Parallel tube stages

[BYacey]

Just from a theoretical point of view, If a number of identical class A tube stages are paralleled together, would'nt this increase the signal to noise by 3 db per stage? Thermionic noise being random noise would remain more or less the same, while the gain from each stage would increase the signal by 3db. Of course this would entail matching the tubes within reasonable limits.
Food for thought; perhaps not worth the effort. Any comments?

 

[PRR]

> If a number of identical class A tube stages are paralleled together, would'nt this increase the signal to noise by 3 db per stage?

No.

? Thermionic noise being random noise would remain more or less the same, while the gain from each stage would increase the signal by 3db.

No.

 

[Rob Flinn]

The output Z would be lower.

 

[NewYorkDave]

Two correlated signals of equal amplitude, summed, give a 6dB increase. Two uncorrelated signals (e.g., noise), of equal amplitude, summed, give an approximate 3dB increase.

The thing is, using two tubes in parallel results in a lower source Z and better current drive, but not significantly higher output voltage except when driving a load that's fairly low compared to the plate resistance of the tubes. So the noise of the two tubes will add in the fashion described above, but the output voltage--at least inside a preamp, where you're driving fairly high impedances--will not increase by a significant amount.

Paralleling tubes does make a lot of sense in preamp output stages, to better drive low-Z loads, and you see this done in many classic preamp designs.

 

[rlaury]

We can double the POWER by paralleling class A stages.
The noise figure stays roughly the same. In one way you look at it
the noise IS reduced. But at the SAME power and load, it's not.
If we can supply double the power (3dB), and the noise remains roughly the same, You can say the noise has been reduced by 3 dB. The output Z is halved.

If we try to double the power in a SINGLE stage the noise will also double.

I think this may be what your referring to.


RonL

 

[NewYorkDave]

That's a good way to put it. Inside of a preamp, anyway, you're generally not concerned with delivering power to a load. The output stage is a different story.

 

[Guest]

PRR, I LOVE duality, why you not do so?

[quote author="PRR"]> If a number of identical class A tube stages are paralleled together, would'nt this increase the signal to noise by 3 db per stage?

No.
[/quote]
YES,

If you paralel devices (with correspondent increase overall plate current)
their noise resistances are also paraleled.
If you connect two amplifiers to paralel, gain remains unchanged
(principle of compensation) but noise currents from each stage
quadratically superpones, and resulting noise (3 dB higher) is
transferred to voltage via 1/2 plate resistance (6 dB lower)
in result, dynamic range will be increased.

Paraleling devices are equivallent to ideal transformer, read Leach once more,
In typical input trafo we have 1:10 transfer ratio, with 1.5 dB transformer
related noise figure.

Nice school example: How many tubes with input noise resistance 200 ohms must be used to replace 1:10 transformer with noise figure
1.5 dB @200 Ohms input ?
Hint: lowest upper bound 10*10 = 100

Merry Christmas

xvlk

 

[NewYorkDave]

All right, Bill, is everything clear now?

:wink:

 

[PRR]

Paralleling decreases voltage noise only when power consumption increases.

Most of the noise of a vacuum tube is cathode dynamic resistance. Paralleling tunes parallels cathode resistance so yes: noise is reduced. However we can often get the same effect by running the one tube at higher current, or by picking a better tube.

Voltage gain generally will not change much in voltage-amp stages, unless they were badly overloaded to begin with. If one 12AX7 gives a voltage gain of 60, ten 12AX7 do not give voltage gain of 179. Total available power does increase, but this is getting power the hard way.

There is a small theoretical advantage to paralleling tubes at the same total plate power. Cathode resistance is 1/Gm, and Gm falls about as square-root of cathode power. So if we have 10mA to spend, and can spend in in one tube or ten identical tubes, each of the 10 tubes will run at 1mA and have a Gm about 0.316 times the Gm of one tube working at 10mA. So the net effect is Gm about 3.16 times higher, noise about 1.77 times lower, or 5dB lower. This is buying low-noise the hard way, after you consider the huge heater power required.

And while grid current noise is usually negligible for a single tube, when you get into dozens of tubes it is non-negligible.

Paralleling can lower noise. In practice it costs a lot. Before you start paralleling the classic audio tubes working at low current, try the same tube at a high current, or switch to a high-current low noise tube like 6BQ7 or 6DJ8.