Split supply, arrangement of gain and phase splitter tubes

[DrewP]

Due tothe vagaries of the power trannies I have available, my el84 amp will have twin power trannys for B+, even though it's an integrated on a single chassis, not monoblocks (bias supply for outputs and the heater supplies will be shared).

In wiring up the 6CG7's for the input tube and phase splitter, have two options:

1- Use one tube per channel, feed a single positive supply to it and use it to run the gain and splitter in the one valve.

2- Run one tube per block (gain block and phase splitter block) but feed the two supplies separately to both sides of each stage.

In theory (theory only) option two sounds preferrable from a channel balance point of view as it would be expected that the two halves of a tube would be fairly closely matched.

That said, were this the case, high end monoblocks would invariably be unmatched sharing no tube halves whatsoever, so the theory seems off before I even start.

Separating the stages also gives the option of tweaking amp behaviour by changing just the gain tube from a 6CG7 to something different, instead of having to play with both input tubes in order to get a single change.

So: separate tube for each stage (and maybe some crosstalk) or separate tube for each channel (and maybe some inbalance)?

[SY]

it would be expected that the two halves of a tube would be fairly closely matched.

Reality does not match your expectation. There are some tubes with guaranteed section-to-section match (e.g., 6SU7), but they are the exception. Most tubes do not have any better match between sections within an envelope than between sections in different envelopes.

Any reasonably designed circuit should be able to accommodate the normal tube-to-tube tolerances. If not, you have two choices- redesign or manually match tubes. The former is the better path, IMO.

[richwalters]

You will probably get no better and the most realistic match from the 6SN7 breed .....a tried and tested medium mu tube.
All the others from A to Z will vary from batch to batch and that's worse to shocking with pentodes: I have many good, hardly used tubes from power to signal that simply just aren't usable as the characteristics are so wild out from norm.
One is my weapons is to CCS the working stage and add a voltage clamp to fix the constant current without upsetting the impedance; this can reduce the rejects to force a tube to work within fixed parameters. A obvious role for wildly out pentodes is to use them as triodes.

richy

[DrewP]

Many thanks all. Given that there seems to be nothing in it except a miniscule risk of crosstalk, in the interests of neater layout and easier faultfinding for my newb brain I'll use the two halves of each tube for both channels of the separate stages.

Looking at the connections I need to make at the bases it certainly appears more straightforward and easier to understand what I've done if I do it that way.

Far neater install of coupling caps between stages too.

[Merlinb]

Originally posted by DrewP

So: separate tube for each stage (and maybe some crosstalk) or separate tube for each channel (and maybe some inbalance)?

Cross talk in dual triodes is generally better than 50dB, so you should be fine whatever you choose.