Parallel operation of triodes.

[ashok]

What's the difference in noise performance of two triodes in parallel ( Anode, grid and cathode connected directly in parallel ) and one with a common anode load and separate cathode resistors with bypass capacitors ?
Unless the sections are very closely matched , the direct connection will not ensure identical operation of the two sections. Which is a better way to operate in 'parallel' ?

Conrad Johnson had a design with 12 tubes in parallel. How did they do it ?
IIRC some designers claim that parallel operation of tubes ( or even SS chips) sound worse than single device operation.

Is this observation confirmed now ?
Cheers,
Ashok.

[Wavebourn]

I use paralleled 6L6 in P-P for 25W of power. Sound very clean.

[glennb]

2 in parallel will give half noise and double the gm (gain). The usual method is to tie the G, K and A directly together. I'm not sure of the good or bad effect of having separate cathode bias circuits.

[ashok]

The gm is doubled but I think the noise improves only by 3db ( 20*log (n^0.5) where n is the number of paralled devices) .

Any info on the Conrad Johnson preamp (ART)? It appears to have 10 paralled sections . Has anyone heard it ?

Cheers.

[SY]

Ashok, that's correct, the noise of each section is uncorrelated, so you only get 3dB noise improvement from paralleling. Separating out the bias won't really change that.

There's no reason from an electrical standpoint that a parallel arrangement should sound worse, but I admit it's entirely possible to cause issues because of layout, grounding, and the like when doing that with actual tubes and wires.

[AndreasS]

Originally posted by ashok
Which is a better way to operate in 'parallel' ?

Depends of the tubes gm in the sum. Separate cathode resistors increases the operation point stability (decreasing gm for DC).
The "borderline" is IMHO a gm > 10 ma/V for the tubes in parallel.

Andreas

[coresta]

IMHO, paralleling tubes equals increasing the max power, the slope, and reducing the internal impedance (in a rate of 2 for tube doubling) . But the "single plate" of a 300B for example is by itself a twin plate as the 6080 is one "right" anode , one "left" anode together tied , around a single cathode

[kevinkr]

I used up to 4 sections of 12AX7 in parallel in some of my early phono pre-amplifier designs, this in and of itself resulted in about a 6dB improvement in noise performance.

The design only proved to be tolerant of mismatch between sections when individual cathode bias resistors were used. (higher distortion and lower gain without individual biasing)

The source impedance of that first stage was 1/4 of what it would be with a single tube meaning that the passive eq network could be scaled downwards accordingly which also resulted in some (significant) improvement in noise performance.

Sounded very good actually, these days I just cut to the chase and use triode connected D3A instead - sound even better and far quieter than even 10 or so 12AX7A sections in parallel and with a lot less miller capacitance - the downside of paralleling so many devices.

edit

[richwalters]

Parallel operation of o/p stages is highly recommended, the near halving of output transformer Z reduces copper, improves b/w and easier to design o/p transformer. Related to another thread, with parallel o/p concept, UL operation will completely outstrip triode performance. Naturally the pSU becomes the other issue as stage current goes up and larger chassis.

Parallel tube operation of the p-p output stage doesn't quite follow the "halving Z rule" for optimum output stage performance. Shall have to read up on this.

richj

[Gordy]

* revives thread *

1) ...but surely tubes have such wide tolerances that two or more tubes of the theoretically same type in parallel will all be exibiting (slightly or widely) different characteristics, so will that increase distortion?

2) When paralleling tubes in a line stage (such as is done by CJ) which parameters require the closest matching for best objective performance of the completed stage?