Do EL34s in pentode mode PP AB1 sound better with lower screen voltage?

[ray_moth]

EL34s are supposed to be happy with a screen voltage up to ~ 420v but they can, of course, be run with lower screen voltage than that. Some people believe that power pentodes sound better at reduced screen voltage but I've never seen any convincing argument as to why that should be so.

I'm wondering, in the case of EL34s in pentode PP Class AB1, with B+ of 420v, whether there is anything to be gained by using a lower screen supply voltage and, if so, what sort of voltage and why? Anybody have any thoughts on this?

 

[artosalo]

Some people believe that power pentodes sound better at reduced screen voltage...

I would believe this, if one also told that the total distortion was observed to be reduced, or something evident and measurable improvement was achieved.

....but I've never seen any convincing argument as to why that should be so.

In general I have seen very little if any arguments why - for example - Tungsol 6L6 sound better than RCA 6L6 and similar.

I have noticed that those hobbyist who maily speak about "how it sounds" do not usually show any test results or do not make tests.
All the above only concerns HiFi-amplifiers. Guitar amplifiers are fully other story.

And those who make lot of measuremets with their amplifiers do not speak so much about "how it sounds".

 

[HollowState]

I'm wondering, in the case of EL34s in pentode PP Class AB1, with B+ of 420v, whether there is anything to be gained by using a lower screen supply voltage and, if so, what sort of voltage and why?

As you should already know, the screen grid of a pentode controls the transconduction every bit as much as the control grid does. Assuming all other potentials remain stable, lowering the screen voltage will reduce total current through the tube. If there was a DC imbalance between the two tubes through the output transformer, then the total imbalance would be similarly reduced easing possible core saturation effects.

Also, reducing the screen voltage will lower the gain of that stage since Gm and cathode current are directly related. But to me, this would be a waste of potential power, although it would increase tube life.

Now, if there is feedback around the output stage, global or otherwise, a reduction in gain should produce a corresponding reducion in feedback. Some folks think this makes an amplifier sound more dynamic. Albeit with less power.

With reduced screen voltage we could, in the case of fixed (or cathode) bias, lower the grid bias voltage to compensate for the lower tube current. This, in effect, would make the control grid easier to drive and so effect input sensitivity of the amplifier overall. But still with a reduction in power out.

I think it just comes down to designers choice. What you're willing to give in order to get something else. It's always a compromise.

 

[M Gregg]

Just for interest.

I made an UL EL34 amp and tried Pentode, triode, UL. Then I was told quote; why don't you do a "true triode" connection. So I connected all the grids (not used for signal) via resistor (to stop grid current) to anode. It sounded very different, I did not take any measurements. However the sound stage became "very" clear. Obviously I had to re-bias the tubes. Just in contrast to the question!

Regards
M. Gregg

 

[M Gregg]

The problem with changing potential on the grids is flash over!
This did run Ok though with no problems at the time!

 

[kach22i]

My Golden Tube Audio SE-100 was running at 350v for a while from neglect/age. Thought I would be a good boy and bias them to the factory 500v, which quickly fried them (Electro-Harmonix).

My EL-34 Tung-Sol's are now run at 450v and sound great.

Don't know if it was the age or the bias, but 350v had no kick in the bass.

 

[Merlinb]

I've never seen any convincing argument as to why that should be so.

At reduced screem voltage you eventually reach the point where the load line passes above the knee of the Vgk curves, and the harmonic distortion becomes basically the same as that of a triode (mainly H2), which may have something to do with it.

 

[artosalo]

At reduced screem voltage you eventually reach the point where the load line passes above the knee of the Vgk curves, and the harmonic distortion becomes basically the same as that of a triode (mainly H2), which may have something to do with it.

If this were real, would it also be measurable ?

 

[kach22i]

At reduced screem voltage you eventually reach the point where the load line passes above the knee of the Vgk curves, and the harmonic distortion becomes basically the same as that of a triode (mainly H2), which may have something to do with it.

I've read that the lower the bias the better quality the sound, and the higher the bias the more power but less quality.

Now I know why that was written.

 

[Merlinb]

If this were real, would it also be measurable ?

With a spectrum analyser, certainly.

 

[Merlinb]

I've read that the lower the bias the better quality the sound, and the higher the bias the more power but less quality.

That's basically another way of saying class-A sounds better than class-AB, which sounds better than class B. ('cause power efficiency and crossover distortion increases with class of operation)

 

[artosalo]

With a spectrum analyser, certainly.

Have done this ? If you did, please let us know the results.