Power supply question

[catalin gramada]

Hello. I wonder why not to use a blocking diode between reservoir, meant plate supply and screen grids supply cap. Playing with power supply dynamics versus bass transients should be easier I think if I isolate the screen caps to not discharge in reverse to the plate. Then a series diode should be the easy think to do. But is not very clear the whole picture to me, should.d I have any concern doing that please ? Thanks.

[catalin gramada]

This is what I mean...

[Helmholtz]

Can you really see the plate supply voltage drop below the screen supply voltage at times?

[mozz]

Give it a try and see if you hear anything different. If yes then measure before and after, let us know what you find. I read once that on the old bassman power supply, interaction and impedance were part of the signature sound.

[catalin gramada]

Can you really see the plate supply voltage drop below the screen supply voltage at times?

Oh yes I can see as the plate voltage goes lower the rising screen currents may damage the screens so it is a drawback as the the screen voltage is not in a safe region or screen limiter resistors are so huge as it compromise the whole circuit ideea...

[Helmholtz]

Oh yes I can see as the plate voltage goes lower the rising screen currents may damage the screens ..

Not clear what you mean.

I think it's good practice to have the screen supply follow and track the plate supply.
So the screen node cap shouldn't be too large.
Also for a high power (bass?) amp the reservoir cap looks rather small.

[g1]

the screen node cap shouldn't be too large.
Also for a high power (bass?) amp the reservoir cap looks rather small.

Yes, why is the plate node lower capacitance than the screen node? I would expect to see something more like 500 > 100, or 50 > 10 there, rather than 50 > 100.
You can still run the preamp supply off the plate node and have the screen node separate.

[catalin gramada]

...then what problems could be, running the power tubes like in sketch circuit please ? There is any risk to destroy the screen grids please ? DC choke value is 100 ohm around. Plate voltage ca 450V dc. My intention is to keep the screen voltage more "stable" not necessary regulated but out of severe variation of plate voltage under power transients. The current of screens is low, making supply cap bigger think helping to make little stiff. But with condition to not depend by discharge over the plates....this was the ideea...to block it in reverse. I don't know if it work as I don't see the whole pictures. The caps could be 100 uF for the plates and 500 to the screens I just asked if it works or not and why please ? Thanks.

[Helmholtz]

I don't understand your motives and you didn't respond to my points.
Edit: missed post above.

Generally a stiffer screen supply tends to increase average screen dissipation but there are also other factors to consider.

[loudthud]

The attached pic shows B+ (red) and screen (green) nodes of a 50w guitar amp power supply from idle to full load. Screen Voltage hovers near the bottom of the saw-tooth mostly due to losses in the DC resistance of the choke. With lower resistance it would be closer to the center of the saw-tooth. You can see the currents used for idle and full load next to the current loads I1 and I2.

The choke being in series with the diode in Catalin's example will act strangely. At times the left side of the choke will dip negatively (relative to the right side) keeping the diode turned on.

[Helmholtz]

The attached pic shows....

Thanks a lot for posting this.
Made me aware of the stepped load feature I hadn't noticed before.

I did a simulation using C1= 50µF, C2= 100µF and L1= 5H/100R.
The current step makes the screen voltage drop evenly without any "undershoot", even though the plate node (mean) voltage gets lower than the screen supply voltage and there is some reverse current through the choke.
I would post the results here if I knew how.

I think inserting a diode would somewhat delay the screen supply voltage drop after the load step.
I also expect increased screen supply ripple with the diode.

[Helmholtz]

Ok, trying to post my PSUD2 results: psud2.pdf

Does this work? (Try the link in the frame below).

The green line is the inductor current (scale on the right side).
Lowering C2 to 50µ causes a temporary undershoot of the screen supply after the load step (idle to full power).

[catalin gramada]

We have same thing in Ampeg b15s with F10 D5 diode but in ground return. They did not used a choke but 1k resistor instead. I think mainly it have the same purpose as I intended to. Otherwise what is the reason of this diode in the circuit please ?

[Greg Robinson]

D5 makes sure that the neon indicator turns off as soon as the standby switch is turned off, rather than gradually dimming.

[Helmholtz]

Maybe, but might also be related to the standby switching.

How does it sound with the diode?