El84 Max screen grid dissipation

[Andy Le Blanc]

Heres what I mean...

the problem with series screen grid resistors is that small changes in screen
current make large changes in grid potential.
when the screen current varies the screen voltage varies.

You end up with voltage swings as you push the tube, the operation isn't stable
your not exceeding current, your lighting up with voltage.

and you have no easy way measure or quantify whats happening
when you try to measure current over the series resistor or referenced to ground.

If you use a voltage divider, to source the screens, the current drop in the
divider stabilizes potential swings on the screens, and sets a voltage difference
between the plate and screen that you can measure.

I just did a traynor that had only one common resistor for the screens
The quick remedy was to add two more g2 resistors and leave the original floating.

[Gaz]

Thanks a lot for those posts, Andy! First off, to avoid confusion, Structo, the schem you posted is not what Andy is talking about, but simply a shared screen resistor before the individual screen resistors. What I believe Andy is talking about is more of a capacitive bypassed voltage divider, which I have a few questions about.

Andy, how do you determine the values of the divider (including capacitance), and what percentage of the plate voltage do you aim for?

When the divider is in place, how do you determine the value of the individual screen resistors? Are they even necessary? I see that they are drawn in your schematic.

Also, it seems like the resistors in the divider would be dissipating a good amount of heat, what wattage is appropriate? I know it depends on the voltage drop of course.

[Jana]

"irregardless" is not a word.

Love,
Jana (MA in English)

[Andy Le Blanc]

I'll have to repudiate...

start with say the common fender values, measure the difference between
the plate and g2 on the tube socket as you set up the amp bias under static conditions.

most old fenders will give you a g2 that is slightly positive with respect to the plate.

Use common values, 470r screen grids, add a 470r , per the marshal floating resistor example.

and from that junction put a 100k to 220k bleeder to ground .
so from the PS node where you would normally source the screens
you have a 3 to 5w 470r, then a bleeder to ground, and from here you source your screen grid resistors.
You can add another filter cap or not..

the bleeder doesn't have to be a large wattage, like the balancing resistors on a common PS filter stage.

increasing the resistor before the bleeder increases the voltage difference
between the plate and g2, giving you control over the difference between
the plate and screen voltages.

reducing the value of the bleeder increases the current drop and voltage regulating of the screen
You also have to raise the wattage to handle the current.

Again you measure the voltage difference between the plate and g2.
set it to where want subjectively, listening, but you'll have a voltage
difference that can be measured and repeated.

the old fender set up is up to about .5 v +, Ive seen it referred to in tube data
but it leads to issues, like the tube eating marshal of all places...
the famous fender tone, that fizz at clip...

You see it dealt with in other ways but usually with series resistors
to set the screen v, but as is stated you cant measure screen grid current
very easily with a simple meter.

This approach allows you to measure ,repeat, and mod circuits, even if you simply stick with series resistors as is common.

At least you can measure between the plate and screens.

[Gaz]

Thanks again Andy for the cogent reply. It's strange you don't see this approach more often, or even discussed for that matter (maybe it's a Hi-fi thing?). This method seems to be a middle-ground between adding a large series resistance, or on the other side of the spectrum, fully regulating the screen supply. I realize any method of lowering voltage at g2 will lose output power and 'that' sound, but it seems your method should be more tonally transparent than simply increasing the stoppers to high values and increasing the feedback, as you explained.

[Ian444]

Another possibility is a string of zeners instead of the bleeder resistor to ground. The max power dissipated (by zeners or bleeder resistor) would be at idle (static conditions). One thing I wonder about is should the preamp be powered from this node with the bleeder resistor, or from before that node, as the bleeder will pull the preamp B+ lower.

[Andy Le Blanc]

I think its not more common because the leading examples (fender) never did it
EVERYONE on the N. American continent starts with a fender, its like a bad rash.

There is no diversity of approach in the current DIY market either...
fender dumble wreck... whatever.... they all share the same design aspect

I heard a funny little distortion in one of my rigs, that fizz, even though every
thing with the thing was technically correct as a fender based 50w.

the only measurement that betrayed any usable information was the
measured difference between the anode and G2, and the screen was
slight less than half a volt positive to the plate, just 1 volt the other way
and it was a totally different critter. The reason for the slightly positive G2
is the position of virtual cathode in the newly developed 6l6 and related
family of beam tetrodes, its a technical hold over from the thirties.

It only pops up in a couple data sheets, its a design assumption.

If your a DIY, and looking a technology you don't fully understand,
It can lead to a wide array of other bad assumptions, IE: cloning amps

WTF........

At least the measurement point between G2 and plate can help
So if you got a tube type with a G2 voltage limitation, say you want to
convert a fender to run el34, the design center has a G2 max of 425v
even though the AB1 example is 450, and you stuck with a tube brand
thats notorious for a short life, like a bad set of jj... what do you do?

The plate will be fine, but big screen grid resistors are joke, you can only
assume that that'll work because you read it in a internet forum
and you can only assume what the actual screen v is because your
monkeying with the bias any way

if the plate are running at 500v where you've set your bias
you measure the difference to the screens and set the g2 v at 400v
There is no assumption that the plate are 500v
There is no assumption that the bias is say 70% of the published max
There is no assumption that 500- 100 = 400
set the g2 -100 with respect to the plate

No assumptions means happy tubes....

getting back to el84, set the screen v at known voltage with respect to the plate
if you know what the plate v is, even if your pushing the voltage....
set the screens within the published max, with respect to the plate v
what ever it is, at the bias point you've chosen, it should be happy.