Isolating FWB (-) from other ground paths with a PT without CT (tube amp)?

[Murrayatuptown]

I think this question may already be answered but not finding it with my search details.

I want to, as a starting point, isolate a semiconductor FWB (-) and reservoir capacitor (-) from the rest of the B+ voltage dropping and filtering capacitor returns (-).

The power transformer has no CT, as it's unnecessary with FWB.

To isolate the FWB (-) from the 'cleaner' return(s), should I use the chassis connection where the PT mounts for FWB(-)/Reservoir(-)?

Thank you

Murray

[martin manning]

Yes, treat the negative DC corner of the FWB the same as you would the CT in a two-phase rectifier, i.e. ground it and the reservoir to the chassis together near the PT. It is the return for all of the HT current.

[bepone]

minus poles after rectifier (anodes of diodes) should not be grounded to chassis! should go directly to negative terminal of the electrolytic cap!

chassis in amp building need to be considered as if it were made of plastic, not connecting point for some current paths, in short.

[Murrayatuptown]

Thank you, Martin & Bepone.

[ChopSauce]

With apologies for deviating from the original question ...

minus poles after rectifier (anodes of diodes) should not be grounded to chassis! should go directly to negative terminal of the electrolytic cap!

chassis in amp building need to be considered as if it were made of plastic, not connecting point for some current paths, in short.

How about Dumble (& Fender, etc) amps having all these ground points, then ...

[martin manning]

How about Dumble (& Fender, etc) amps having all these ground points, then ...

Of course! A multi-point ground scheme using the chassis for returns can be just fine if done correctly. The first and most important thing is to connect the reservoir negative directly to the main return to isolate the high current pulses. Then there are a couple of options for dealing with the other returns.

[bepone]

With apologies for deviating from the original question ...

minus poles after rectifier (anodes of diodes) should not be grounded to chassis! should go directly to negative terminal of the electrolytic cap!

chassis in amp building need to be considered as if it were made of plastic, not connecting point for some current paths, in short.

How about Dumble (& Fender, etc) amps having all these ground points, then ...

and all this is of course wrong.. and all the millions people around doing the same. why you think that current paths through the chassis is proper way? it is the worst way.
chassis need to be just extension of the neutral earth line, for protection and shielding, nothing else, conected to signal gnd only at the input jack in 1 point.

all confusion in grounding starts from there, that people without experience trying to do some circuit wiring through the chassis, and doing so, mixing high and signal currents in chassis, resulting in high noise and humm. and after speanding hundreads hours in troubleshooting and loosing momentum and long term interest (time also) in tube circuits.. seen absolutely every time when someone start to build something.

so proper way must be, think about chassis like it is plastic chassis, people will be forced to make connection in proper way, like wires from rectifier directly to capacitor not through the chassis and all other common mistakes.. results will be automatically good and correct. it is very simple and straightforward to understand this.

[bepone]

then next level when you understand high and small signal currents in the ground plane or wire, then is easy to apply chassis for return currents, like giant buss bar, where still need to be applied the same rules from not mixing currents in gnd, but for beginner it is better never to mention chassis at all, he will just make too much mistakes easily.

btw. the quietest amp ever what i did, no humm, noise, hiss, high gain Soldano style, was pcb build, without any chassis, all amp intergrated on pcb (not any random pcb, pcb designed for that application with all the rules applied!)

[R.G.]

@Murrayuptown:
In my opinion, you should
(1) NOT use the chassis as a ground for anything except the third-wire AC mains safety ground
(2) use the chassis as an RF/capacitive shield
(3) make a distinction between signal ground and all other things connected to ground
(4) wire your signal ground network on wires using "star ground" techniques
(5) treat the connection between the rectifier negative wire to the first filter capacitor in a special way, described below

The confusion with grounds and grounding for low noise and hum starts with schematics. All schematics use the convention that connections are superconductors, wires with no resistance, inductance, or stray capacitance. This is, of course, flatly not true.

All real wires (and the chassis itself, and buss bars, and so on) have resistance, although it's small compared to the resistors in the rest of the circuit. So currents through the wires cause a voltage, according to Ohm's Law: V = I * R. The circuits in tube amps have zero rejection of ground noise, so any voltage wobbles on their "ground" is amplified by the gain of the rest of the circuit. Even a few millivolts of signal injected by ground wobbles gets amplified by the full gain of the amplifier.

Wiring the rectifiers to the first filter cap: These wires carry BIG current pulses. A big current, even through a low resistance, makes a measurable voltage drop. The rectifier negative wire to the first filter cap should be >one< wire, and should have no other connections. The One True Ground is the negative terminal of the first filter capacitor. All other "grounds" should branch out from here. This technique is called "star grounding" in the trade.

You will find many other schemes for amp grounding. Some of these work well, but often they work by what I call the method of offsetting errors. They connect the ground reference from various places and combine the ground currents in a way that more-or-less accidentally causes ground-induced voltages to cancel to some extent. You will also find that people that have had good luck with a grounding scheme that's not star grounding will be passionate about not needing star grounding. And they're right - on that one, single amplifier. Star grounding is the only easy-to-understand way to get low ground noise in a predictable way.

This is a GREAT article on star grounding and many other aspects of wiring an amp: https://audioxpress.com/article/Glass-A ... of-a-Champ

[martin manning]

You will find many other schemes for amp grounding. Some of these work well, but often they work by what I call the method of offsetting errors. They connect the ground reference from various places and combine the ground currents in a way that more-or-less accidentally causes ground-induced voltages to cancel to some extent. You will also find that people that have had good luck with a grounding scheme that's not star grounding will be passionate about not needing star grounding. And they're right - on that one, single amplifier.

I typically use a "multi star" arrangement, where there are only a few chassis ground points in strategic locations. I have applied the same basic principles many times with success, and I'm confident that grounding scheme artifacts are not dominating the noise floor.

[Murrayatuptown]

Thank you, bepone & R.G.

Earlier I came to the conclusion the amp was dramatically quieter with all the chassis and PCB connection points isolated, using clip leads to experiment for different local ground connection points. But still not quiet enough. At one point I swapped two different NOS 5751's for one of the 12AX7's and the noises got much worse again (hum & random grumbling noise). I concluded there was more wrongness to pursue and also wondered if 'quieter' was circumstantial...would another bad practice be reinstalled when I put it all back into the cabinet?

A scope or DMM probe in a 'wrong' place (I think the PP EL84 plates) was confirmed by a dog in the house that I thought I heard some oscillation. No other test points caused that & I don't have a need to measure there for this effort right now (problem avoided).

There is some 'culture' to overcome with avoiding bad grounding practices. I have read multiple recommendations but have to get the assumptions out of my head, particularly thinking something seems missing.There are books on the subject because even people with education & experience make mistakes.

I'm overthinking it because I am going to cut ground traces after verifying what stages are currently connected where on the PCB. I don't need to reassemble the whole thing to see what improvement separating the FWB (-)/reservoir(-) achieves. I'll hold off cutting the tube stage ground traces until I look at where my grounds resulted from adding a bias pot and 1 ohm EL84 cathode resistors with DMM jacks. The jacks are insulated but some choice had to be made for the common resistor 'ground' ends. I was going to install a turret terminal at each point I isolate a 'ground' and use shielded twisted pair cable to relocate local grounds (experimentally) (as detailed in Blencowe's book with the Grounding chapter (PS book, I think).

I needed to hear a little more loudly than what I've read so far, that there must be separate ground paths...so thank you for that support.

I've had enough unintentional floating ground problems unintentionally induced that I can forget all logic.

Extra ground connections can be bad. Last time I put a Strat clone back together with too many installed extra ideas (4th time?) I had a new hum and indeed was too disgusted to take it apart again. (Just as bepone described). It's my daughter's but shes a violinist now & doesn't like frets!

The amp has more urgency because it's a friend's ;@).

[R.G.]

I typically use a "multi star" arrangement, where there are only a few chassis ground points in strategic locations. I have applied the same basic principles many times with success, and I'm confident that grounding scheme artifacts are not dominating the noise floor.

If it works for you, use it.
The idea is to arrange ground wiring so that return current from either power or signal circuits do not cross-contaminate one another. There are many ways this can work.

Most nominally star grounded amps are really collections of local star-ground "domains"; it's nearly impossible to run a ground wire for every cotton-picking component that connects to ground. At some point the whole chassis gets 100% filled with ground wires, no open spaces. So smaller star collections are made, with a ground wire from each local star to the One True Ground point. This works especially well when the smaller star has ground currents that are all either cancelling or reinforcing the signal in that star. For instance, a single triode stage, or a triode stage plus a follower, or a triode stage, controls, then a recovery stage. These are all working on the same AC waveform, so local ground wobbles are entirely composed of DC plus the signal. Local ground interactions can't do much except make the gain a little bigger or smaller, and there isn't enough gain in a tightly related circuit to goad it into local oscillation in most cases.

If I understand you correctly, the difference you're making is that you use local stars and then connect them to the chassis. That can be made to work fine (obviously - it works for you!) if the strategic chassis ground locations are in places that don't conduct currents big enough to make for ground offsets in the chassis metal resistance.

Counter-indications are for situations where big current pulses are being carried by the chassis. Bad examples are when the rectifier negative point (CT or FWB negative) is tied to chassis and the first filter cap negative is tied to chassis, when speaker current return is carried through the chassis, or when AC mains leakage current is carried on the chassis, and then inputs are also grounded to the chassis. These situations have the potential (yuk, yuk) to cause noise and/or oscillation if the current path on the chassis can raise and lower the signal ground voltage because currents running from one place to another are tapped off by an input connection. Another bad situation can arise when chassis connections are on screw-down terminals and the screws get loose, or if a critical path is through screws into a cover plate and out, and the screws get loose. For instance, using the input jacks for ground referencing the chassis, or grounding tone/volume control circuit grounds to the body of a pot, and expecting the bushing nut to always be tight.

If you're picking good places for connections, it can work very well. Star grounding is NOT the only way to get a quiet, stable grounding arrangement. But it is the only way that is predictably quiet, not requiring luck or thought and experiment. Other techniques will usually require some work to pick out a suitable set of strategic locations.

[R.G.]

There is some 'culture' to overcome with avoiding bad grounding practices. I have read multiple recommendations but have to get the assumptions out of my head, particularly thinking something seems missing.There are books on the subject because even people with education & experience make mistakes.

Boy, are you ever right about that "culture" thing. I regularly get pounced upon by someone from a different ground culture.

Down at the bottom of all this is the idea that you need to understand what currents are going through which conductors. It's a lot of work and thought to do this. Most people, myself included, memorize a few hard cases and develop habits that sidestep those hard cases. The rectifier pulses into the first filter cap is a big one that every amp has. The speaker return current is often a problem if you just put in a jack and let the chassis carry the amps of speaker return current. Using a specific wire and forcing those currents to flow in the specific wire means that the current-induced voltages don't impinge on signal by wobbling the signal ground up and down.

[martin manning]

If I understand you correctly, the difference you're making is that you use local stars and then connect them to the chassis. That can be made to work fine (obviously - it works for you!) if the strategic chassis ground locations are in places that don't conduct currents big enough to make for ground offsets in the chassis metal resistance.

Here's a recent example where the AB763's scattered grounding was reimagined (wow, been two years already!). I'm basically following the Dumble-style ground scheme: https://ampgarage.com/forum/viewtopic.p ... 77#p428877
I did not want to go to the trouble of isolating the audio jacks, but they all have star washers for good electrical contact, and to keep them from loosening. I did get rid of all the pot case grounds, and the filter grounds ere separated so that all of the tubes supplied by the last filter cap are grounded near the inputs, along with the negative lead from that filter. The reverb grounds are still at the rear of the chassis, near the speaker ground, and the PI is also grounded there. The chassis does not carry the speaker current since both OT secondary leads go to the speaker jack.The other change was to separate the trem tube neon side ground from the PI ground. The local stars are not exactly single ground points, but still this amp is very quiet.

[Murrayatuptown]

I'd like to think I took the thought and work approach, but part of it could easily have been dipping a toe into the paranoia/superstition pool (even if l was the only one at the shallow end)!

I got obsessed trying to figure out why some transformers have different numbers of nylon or fiber washers on the bolts that clamp the laminations and end bells (if present). I read one person's opinion it's worth assuring only one bolt makes contact with the transformer and chassis, and another said it might matter occasionally but how can you know?

Then I saw a photo of a transformer on the Hammond website. 3 of the 4 nuts were visible. 2 had insulating washers under the nuts, one had no insulating washer, and the other was obscured. Couldn't see the screw heads, but I got curious and sent an inquiry to them about what it accomplishes. The application engineer who replied said it was a good question and he remembered reading about it many years ago but didn't remember (apparently not critical).

I've read opinions about avoiding shorted turns, but the bolts thru an EI transformer don't pass thru the main flux like a toroid mounting bolt. There is no question what happens when allowing a toroid mounting bolt to clamp an uninsulated steel mounting washer thru the core and chassis.

I have also seen transformers with welded laminations. Those inspire separate folklore that all transformers should be that way (well, at least one former co-worker believed that). That's a head-scratcher. It must be for a specific purpose more functional than saving the cost of bolts or (really?) actually stifling magnetostriction (core vibration). Or I hope so.

That it's done deliberately (if infrequently) makes me assume it is done when a need is recognized...but for what purpose?

There are too many variations in transformer design to generalize, but maybe it alters the leakage flux or how it couples (or doesn't) to the chassis. Maybe similar to flux bands outside a transformer core (but I've only seen non-ferrous types with copper or aluminum tape. Or I'm making things up.

So I jumped into the deep end. I removed the power transformer nuts and bolts (it was varnished so nothing moved), found heat shrink tubing that fit over the bolts and put them back thru the laminations, bought the last nylon washers my local hardware store had, and put steel toothed lockwashers between the transformer and the stack nut and between that nut & the chassis. I'll never know whether it helped but I hope having multiple bolts wasn't shunting stray flux to make the field bloom away from the chassis instead of coupling. (I'll just have to say "what the flux, man?" if I relocated a problem. I have no options for rotating any magnetics now. I squeezed in a larger output transformer and a choke. The choke can go if it ends up being a problem near the speaker magnet (it fits).

I don't remember how long it took to rearrange the power transformer hardware, but it was definitely tedious and had me questioning what I was thinking. I have had bolts shear off instead of the nuts loosening, with tougher-than-varnish potting material (Midwest EIA 549).