what order what stages are connected to what sub-bus, and what part cancellation of hum and ripple might play.

These things are not violations of the theory - just validations that theory too has many levels. At a gross level, theory says you can get rid of interactions by forcing the grounding net to carry only isolated sections per wire. At a more refined level, you can combine sections to get some form of cancellation, or combine sections that introduce hum at such a low level that it's lost in other noise phenomena.

It's probably true that many PCB based amps use busses. However, when I designed the Workhorse amps, I ran a star ground on the PCB. Worked first time, never had hum issues in one that was not otherwise defective. There, that's MY anecdote. As with all anecdotes, it reinforces my biases.

As to hum and hiss at low levels versus clipping, clipping only reduces gain when the signal is actually clipping. Otherwise, for signals below the clipping level, gain is the full, unadulterated gain. So any front end hiss and/or hum is amplified by the full gain of everything behind it. That's why noise performance of an amp is so very dependent on what you do in the first stage, and the first rule of low noise design is to design a first stage that is itself quiet, and has impedances that minimize the self noise of the signal source and its own noise.

[catalin gramada]

keep small signal return away from power return. No.1(top pic.) by my choice

[Mike Sulzer]

As to hum and hiss at low levels versus clipping, clipping only reduces gain when the signal is actually clipping. Otherwise, for signals below the clipping level, gain is the full, unadulterated gain. So any front end hiss and/or hum is amplified by the full gain of everything behind it. That's why noise performance of an amp is so very dependent on what you do in the first stage, and the first rule of low noise design is to design a first stage that is itself quiet, and has impedances that minimize the self noise of the signal source and its own noise.

A guitarist using a high gain amp set for at least a reasonable fraction of its full gain is probably overdriven hard, with the signal in the linear region for only a fraction of the time, and so the signal to noise is probably not degraded much by the high gain, not that you have much chance anyway of hearing anything except the guitar at such a loud volume.

As for how quiet a preamp needs to be, a guitarist playing high gain is probably using a pickup with at least 10K of resistance. So this gives a hint as to how quiet the first stage of the preamp needs to be. Of course, the bandwidth coming out of the guitar might be as low as 1KHz, thus making it a lot less noisy than implied by the R value. 2.5KHz might be more likely for high gain playing, still less than the 5KHz we expect from the amplifier/speaker combination. If the volume pot is turned down a bit, the resistance looking back into the guitar could be as high as 250K, just plain bad.

If you use low output, low impedance pickups, you need a quieter preamp (or use a transformer). That is the only reason I see for having a low noise first stage.

[R.G.]

True, Mike. On stage, all the machinations we go through to make amps quiet and well behaved are generally lost. It is generally possible for the electronics to be much better in terms of absolute performance than the guitars, cords, and general havoc going on on stage.

But still, the bedroom players ask. There are many more bedroom players than stage players, and I suppose that's good if it drives the design of more highly performing amps.

But in general, I have a bit of despair and amusement at the same time about our continued process of designing out the last little bits of hash, noise, grit, grunt, and huuuuumm. To many players, amps stopped evolving after Fender or Marshall came up with their stuff in the 60s and 70s; even now, those very amps, in "original" condition are highly prized.

Sigh. It's the people that are the problem, I tell you!

[Malcolm Irving]

The ultimate in star-grounding would be for all the guitarists on stage to disconnect the screen of their guitar cable at the amp end and have a dedicated ground wire from the 0V on their guitar to one global star point on the stage (which is also connected to various amps etc.). Haven't tried it, but I suspect it would hum due to interference-induced voltages on those ground wires.

I think there is a reason we like signals to be carried by a pair of conductors as closely coupled as possible (as in a screened cable or a twisted pair).

I know the above is a 'ridiculous scenario' but I hope to make an analogy with how we do the grounding inside an amp.

[R.G.]

It practically begs for RF pickup, at the very least, and fluorescent and diode RF emissions if even one thing is wrong. At some point, the word "ground" no longer applies. Ask an antenna designer how grounded the shield of the coax to the antenna is along the way.

Hey! Here's an idea - let's get the guitarists to perform inside a Faraday cage - a wire mesh box with the wires grounded and all carefully attached along the seams. I know that some C&W bars use chicken wire between the audience and the band, but that's more for keeping beer bottles out than RF, I think.

[Malcolm Irving]

... I know that some C&W bars use chicken wire between the audience and the band, but that's more for keeping beer bottles out than RF, I think.

Yup. Those are the bars where they have both types of music.

Always check that the chicken wire is grounded.

[Malcolm Irving]

Just looking at the options for the ground connection between the preamp supply cap and the PI supply cap. Let’s assume there are one (or two) gain stages supplied by, and in close proximity to, the preamp supply cap. The signal is then AC coupled to the PI stage, which is close to the PI supply cap.

The preamp supply cap is kept charged up by current through the dropper resistor. Where does that charging current come from and return to? I would say the majority of it must come from the +ve of the PI supply cap and return into the –ve of the PI supply cap. Because, that current loop has a lower loop impedance than alternative loops which necessarily include the next dropper resistance (between PI cap and Screen cap). In ‘option 1’ the charging current is returned directly. I can’t see any advantage in making that current go to the global star point and then back again to the –ve of the PI cap.

When the signal is AC coupled from the stage(s) fed by the preamp cap to the PI stage, it presents itself as an AC signal voltage across the grid leak (and tail resistor if there is one) of the PI stage. Suppose the signal is 5 Vac and the grid leak (+tail) is 1M then we need to be supplying a signal current of 5 micro amps. Yes, it is tiny, but like all currents it still has to go around in some loop. In this case, that tiny current has to return to the preamp stage via the ground connection. Pure DC current mixed in with that tiny current has no audio effect, but any ripple current sharing the same ‘signal ground return path’ will be turned into a ripple voltage (by the non-zero resistance and inductance of the ground path) which adds to the true signal and will pollute the input to the PI. This is another way of saying that we want the 0 V signal reference at the output if the preamp stage to be exactly the same as the 0 V reference at the input of the PI. The best way to achieve that is a short, low resistance, low inductance, direct connection; not by going to a remote global star point and back again.

As mention above, the signal current travels in a loop: sending stage anode -> coupling cap -> grid leak of receiving stage -> ground return -> cathode circuit of sending stage. We want to keep the area of that loop to a minimum, to avoid interference from any AC magnetic field passing through the loop. Again, option 1 is better, because the loop does not have to go via the global star point and back again.

[SoulFetish]

This is Great! I'm glad I've been able to check back in on the forum. I didn't realize there were so many more thorough responses.
Regarding grounding noise in the preamp, and at the input in particular. I had an idea. I'm wondering if we could benefit from... the benefits of CMMR by employing a balanced input with a bipolar supply?
In this setup, we get power supply phase cancellation (assuming equal +/- supply filtering at the node) and close to identical balance at the output (using real-world components). Then, feed a fully differential second gain stage. In this case, inspired by the cathode coupled amplifiers found in the Direct Coupled chapter of RDH4:

and (more currently) Mike Sulzer's http://music-electronics-forum.com/t41822/ with minor changes.
In this case, using the non-inverting output to control the second triode instead of a grounded grid input. This may be diminishing returns for the effort, but at the very least brings to the topic the idea of grounding a bipolar supply as a possible way to keep power supply grounds and signal ground separate

[catalin gramada]

Did you ever thinking to use a stereo cable as guitar interconnect? Keeping the screen tied at chassis end and further all guitar shields..?

[Malcolm Irving]

There's a 'bare-bones' balanced amp circuit here:

https://www.ampbooks.com/mobile/amp-...-schematic.gif

The input from the guitar is unbalanced, but the first stage is a phase inverter and everything is balanced after that.

[nickb]

The input stage only give you about 6dB more power supply rejection that a conventional common cathode stage. Just doubling that stage's filter cap will give you about the same effect for no added complexity. Of greater concern to me is that the noise performance will be significantly negatively impacted due to complete lack of gain in the first stage.

[SoulFetish]

The input stage only give you about 6dB more power supply rejection that a conventional common cathode stage. Just doubling that stage's filter cap will give you about the same effect for no added complexity. Of greater concern to me is that the noise performance will be significantly negatively impacted due to complete lack of gain in the first stage.

Interesting point. So, if the only benefit (if it’s even a benefit) of the stage is a balanced output, then an well constructed audio transformer is probably a much better option?