screening sweet spots

[Andy Le Blanc]

I've been noodling with the screen grids of pentodes for a good while.

I've been using voltage dividers to set the screens voltage in relation to plate voltage.

So far with small signal pentodes ( 6sj7, 6au6, ef86,... etc) you can do it by ear with and adjustable pot., use your ears.
Its very easy to dial it in, the gain peaks and the tone fattens up, fairly simple.

With power pentodes it a bit more of a challenge, a lot depends upon how the screen supply is implemented in the circuit.

Are the screen tied together? Are there series screen grid resistors after the divider?

The screens are so far back in the circuit path, that apparent tone change is less of a concern.

So far it seems to matter more that the screens are negative to the plate.
So far to my ear some where -5v to -25v, and like I mentioned just how much separation between the screens there is in the circuit seems to have a greater impact upon tone than an arbitrary voltage setting.

Say you have the screens sharing a common voltage source and are physically tied together. This can be demonstrated in old fender, vox and marshal schematics...

The screens are tied, then in latter models, screen grid resistors float and separate the screens.

I stumbled upon this thru my own noodling by trying to "fix" positive screens by adding bleeders after the screen grid resistors, giving the screens on both sides of a P-P pair their own voltage divider, keeping the separation.

Thru attempting to replicate data charts for tube types you see the various conditions for class A and AB P-P circuits and the conditions that will provide the most watts with the least distortion.

For guitar amps I noticed that if the screens voltage is set in accordance with these figures, with the screens strongly negative, the top end of the tone becomes hard, harsh when the amp is pushed

It seems that for guitar, (to my ear anyways) it seems better to keep to the class A, AB conditions vrs, AB1 AB2... for a non-master volume amp where your relying upon the natural tones, the natural distortion of the amp.

These are arbitrary observations of course...

Series screen grid resistors are de-stabilizing and can put a P-P power side into dangerous ground with large signal swings, but.. they separate the sides of a P-P circuit with a common voltage source.

I think that the most flexible approach may be to use separate voltage dividers for screens on each side of a P-P circuit, set sufficiently negative to accommodate a range of bias setting that might be used for the tube type.
And then quite simply an adjustable pot between the screens on both sides of the P-P circuit to mimic a "tied screens" condition and then a variable separation of the screens to allow the circuit to self balance around the screen.

thinking out loud of course... time to noodle on me bench.

[David Root]

With small signal pentodes, it's normal practice to have the screen much lower than the plate, and what little work I've done with them confirms tonal changes too. Merlin's preamp book has a good section on these tubes.

Did you examine the 5879 at all? That's a nice preamp pentode that I've used and easier to work with than the EF86, not to mention much cheaper.

With power pentodes, are you implying that, for example, a Music Man topology EL34 with 700V plates and 350V screens will have a hard-sounding top end? Or an Ampeg V9/SVT? I've not yet built this topology so I don't know, but I know Dumble did some amps using Music Man iron and his 300W bass amps used 700V plates I believe.

[jazbo8]

It seems that for guitar, (to my ear anyways) it seems better to keep to the class A, AB conditions vrs, AB1 AB2... for a non-master volume amp where your relying upon the natural tones, the natural distortion of the amp.

Not sure what you meant WRT the classes of operation, there are very few guitar amps that run in class A or AB2, most operate in class AB1, so are you suggesting to run them in class A instead?

Series screen grid resistors are de-stabilizing and can put a P-P power side into dangerous ground with large signal swings, but.. they separate the sides of a P-P circuit with a common voltage source.

Lost me again, aren't the screen grid resistors there to protect the screen grid during large signal swings? Without them, the screens take even more beating. Or did you mean something else?

[roberto]

Are the screen tied together? Are there series screen grid resistors after the voltage divider?

One option you suggest to try is to place a grid resistor in common for two tubes of the same side of the PP, and after that a voltage divider and the screens connected directly to the voltage divider. What this would change tonally?

How much separation between the screens there is in the circuit seems to have a greater impact upon tone than an arbitrary voltage setting.

you mean on the screens of the same side of the PP or between the 4 screens?

The screens are tied, then in latter models, screen grid resistors float and separate the screens.

clear, nowadays all the amps have tubes with dedicated screens resistors.

Series screen grid resistors are de-stabilizing and can put a P-P power side into dangerous ground with large signal swings, but.. they separate the sides of a P-P circuit with a common voltage source.

not clear. The purpose of those screen resistors is to limit the current of the screens.

Thank you in advance for your time!

[Andy Le Blanc]

the problem is that under some conditions, less than optimal loading, when you push the amp,
what the screens supply sees, actually becomes un-loaded, voltage tends to rise and the drop across the series resistor
and actually increases the voltage at the screen-grid even further.

the screen grid resistor is exasperating the condition.

look at old scheme, what do you see?

power pentodes where designed for radio receivers, the plate voltages where generally lower and the screen where often connected directly to the same tap as the opt, this puts the screens positive to the plate, which is what they were designed for.

with an increasing desire for watts and fidelity, you see and increase in plate voltage and the screens are either connected to a voltage divider or a tap further down the power supply, to keep with in the voltage limitations, with the front end of the circuit providing the voltage drop as the current is utilized.

but the trend is for the screen to be simply tied together sharing the same voltage source.

look at this old gibson, the screen are tied and using the same tap as the anode circuit.
But also notice that the field coil speaker doubles as a choke provide regulation to the circuit

[Andy Le Blanc]

fender princeton were a familiar hold over, the screens are tied.

you can see the economy expressed in the design.

no series screen grid resistors.

think the tone difference between a deluxe and a princeton.

[Andy Le Blanc]

and here is the deluxe, this shows the trend that fender established, and everybody has mimicked for decades.... the pre amp, the invertor, the feedback... and the power side

there are series resitors after a choke(think field coil speaker).

there is a big tone difference between an amp with the screens tied and with series screen grid resistors

its one of the differences between ampeg and fender, many ampegs had tied screen grids

[Andy Le Blanc]

heres an Ampeg example

[Andy Le Blanc]

and there are examples that show a wide variety of different approaches of the screens supply.

tied, connected to the ct of the opt

tied, after a voltage divider, connected to the ct of the opt

tied, after a series resistor, connected to the ct of the opt

split with series resistors, but tied to the ct of the opt

you can find and see just about every example you can think up.

and nearly every brand of amp goes thru a similar evolution, everybody made a fender at some point as product competition drives the market evolution...

some notable hold overs were ampegs... magnatone... fender princeton

every design has its own tone character, thats really what your after.

I find the tied screens tend to be more solid, more up front.

it depends upon the voltage gradient that you set up over the pentode..

the screens voltage relative to the plate...

and whether or not, or how much you allow the screens to float independently of each other, between the legs of the push pull circuit.

[roberto]

I find the tied screens tend to be more solid, more up front. it depends upon the voltage gradient that you set up over the pentode..

I noticed that on an old 50W Marshall without screen resistors. Sounds more in front of the mix. But I'm worried about reliability.

[jazbo8]

the problem is that under some conditions, less than optimal loading, when you push the amp, what the screens supply sees, actually becomes un-loaded, voltage tends to rise and the drop across the series resistor and actually increases the voltage at the screen-grid even further.

the screen grid resistor is exasperating the condition.

Setting aside how the amps sound for the moment... are you suggesting that having screen stop resistors is worse for the amp under some conditions? So under those conditions, it would be better NOT to have them at all?

[unbiased]

Regarding the Fender Princeton, screen nodes need to have the current limited. In Fender amps that use a choke, the screen tap is right off of the choke. The choke has very low resistance and therefore cannot provide enough current limiting so each screen has its 470 ohm resistor in series. The BF Princeton (and Ampeg Reverb Rocket, et al;) does not use a choke in the power supply. Instead they have a large dropping resistor in the power supply chain which conveniently limits the current to the screens so the series resistors are not necessary. I have worked on BF Princetons in which the 470 resistors have been added and I haven't heard a difference in tone that I can attribute to their presence. The difference in the sound of the BF Princeton versus the BF Deluxe is more likely due to the cathodyne PI in the PR versus the LPT PI in the DR.

[SilverFox]

I'm thinking of this in terms of the differences in change in the state of a sort of floating filter, created in the output section resulting from the alternative approaches to screen connections.

Isn't there going to be a "change in state" in the output section through the alternative screen approaches? This state change will be present for a time, x, of the cycle, an RLC lead-lag effect. So in a sense the alternative methods create a change in what amounts to a variable filter within the output section, thus adding subtle variations of tone. As a system, the changing signal of the musician will create a uniquely nuanced final filter in the output section for any particular moment- Within limits. The different approaches then only set up the preliminary unique environment.

If this is the case, changing values and listening to the results should be a rather mind boggling experiment. I think what's needed here is calculus to be able to predict any particular outcome. What sounds good to your playing style may have a subtly different quality to someone else's signal sent through the amp. Touch sensitivity? Perhaps analysing a design that is in the neighbourhood of what you think is good sounding is a good starting place.

I'm only thinking this through myself so don't take this as a criticism. Just an observation and thought process on my part. It's got me thinking now...

silverfox.

[teemuk]

This topic somewhat tickles one of my pet peeves...


One of the "tubey" characteristics raved about is their dynamic "gain compression": When signal swing begins to reach its limits the gain of the tube decreases and you get some nice gain compression instead of rail clipping.

But in reality, with typical power pentodes/beam tetrodes pretty much the only way to get "gain compression" is via screen voltage as pentode/beam tetrode's gain itself is rather independent on plate voltage voltage variations.

So, with "stiff" screen supplies we get almost no gain compression at all. Likewise, "stiff" screen supplies don't react quickly to B+ sag either.

Then we need to separate the behavior at screens to two categories: In practice, are they bypassed or not. Bypassed screens (such as those connected directly to screen's DC supply (independent on how the supply is constructed) will only react to variations within that screen supply. Usually this takes place s...l...o...w... enough that compression won't apply to signal transients. A "looser" screen supply could react to things like B+ sag and higher screen current draw during overdrive but it would do so gradually.
Non-bypassed screens (such as those with individual screen resistors) will generate an additional AC signal to the screen (in pretty much similar way to plate) and will therefore introduce much quicker dynamic compression characteristics. But to make them blatantly obvious you effectively need to use screen resistor values in the range of 10K or above. Anyway, much higher resistances than "standard". 500R vs. 5K... pffftt.. Won't make a difference. The resistance must be high to generate a substantial AC signal at the screen.

Yet, take a look at revered Fender "tweed amps": Direct referenced screens and the reference is generated with rather stiff RC or LC filters. Aside for "power attenuator" schemes there are practically no guitar power amps that would employ "higher than standard" -resistances in individual screen resistors. Well... Mesa's "DynaWatt" enhances sag characteristics with "looser" screen supply, but they mainly do it to squeeze more burst power marketing watts from their amps. A common modification to them is a "stiffer" screen supply. Additionally, "pentode characteristics" (no screen NFB) are sought after instead of "triode characteristics" (plenty of screen NFB).

So... IMO, this "gain compression" in the output stage seems to be one of those tube myths that rarely apply to any real guitar amps out there.

What say you?

[matt h]

(deleted)