Regulated screen grids in guitar amp

[catalin gramada]

For sure a common resistor for all grids induce a balancing effect in power section The output swing more simetrical onset clipping. That may be responsible for reduced distortions maybe and different response translated in sound.

[Helmholtz]

Screen grid compression is best scoped using a sine burst signal.

[J M Fahey]

Yes, and burst duration must be, say, 5 to 10 seconds long, since many time constants involved are *slow*

[catalin gramada]

If a supply node derived from the plate is used (with undecoupled screen grid resistors) does it make dramatic diference if 10ohm/1000u or 100ohm/100u used ?

[Helmholtz]

If a supply node derived from the plate is used (with undecoupled screen grid resistors) does it make dramatic diference if 10ohm/1000u or 100ohm/100u used ?

That would be just one isolated time constant.
In reality the filter stages interact and time constants are modulated by plate and screen currents, which represent varying loads.
While your examples result in the same RC product, they result in different source impedances.
So yes, I'd expect a significant difference.

[Helmholtz]

Yes, and burst duration must be, say, 5 to 10 seconds long, since many time constants involved are *slow*

If it's not about really measuring attack time, a shorter burst length of maybe 100ms should suffice to evaluate the effect.
To a player the difference shows in attack response and a guitar signal decays fast.
To also see the release, the signal should be switching between high and low level.

[J M Fahey]

For a better picture of voltage variation, which directly affects transconductance/gain, a somewhat longer burst duration is needed.

We do not only want to see the attack envelope by itself but also to what voltage does it stabilize to after some time.

Just checked, time constant for a typical screen supply node made out of a 1k "choke substitute" and a 50uF capacitor to ground is 50ms

Since voltage at time constant is not the final voltage by any means, but only 63% of it, and then change becomes slower and slower, I´d watch envelope on a scope screen for a quite longer time, just to make sure.

[Mick Bailey]

To clarify, I can fully appreciate the difference between a stiff or regulated screen supply and one that allows some dynamics. I thought that the comparisons that made a big difference were those described in post #9.

Out of interest, the regulator in #12 is configured as a constant current source.

[Helmholtz]

There are 2 different screen compression effects mixed in this thread.

1) First one is screen compression by sagging of the screen supply node voltage.
This effect is slow, has attack and release time but doesn't alter the waveshape of the signal.

2) Second one is instant compression due to voltage drop across a non-decoupled, individual screen resistor. This type does act on the waveshape of the plate signal.
As a tube has an intrinsic expansion characteristic (gm increases with plate current), a non-decoupled screen resistor tends to linearize the signal.
This is due to the AC component of the screen current, which makes the screen voltage vary in opposite phase to the plate signal, resulting in NFB.
When 2 tubes at opposite sides of the OT primary use a shared screen resistor, this effect gets spoiled.

[Helmholtz]

Calculating individual screen resistor power dissipation:

While this has been discussed before, I thought it might make sense to repeat the procedure here.
The resistor is heated by the screen current which looks like the plate current and both look like the half-wave rectified signal (+some idle current) at large output (class B). I.e. essentially half-sine pulses with a sine signal.
The screen current produces a voltage drop having the same waveshape as the current.
The voltage across the resistor thus has a DC and an AC component, each of which separately produces power.
To find total dissipation, DC power is calculated from the (averaged) DCV drop and AC power from the ACV drop. Both results are added for total power.
Measurements should be taken at full output.

[Helmholtz]

Continuing my post above.
Formulas for screen resistor dissipation:
Ptot = Pdc+ Pac = Vdc˛/R + Vac˛/R (current is not needed for calculation).
Vac measurement requires a "True RMS" meter.

When a shared screen resistor is used for 2 tubes at different sides of the primary, things get tricky (though the above formula using measured voltages works the same.) as the tubes draw screen current at different times,
Means that total RMS current is not simply doubled.
The AC part of the voltage across the resistor now looks like the full-wave rectified signal.
The RMS value of a full-wave rectified signal is 1.41 times the half-wave RMS voltage, so Pac is doubled.
The DC currents just add, meaning that Pdc multiplies by 4.

[catalin gramada]

May a series fuse in line with undecoupled limiter resistor to serve as effective protection for the screens please ? Thanks.

[Helmholtz]

Could an admin move this thread to the "Theory&Design" section, please?

[Chuck H]

Could an admin move this thread to the "Theory&Design" section, please?

[SoulFetish]

I, having had to repair the damage caused by screen failures and over-dissipation, would highly recommend fusing the screen supply. Only in that it’s such a common mode of failure in Output tubes. Probably the most common.
The damage ranges from screen supply dropping resistors going open with no visual sign of stress, to 20% of the amplifier looking like it was hit by a mortar.
I personally don’t like relying on resistors to fuse open as a protective measure in this application. My opinion is to fuse the screen circuit/circuits, and let fuses do what they’re made for.