Daft double triode question.

[Tractorfan]

Hi,
A random pointless thought just popped up among my lonely little grey cells.
If the electrodes of a double triode were to be strapped together, each one to its twin then, electrically, you'd have a bigger single triode.
How, then, would the characteristics be affected? For better or for worse?
Would there be any advantage?
Cheers, Pete.

[ms660]

If the characteristics of the two triodes are equal then the new ra = ra/2 and the new gm = gm*2.

Lawrence.

[turretslug]

This was done in effect with some pre-war medium power triodes, ISTR the 6B4 was an example, with the results that Lawrence describes- short internal straps between two existing electrode assemblies saved on development and production costs. One snag that can crop up with external links is that what appear to be simple parallel straps can look like tempting tuned lines to frisky triodes, resulting in a push-pull UHF power oscillator!

[Radio Wrangler]

Capacitances would be doubled too, but with the halving of resistive impedances, turnover frequencies would stay where they were.

David

[turretslug]

If you're feeling bored, check the PSU circuit for the Collins R390 (not -A), this has two 6082 double power triodes strapped as 4 triodes in parallel for high current and low impedance, but there are more 47 ohm parasitic stoppers dotted around than one can shake a stick at.

[Syrinx1]

In addition, you should get an "ideal" 3db improvement in signal to noise ratio.

[Lucien Nunes]

The amplifier described in this post uses both halves of an ECC83 paralleled to drive a reverb spring tank. Circuit in the PDF.
https://www.vintage-radio.net/forum/...4&postcount=30

[bikerhifinut]

Or as has been the subject of Joe Bogs post here, if you need a single triode "ECC83" then simply use a 6AV6/EBC90.

Andy

[bikerhifinut]

Originally Posted by Syrinx1

In addition, you should get an "ideal" 3db improvement in signal to noise ratio.

Wasn't that technique used in certain low noise Valve preamplifiers to lower the noise?

Andy

[Aub]

Some years ago I saw, on the internet somewhere, a headphone amplifier which uses several paralleled ECC83s, with the headphones as the cathode load, so no transformer. I always
fancied experimenting with a set up like that.

Cheers

Aub

[Craig Sawyers]

The calibration procedure for AVO valve tests calls for a double triode strapped together. CV491 (AKA 12AU7 etc).

Basically you use a DC anode supply of 200V, grid supply of -9V and anode current of 16mA and work out the gm by incrementally changing parameters according to their procedure (in the range 4-5mA/V they say - so twice the nominal single triode value of ~2.2-2.5 mA/V)).

Craig

[barrymagrec]

Leevers Rich used a pair of ECC82s in parallel push pull for the bias / erase oscillator, in order to obtain the required power.

[GrimJosef]

Originally Posted by Aub

... a headphone amplifier which uses several paralleled ECC83s, with the headphones as the cathode load, so no transformer. I always
fancied experimenting with a set up like that ...

A more extreme example, albeit with output transformers, is Tim de Paravicini's EAR Yoshino V20. The picture (lifted from the internet, although I have worked on one) shows 15 double triodes in each channel. Three ECC83s are used in the small-signal stages, two (ECC82s) are used as drivers and ten (ECC83s again) are used in parallel push-pull in the output stage. The output valves are somewhat over-run but the clever and quite complex feedback circuitry and the split-primary output transformer (half in the ECC83s' anode circuitry and half in the cathodes) means that the measured performance can be really pretty good. I measured hum and noise below 0.2mV RMS (20Hz-30kHz into 8ohms) and total harmonic distortion ~0.05% at 1W out. At 20W out this had risen to 0.4%, but mostly low order. Given that the feedback doesn't include the output transformer the output impedance of 0.8ohms or so was pretty creditable too.

The whole thing was a little 'highly-strung' though. It was a bit sensitive to drift in the DC characteristics of one or two of the more critically positioned valves.

Cheers,

GJ

[Aub]

Wow looks great. I dread to think what price that is.

Aub

[G6Tanuki]

Yes, I remember a 'low noise' audio-preamp that used two halves of a double-triode strapped in parallel. Can't remember if it was for use with a ribbon microphone or a tape-recorder replay-head.

I also remember coming across a small Japanese VHF transceiver whose entire transmitter was implemented using the 6J6 double-triode.

One was used as the xtal oscillator/tripler, another served as a push-push multiplier, another [with both halves strapped] as a doubler, feeding into a neutralised push-pull power-amplifier to produce a few Watts of RF.

On the audio side there were 3 stages of 6J6 triode audio amplification, another one worked as a see-saw phase-splitter, feeding another one a push-pull modulator.

The 'spare' triode in this chain was wired as a diode and rectified a smidgin of the RF to drive an output-power/modulation-meter.

I guess someone got a bargain deal on a load of 6J6 !!

[Radio Wrangler]

The idea behind the noise reduction belief is that the noise of the two halves will not be correlated, whereas the signal contributions are correlated.

This is true, and each of the to triodes can be considered to be two half-sized triodes stacked one above the other. Correlated signal, uncorrelated noise again. This can be repeated endlessly, so giving an endless series of 3dB reductions in noise contribution... Oooo-Errrrr that seems wrong, the noise going down just by changing how we think about analysing it.

What really happens is that running the two triodes in parallel does not give an improvement in noise in the device, what it does do is lower the source impedance at which the best noise behaviour is given.

So multiple paralleled transistors are used in moving coil disc preamps in order to make an amplifier on its best noise behaviour when fed from a low resistance source, such as the MC cartridge. A single device would have been further away from the optimum environment for noise and so would have been noisier. A transformer converting the apparent source impedance would have been a suitable alternative in optimising the noise match without multiple devices.

David

[thermionic]

Just found this circuit on the net...

Essentially a strapped 6SL7GT ( ignoring the 47 Ohm grid stoppers )in the first stage of a phono amplifier.


Cheers.

SimonT.

[Radio Wrangler]

Yep, that's what they're up to.

You can model the noisiness of an amplifier as an effective noise voltage source and an effective noise current source at the input. Together these set an optimum source impedance. Feed the stage from a higher impedance source and you lose noise performance as In starts to dominate. Feed it from a lower impedance source and you lose again, this time with Vn dominating.

This gets harder at RF where you can't ignore capacitive and inductive effects, so all impedances become complex numbers, but, helpfully, you can plot noise figure contours on Smith charts and see what you're doing.

But, in a modern phrase that's getting sickeningly common, your cartridge is what it is, and it's the amplifier you need to suit it.

The sort of source impedances that most valves work best with tend to be rather high, so although the paralleled pair of triodes won't get you optimised, it still brings a real improvement by bringing you closer. You could, in theory, get the same effect by using a transformer to transform up the cartridge impedance to suit the valve. In reality this comes crashing down in flames. Transformers may be feasible for low-Z MC cartridges and low-Z transistor amplifiers, but doing one for these impedances and any attempt to cover the audio frequency range is not going to turn out well.

I mentioned RF noise design and Smith charts. People doing these things lear one very inconvenient fact.... the source impedance for a proper match, and the source impedance for lowest noise figure are not the same. They don't even have the same real terms.

Normally you'd be inclined to do a proper match on an input because the max power transfer theorem says you get the most signal, and so you just let the noise look after itself.... Unfortunately the noise grows worse than the gain benefit so you lose net signal to noise ratio coming out the output. <Insert favourite swear words here, robust ones>

There is another approach to paralleled valves, you can use one high power valve and get a lower optimum source-Z that way.

Welcome to the frustrating world of signal/noise optimisation!

David