cathode bypass and grid stoppers/ when where and why

[Malcolm Irving]

...
If the hum is due to poor balance between the El34s, should it not be just as audible with the volume all the way down?

When you make the cap bigger, you are reducing the hum at all the successive stage of the RC filtering (if the circuit is the usual one or a variant), and so the question of where the hum is introduced is still open, I think.

Good points, I agree. With the vol. all the way down, the output drops to 0.5mV ac. The hum (as well as the hiss) must be coming from the first stage (which is the only one prior to the vol. control).

[Mike Sulzer]

Good points, I agree. With the vol. all the way down, the output drops to 0.5mV ac. The hum (as well as the hiss) must be coming from the first stage (which is the only one prior to the vol. control).

So perhaps the hum is from the filament (how they are heated, that is)?

[pdf64]

But Malcolm notes that it's 100Hz hum (post #26).

[ric]

But Malcolm notes that it's 100Hz hum (post #26).

USA is 60 cycle 120v for residential, you are 100hz 220v? So you are talking possible mains hum?

Malcolm is using an aluminum chassis, not as good as steel for magnetic shielding. Could this be how hum is getting through? Are preamp tube shields being used? If aluminum, would steel shields make a difference?

[Malcolm Irving]

The amp uses AC heaters. We’re on 50Hz here in the UK, which makes the hum 100Hz on B+, so I expect it is coming from the B+ rail. As Mike said, doubling the reservoir cap reduces the hum throughout the B+ supply, including that to V1. The EF86 has an internal shield, which in this case is grounded.
I’ve tried 4 old EF86s and 2 new ones, producing a range of hiss and hum output levels, at full volume, from 13mV to 20mV. (All with 2 x 47uF reservoir.)
It is interesting that the hiss and hum are subjectively at about the same level, while the earlier test suggests that hum is a bigger voltage component of the output. (Doubling the reservoir reduced the hum but does nothing to the hiss.) I expect this is due to human hearing being much less sensitive to 100 Hz than to the higher frequencies in the hiss, at these low sound pressure levels.
The hum (and hiss) are already at an acceptable low level, but I will try tacking on an additional smoothing cap on the B+ for V1 and report back, when I have some time, just out of academic interest.

[Malcolm Irving]

I’ve increased the B+ smoothing cap which supplies V1 (and the LTP) from 22uF to 44uF.
Noise output at full volume is now down to 8mV. Subjectively, the hiss is now much more noticeable than the hum.
Tried grounding to the chassis near the input jack (option 1) or at the reservoir cap 0V (option 2) – still no noticeable or measureable difference.

[Mike Sulzer]

But Malcolm notes that it's 100Hz hum (post #26).

My experience is that when you get the hum down to the level Malcom has, and it it is fundamental, and you are using ac heating, then the problem is the heaters. But it might be different with the EF86 pentode rather than the 12AX7 triodes I am familiar with.

Speaking of pentodes, has anyone tried using them further along the preamp chain rather than at the input? The idea would be this: Since amps today often have power stages that are too powerful for the application, there is a greater emphasis on preamp distortion. But that is usually all triode, while the output stage is almost always pentode or, similarity, beam power tetrode. I am thinking that an interesting thing to try would be an EF86 phase splitter before the triode phase splitter. Each EF86 output would feed one input of the triode phase splitter through a "stereo" master volume control. The EF86 splitter would also have an "imbalance" control so that you could adjust the ratio of even and odd harmonics. Has anyone tried something like this?

[Mike Sulzer]

I’ve increased the B+ smoothing cap which supplies V1 (and the LTP) from 22uF to 44uF.
Noise output at full volume is now down to 8mV. Subjectively, the hiss is now much more noticeable than the hum.
Tried grounding to the chassis near the input jack (option 1) and at the reservoir cap 0V (option 2) – still no noticeable or measureable difference.

So I am wrong as usual: it is (or was) still the B+. But at some level it must be the heater.

[Malcolm Irving]

The EF86 only uses pins 4 and 5 for the heater, so I was able to get a very tight twist right up to the valve base. The heater supply is centre-tapped and elevated by connecting the centre-tap to the EL84 cathodes (they are cathode-biased).

[Mike Sulzer]

The EF86 only uses pins 4 and 5 for the heater, so I was able to get a very tight twist right up to the valve base. The heater supply is centre-tapped and elevated by connecting the centre-tap to the EL84 cathodes (they are cathode-biased).

It would seem that yo have done all that you can do. I wonder if there is something about the construction of the EF86 that allows good internal rejection.

[Malcolm Irving]

There were some interesting claims from Mullard about the EF86, reproduced here:

EF86 @ The Valve Museum

[Enzo]

If the hum is 100Hz, that is power supply tripple, but that doesn't necessarily mean a lack of filtering, The ripple charging currents can travel on ground lines and inject their noise into sensitive stages. Adding more cap doesn't help that. Making sure that sensitive stage is not sharing ground copper with the filters upstream can help.

[Malcolm Irving]

Yes. That’s an important point, but I think I’ve got it covered. Each power supply filter cap is physically close to the stage(s) it supplies and the 0V of that cap forms a local star ground. The local stars are then daisy-chained together using dedicated ground wires from local star to local star.
Just as a reference point, I’ve made some measurements on the only commercial amp I have to hand, which is a Blues Junior. It produces 7mV across the 8 ohm speaker with vol. and master full up, and 16mV with the ‘fat’ switch engaged. Not a like-for-like comparison though, as the BJ is what I would call a medium-gain amp, while my home-brew is a low-gain amp. On the other hand, my home-brew is ‘open chassis’ on the bench while the Fender has the benefit of a foil screen to close off the chassis.
Fully closing the screening on a chassis can make a big difference at these low noise levels. I had a Quad II hi-fi amp to work on (2 x KT66) and was surprised how buzzy it was on the bench, but as soon as I screwed the base onto the chassis it went 'silent'.

[Malcolm Irving]

Just tried a layer of aluminium foil to close off the chassis on my home-brew amp. Very little difference, just 1mV (approx.) reduction in noise output.

[Chuck H]

Just tried a layer of aluminium foil to close off the chassis on my home-brew amp. Very little difference, just 1mV (approx.) reduction in noise output.

What if you move the open chassis around the room until you find the noisiest spot and THEN employ the foil shield.?. Since we don't always have a choice about where to put an amp or how much EMF is in that environment this may be closer to a fair test.