Screen resistors - please weigh in

[novosibir]

Ok, heres what i found. with 3k verses the original 1k per tube, the tone changed in the following ways.

PROS: smoother top, more compress top.

CONS: loss of volume, loss of highs

This matches my experiences, as years ago I've tried a huge rheostat in front of the independent screen grid resistors in the first series of my British Purist:

[img:800:533]http://www.larry-amplification.de/briti ... pur096.jpg[/img]

The rheo was 2.5K/50W and got an additional filter cap between it and the independent screen grid R's.

With the rheo on zero the amp was an aggresive beast, which was cutting through any band mix.

With the rheo on full resistance it was nicely smooth and round sounding, more kinda 50W amp than a 100W in dynamics and really nice for playing at home - alone. But no longer useable for live on stage - it has lost its teeths and you hardly still couldt cut through the band mix with it.

Larry

[novosibir]

Here still another, more detailled take of the rheo in the amp:

[img:800:533]http://www.larry-amplification.de/briti ... pur091.jpg[/img]

Larry

[roberto]

Thanks Larry for the explanation. I've seen that rheostat on your website and now I know how it works. Great! The added supply cap it's something around 47uF?

[novosibir]

The added supply cap it's something around 47uF?

The 1-st screen filter cap is a 50µ (below the board), the 2-nd after the rheo is a 20µ/550V from TAD.

Larry

[roberto]

Great info! I will try it soon! Thanks again!!

[Structo]

Back to the subject for a moment.

I remember when I was tweaking my D'Lite that Scott had mentioned that:

"You may not hear much of a difference swapping screen resistors from 470 ohm to 1.5K, but you should feel a big difference in the bass. I can feel a big difference between 470ohm and 1K! The bigger the screen resistor, the mushier the bass."

If I remember correctly I used 500 ohm screen resistors on mine.

[David Root]

This is slightly off-topic, but I have seen a diode in the screen supply recommended from the hi-fi world, with or without a screen resistor.

The proposition is that the the diode (set up to pass DC to the screen) allows proper DC screen supply but blocks wastage of AC signal into the screen circuit, forcing that onto the plate where it ups the output power.

This doesn't sound like it would do much for touch, but I'd be interested to hear what folks think about that, especially in a distortion situation i.e. cranked, no master vol.

[Ears]

This is slightly off-topic, but I have seen a diode in the screen supply recommended from the hi-fi world, with or without a screen resistor.

The proposition is that the the diode (set up to pass DC to the screen) allows proper DC screen supply but blocks wastage of AC signal into the screen circuit, forcing that onto the plate where it ups the output power.

This doesn't sound like it would do much for touch, but I'd be interested to hear what folks think about that, especially in a distortion situation i.e. cranked, no master vol.

Wouldn't that just rectify the signal dependent component of Isc?
Edit: Forget that, I guess the question isn't applicable in push pull.

[Ears]

http://www.valvewizard.co.uk/
The above site has published some characteristic plate traces for EF86 illustrating some of the effects of differing screen potentials and grid-stopper resistor. Interesting to see how the knee is modified by the grid stops.

[Andy Le Blanc]

The proposition is that the the diode (set up to pass DC to the screen) allows proper DC screen supply but blocks wastage of AC signal into the screen circuit, forcing that onto the plate where it ups the output power.

the idea isnt new..... Ive see a couple articles in Glass Audio and something
by Nelson Pass..... about placeing diodes throughout a power supply instead
of resistors to decouple and isolate two amps working from the same PT....
or supply voltages to reduce ripple... and to control large current swings

the diode would decouple the screen potiential from the rest of the power supply
it would "prevent any backward flow of'... current caused by a large current pulse"

you could place a diode before the screen grid resistor..... or after the
plate tap to keep large current swing from effecting the screen or pre....
but sag in a guitar amp is desirable..... if it was only applied to the
screen grids ... youd have to model the valve as to see how it affects the tube.....

the GA article quotes "Electronics Circuit Designers Casebook" , Mcgraw-hill, Takao Kawabata.....
saying that when used in a power supply "the time constant of R1C2 is made desirably infinite"
if applied to just a screen..... R1 would be the screen grid resister after the diode and in series.....
and C2 would be the capacitance of the tube ... it could be trying compare apples and oranges but it sounds like it might be useable....

[Andy Le Blanc]

well thast a neat trick..... just tried placing a diode before the screen grid resistors
it did not hurt the tone...... I was expecting gobbs of non-linear distortion
it did change the amps presentation..... it seems to have cleaned up the distortion a bit
changed the detail in the cranked tone...... a really simple mod
cant wait to gig the amp..... then Ill know if its really worth it....

[Andy Le Blanc]

diodes on the screen grids does effect the definition..... although a guitar amp
may not be the best test bed...... seems to clean up and define ...
its like going to a mu-follower after listening to a simple srpp.....a change in clarity

I tried a single diode to both screen grid resistors...pretty sure that placeing a
diode for each screen resistor will increase the effect.... also that glass passivated
or super fast recovery diodes would be the better choice.....this mod is good for the details
its more of a hi-fi application and may be wasted in a guitar amp....although
for the jazz player it may be a good thing ..... clean detail.....

Ive been playing with the value of the screen grid resistor too....
one thing Ive noticed is that the closer the tube is run to maximum dissipation
the less the apparent effect of the value of the screen resistor is.....
so the operating point of the tube and the size of the resistor
are conjoined.... I tried 4.7k for giggles and got the expected compression
which became less apparent as the bias was reduced and the static dissipation
was raised to maximum for the tube type (EL34).......

[barfoden]

Old thread, but here are my thoughts..

If the tubes are operated within specs in terms of max plate/screen voltage, from the same supply line, and biased at sane levels (60% of max at idle for a grid biased amp), with a small resistance (100-500 ohm) in the B+ screen supply line,, the value of those screen resistors should reflect the tubes ability to draw screen current..
Beam forming tubes (6v6 and 6l6) draw significantly less screen current in relation to plate draw compared to pentodes such as EL84 and EL34 tubes..


Compare a EL34 to a 6l6GC in a high voltage class AB fixed bias setup. The old datasheets will tell you that a EL34 will draw around 20% screen current in relation to plate current whereas the 6l6GC will draw ~10% screen current in relation to plate current.. The tubes are in the "same" power class and a 6l6GC is seen often with 470 ohm resistors and the EL34 with 1K resistors..

A pair of EL84 will draw on the screens a total of 22mA for 17 watt output from two of them in cathode bias (datasheet). This is the same as for the pair of 6l6GC tubes at 450V plate, 400V screen in grid bias for 55 watts of output.. Vox uses 470 ohm screen resistors in their modern 2xEL84 amps when run at 325V plate in cathode bias..

The problem starts when people run their tubes out of spec..

I saw an article where a a pair of grid biased EL84 tubes were run at 368V plate and just slightly lower screen voltage.. He had to add 2.2K - 2.7K screen resistors to prevent the screens from lighting up when pushed. But he also grid biased them far to hot at idle.. I think it was around 85% of max plate dissipation at idle..

There is a reason why Carvin biases their Bel Air at 75mA total current across the standby switch for four grid biased EL84 tubes.
That is 18.75mA pr tube if we ignore the 12ax7 preamp draw..
Multiply with the crazy high plate voltage (~430V) and you get ~ 8 watts pr tube. Assume 10% at idle is screen current and you have 7.2W. This is around 60% of max dissipation on the plate at idle.. A little less if you include the current draw from the 5 12ax7 preamp tubes..
This amp gets away with a 350 ohm B+ dropper and 350 ohm 5 watt individual screen resistors on each EL84.

This amp has tons of headroom for a 4xEL84 platform.. And amp tech measured ~35W clean with just 1% harmonic distortion. The peavey classic 50 uses the same approach.

My normal assumption is that with a typical guitar amp setup with the screen feed downstream the plate supply with little resistance between the plate and screen nod you can get around the sum of the plate power rating in output from a colder grid biased amp with a high plate voltage..
(assume the amp has diode rectification and a strong power supply)

2x6l6GC (2x30W plate rating) = 60W output
(fender hot rod deville 490V plate, choke 485V screen)

2xEL34 (2x25w plate rating) = 50W output
(take you pick of marshall plexi amps with voltages around 400-450V and a screen choke).

[Andy Le Blanc]

There is a relation ship between the plate and screen grid and it changes with the operating point.
Where you intend to bias the rig effects the design you fall upon in the end.
I like the tone of cathode bias power side biased to max plate dissipation.
And that effects how the screen grids are set relative to the plate.
You have to be careful with the assumptions you make around fixed bias.
Some designs have to be biases notoriously cold to maintain the screen grids.
If you bias hotter the plate voltage drops relative to the screen under static condition,
and when you apply signal the condition changes, the plate swings first to "0" and then positive,
setting up screen current, distortion and fizz, shortening tube life.
If you bias cold the way the screens like it, the amp sounds "hard"...

As long as a design keeps within the tube max limitation your OK.
The issue becomes, that the voltage gradient across the valve is more dynamic than simple assumptions,
or a focus upon bias and screen grid resistors as the end all and be all of design set up.
You also have to look at how those assumptions impact the over all design, and design accordingly.

Data sheets are misleading in that they do not inform you how the voltages were achieved,
only the idealized operational results under those conditions.
Screen grid resistors are the cheap way out for a manufacturer.
You also have to be care full where you reference your measurements.
I stumbled upon measuring the difference between plate and screen.
It was a different ball game as soon as I did, it made it much easier to observe and maintain screens relative to plate
and how it was effected by bias .

The tube type no longer mattered, except for idealized loading, as long as the voltage relationships across valve are maintained,
you get the same result no matter what tube type or operating point you choose.

You end up setting the screens as you would the bias for the operating point, loading and tube type, for your own desired result

[Roe]

I prefer a flying resistor in my jtm45, although I'm using a 470R instead of the original 1k. I tried bypassing the flying resistor but the sound is too mushy and it doesn't sound righy in my 45 running gold lion kt66s at 6k6 and 440v with a RS OT