Connection from input jack to V1

[Malcolm Irving]

So using dc on the filaments can get rid of hum and eliminate a headache.

When I build amps I do everything I can think of to minimise noise, buzz and hum, and generally I just get some hiss from the first grid stopper (10k to keep this low) and very little hum or buzz. The only thing I don't do is DC heaters. Not sure why not. Maybe a psychological block!

[pdf64]

I think that the 'worst case' for picking up hum here may be with a very high impedance instrument, eg piezo type PU.
That might be simulated by plugging an unterminated (shielded) plug into the input.
That may show the improvement provided by this wiring arrangement to best effect, hopefully it would be a reproducably measureable lower hum than with regular wiring.

[Malcolm Irving]

By the way, keeping the signal ‘go’ and ’return’ conductors close together in this way minimises interference from non-uniform electric fields as well.

There are only three ways to couple interference into a circuit: magnetic (i.e. mutual inductance), electric (i.e. stray capacitance) and shared resistance. So, the proposed scheme, in combination with a good grounding scheme, should minimise interference into the ‘input jack to V1 circuit’.

[Chuck H]

Mesa took the easy route and just used an un-terminated lead off the filament circuit to intentionally impress hum on an opposite phase circuit downstream. I guess you just move it around until the hum is minimized and then goop it in place (Mesa style). Pretty clever.

[uneumann]

By the way, keeping the signal ‘go’ and ’return’ conductors close together in this way minimises interference from non-uniform electric fields as well.

There are only three ways to couple interference into a circuit: magnetic (i.e. mutual inductance), electric (i.e. stray capacitance) and shared resistance. So, the proposed scheme, in combination with a good grounding scheme, should minimise interference into the ‘input jack to V1 circuit’.

I've concluded pretty much the same thing. My experience was with hum in a bassman with DC filament power and a shielded grid connection (single conductor cable). After much head scratching I followed that same "magnetic inductance" logic and used a dual conductor shielded cable to keep signal and ground together in the cable (for zero loop area). It made the difference. Hum was gone. No numbers to quantify it, but it was clearly audible. I thought to revert to AC filament power after the cable fix to see what that impact was - but I never did. IMO - it's likely that all 3 coupling methods matter.

[Malcolm Irving]

... a dual conductor shielded cable to keep signal and ground together in the cable (for zero loop area). ...

Yes, that should be even better, signal 'go' and 'return' conductors (probably twisted) inside a screen, with the screen grounded at one end only.

Some of the best guitar cables use that approach.

[Enzo]

We really need some quantification. I get the impression like I am setting up the suspension on my mom's car so it is stable to 140mph. She rarely reachs that speed on the way to the grocery store.

[Malcolm Irving]

We really need some quantification. I get the impression like I am setting up the suspension on my mom's car so it is stable to 140mph. She rarely reachs that speed on the way to the grocery store.

I'm just about to start building an 'exact copy' of an amp I built earlier. The earlier one had the conventional 'screened cable grounded at one end'. I'll try the proposed scheme on the new amp for comparison and report back. Might take me a few weeks to build the new one though.

One significant source of low level buzz is heater-to-cathode leakage in V1, so I might have to do a bit of tube swapping to get a fair comparison.

I will probably need to compare the two by looking at the output of V1 on a scope, as even 'exact copies' are never really exact, and other factors may intrude further downstream in the amp.

[mozz]

I have seen one amp, a Bogen perhaps, that had 3 filament wires. One wire was tightly coiled around the other slightly twisted pair, it went to ground on the far end only. How about using a shielded input wire, grounded on both ends, but breaking the shield somewhere near the middle?

[Justin Thomas]

Sounds like the way Fender tried to shield the signal wires to the tone controls in the 70s SF amps... hell, just clean up the execution of lead dress instead of wasting all that wire!

Justin

[nickb]

A good reference in this area is ‘Noise Reduction Techniques in Electronic Systems’ by Henry W. Ott.
On page 40 of the 1976 edition, for example:
“If a non-magnetic shield placed around the conductor causes the current to return over a path that encloses a smaller area, then some protection against magnetic fields will have been provided by the shield. This protection, however, is due to the reduced loop area and not to any magnetic shielding properties of the shield.”

I'll second that book rec. It's the most often referred to volume in my library and covers a huge range.

[Ken Moon]

In Merlin's grounding article, he shows the ideal setup, which is a shielded twisted pair or shielded coax from the input to the first gain stage, with the shield grounded only at the input side (he doesn't show a grid stopper in this pic, but usually recommends one of 10K):

[Malcolm Irving]

That’s very similar, but it doesn’t show the ‘go’ and ‘return’ conductors close together at the input jack and at the tube (perhaps that’s just a diagrammatic difference). Also I keep the grid leak at the input jack, so it doesn’t mess up the ‘layout’ at the tube end.
I’m regarding the grid stopper, grid-to-cathode of the tube, and the cathode resistor (or its bypass cap) all as parts of the overall ‘go’ and ‘return’ signal loop between input jack and V1. It’s this whole loop that should have its area minimised.

[Chuck H]

She rarely reachs that speed on the way to the grocery store.

Am I the only one to recognize that "rarely" doesn't mean never?

[uneumann]

Am I the only one to recognize that "rarely" doesn't mean never?

Yeah - it's also a case of "why not" do all these things. The methods we're talking about are all essentially "free". It's just a question of layout and wiring. Once you know the optimal way to do it - why not do it as well as you can?