Heater DC elevation question

[catalin gramada]

Hello. I have an independent 60V DC source intended to use as elevation for heaters. The amp is grounded with its inputs at chassis. Please let me know where is supposed to ground the source used for heater elevation for minimum noise ? (chassis input, first stage cap ...?) Thanks.

[xtian]

At the same place as heater winding is grounded.

[catalin gramada]

I don't have a rule in respect with this. Normally I do it by trial and error and choose a convenient point to chassis .May I do the same with a power source pls ?

[pdf64]

… first stage cap ...

If that means the reservoir cap, it probably wouldn’t be a ideal choice, as it may be the noisiest point of a typical amp’s 0V common arrangement.
But then it’s all relative, eg it’s no different really to referencing the heater elevation point to the output valve cathodes when they’re cathode biased, which is how it was sometimes done in vintage designs, and gets recommended for new builds / designs.
For my last 2 builds I’ve taken the elevation potential divider feed from the screen grid HT node, used a 220k / (47k//22uF//SIDAC) arrangement for the elevation point, and used the LTP HT node cap’s negative terminal as the 0V common return point. And they’ve had a negligible degree of hum.
The SIDAC prevents the elevation point being pulled to a high voltage, in case an output valve somehow fails in a way that would cause that to happen.

[catalin gramada]

Not at all. There are definitely some noisy points which may should avoided. I thought more like input stage filter cap...
But still you mentioned: there is no limit for max h-k resistance used as 20k or so for a 12ax7 which may put a ht divider in a situation to draw "signifiant" current (well depends of voltage wanna get)?

[Helmholtz]

What are the noise voltage differences between possible grounding points in your amp?

The source of heater noise is the distorted AC heater voltage.
With DC elevation the voltage between heaters and ground is the sum of AC heater voltage and DC bias.
Ground noise will just add to the heater voltage and assuming that the ground noise is in the mV range, it should be insignificant in relation to the heater voltage.
The DC elevation effectively suppresses noise from all AC components.

So I'd say the grounding point should be noncritical.

[pdf64]

That’s a very good point

[trobbins]

I can suggest that any additional noise component may be most noticeable if it entered in to the input stage - to minimise that, it may be better for the heater elevation to connect to the input stage 0V node (ie. where cathode bias and local B+ bypass join in a star) - that way the noise loop doesn't include noise from along the 0V bus. But that presupposes there is noise along the 0V bus, and that heater-cathode noise doesn't ingress from other more influential paths - such as a cathodyne or long tail pair configuration. It can be tricky to identify what the appropriate 0V connection is, or if it is even noticeable or dominant (given there are so many ways for hum and noise to ingress).

But first it is worth appreciating that it may take quite a poor valve (ie. quite low heater cathode resistance) to introduce a noticeable noise. The act of elevating the voltage difference between heater and cathode away from a nominal 0V differential is the principal reason for using this technique, as it intrinsically raises the heater cathode resistance complicit in any noise loop. So heater elevation can be used as a general design tool to avoid noise if a poor valve is used or ages into one. Alternatively, tube swapping is another way to replace a poor valve with a good valve.

Note that it is the voltage difference between heater and cathode that provides the benefit. Inserting an elevation voltage in an amp that has a cathodyne stage may end up reducing the heater-cathode voltage difference for that particular valve.

[catalin gramada]

Yes is a 12bh7 cathode follower stage with 200v into cathode. I tried to elevate the heaters a bit. +100v dc was too much and isolation was broked making an unusable tube. +70v it seems is ok but I wonder how much stres it provide ?