Screen capacitor... is it a current source to power stage??

[reubster]

Been wondering about this over the silly season.
Take a typical guitar amp with a CLC pi filter or a CRC pi filter.

I've tended to visualise that the 1st capacitor (rectifier smoothing) supplies all of the current required by the power tubes.
And furthermore that the screen capacitor supplies only the screens and the preamplifier circuits beyond
But I'm thinking that when power stage requires current, it will also be pulled from the screen capacitor back through the pi filter..

Is this correct?

[Enzo]

Only at any point where the voltage in the screen node cap is more positive than the reservoir cap.

The screen cap does provide additional filtration, so the PI and preamp supply is smoother, but it also supplies some decoupling as well. The further cap stages for the preamp do little ripple filtering, and are mostly for decoupling.

[reubster]

Thanks Enzo

Your reply re screen node voltage being greater than reservoir cap makes perfect sense.

Obviously the condition (V screen > V reservoir) is dependent on a number of conditions, ie output impedance, reservoir capacitance, screen capacitance, pi filter impedance & how hard you are pushing the output stage etc etc.
When cranking amplifiers, I suspect this condition (screen supplying current to output stage) is actually occuring quite regularly and plays a bigger role in the tone and feel of the amplifier than we (I) give it credit for.

It also seems to explain (at least as I see it) a couple of observations I've made over the years.
I'm interested in hearing peoples thoughts/experiences on these.

Screen capacitance.
If you consider that screen current is typically 1/5th to 1/10th of the idle current to the output stage, you would not expect to need a large value screen of capacitance compared to the reservoir capacitor.
Furthermore you would not expect that changes in screen capacitance would have much effect on sound.
Yet in my experience, the amount of screen capacitance has a bigger effect on the sound and feel of an amplifier than I would expect.
Case in point,
I built a 50watt marshallesque amplifier with switchable filter capacitances: 50 or 100uf for the reservoir and 50 or 100uf for the screens.
As expected, changing the reservoir capacitance between 50 & 100uf makes a big change to sound and feel of this amplifier but I'm very surprised (especially considering a 50uf screen capacitance is not trivial) by how significant (to the tone and feel of the amplifier) changing the screens from 50 - 100uf is.

Could this be down to how much "support" the screen capacitor is providing to the reservoir capacitor when the power stage is cooking?

CLC vs CRC pi filters
I've converted 3 amplifiers with CRC pi filters to CLC.
In all 3 cases, there is an improvement to the tone: it becomes much smoother and less harsh.
Over the weekend I replaced a 1K pi filter resistor (8 volt drop) with a 4H 100ohm choke, the reservoir and screen capacitors (both 22uf) remained unchanged.
As expected there was a small rise in the screen and preamp voltages but why does the tone get smoother?

Could it be down to the screen capacitor supplying current back through the pi filter?
In a CRC pi filter, the impedance path for the screen current is R which has a fixed impedance for all frequencies.
In a CLC pi filter the impedance path is determined by L which has a low impedance path for low frequencies and a high impedance path for high frequencies, ergo the lower frequencies are more reinforced.

Any thoughts?

[soundguruman]

Been wondering about this over the silly season.
Take a typical guitar amp with a CLC pi filter or a CRC pi filter.

I've tended to visualise that the 1st capacitor (rectifier smoothing) supplies all of the current required by the power tubes.
And furthermore that the screen capacitor supplies only the screens and the preamplifier circuits beyond
But I'm thinking that when power stage requires current, it will also be pulled from the screen capacitor back through the pi filter..

Is this correct?

yes

[Alan0354]

I have no answer and I am surprised that it make that much difference. For one, have you look at the supply of the power section and the screen supply section with a scope in 1) DC mode to see the droop and 2) AC to look at any signal riding on the high voltage.?

Look at how much each point droop when hitting the guitar hard and compare the difference between different value.

I tend to assume the higher frequency signal is well filtered out on the supply at all points. For a 4H choke, it's reactance is 2\pi f L=1.5K @60Hz, 3K@ 120Hz, and 6K @ 240Hz. So it's is equal to a bigger resistance of 1K, Maybe try 2K or 3K!!!

[Enzo]

I doubt it is from feeding current back to the reservoir. Screen voltage affects tube gain. Screen current affects screen voltage. A stiffer cap there will hold the voltage more steady, which results in the tube gain being more steady. I bet that is what you are hearing.

Try this, stick a diode after the reservoir cap. Now the screen cap CAN'T run current back to the first cap. Report back.

[Alan0354]

I believe current flow both ways, if the power amp draw down the reservoir cap, there is no reason the screen cap will not provide current if the voltage on the reservoir cap is lower.

I think it all comes back to my question about cheap caps and loss tangent and using a small parallel cap to extend the useful frequency.

[Enzo]

I don't think there is any disagreement when the reservoir cap is lower in voltage. The question remains, in practical use, does the screen cap actually have much opportunity to do so?

[Alan0354]

I just look at it as the current flow to the point of lower voltage. But I don't even see what is the issue whether the screen cap supply current back to the reservoir or not. The only think puzzle me is why change the sound. To me, as long as you don't see ripple in any of the high voltage notes, you only see droop when high current demand, there should not be difference in sound......That is if you have big enough cap to drown out all the ripple, that you only see droop on high demand.

[Enzo]

But I don't even see what is the issue whether the screen cap supply current back to the reservoir or not.

because that is the question the original poster asked. That was the reason he started the thread. You already have your thread on buying caps and why they may or may not sound different. So please don;t hijack reubster's thread.

[Alan0354]

I am not hijack the thread, he is talking about the screen cap effect and I am questioning the reasoning. I am no more hijacking than you here.

[Enzo]

Then I apparently misunderstood you. I apologize. I thought you were saying you didn;t care which way the current flowed, which was his question.

[J M Fahey]

FWIW, although the screen filtering cap *might* supply current back to the main cap, I think it will be quite irrelevant to the sound.
1) to begin with, for this to happen its voltage must be higher than main reservoir , which may or may not happen.
2) in any case, even if existing, the voltage difference will be small and it sits "on the other side" of a large resistor or an even larger impedance choke.
Since current is deltaV(small)/impedance(large) net result is that current contributed by such capacitor will be very small, in the order of a few mA .... and that *if* it flows at all, which implies its effect will be nil.
3) of course, the way to prove it is to add a diode, as Enzo suggested, and short across it with a switch or relay.
Which should be operated by somebody else, away from the musician testing the amp, to avoid subjective impressions.

[loudthud]

What is being overlooked is that a choke stores energy and it can push down on the screen cap while pulling up on the main resovour cap. The easiest way to see this is with a simulation like PSUD2. You can look at the current in the choke and tell when it reverses.

An LC filter acts like an averaging filter except for the resistive losses. If the main B+ has 40V of ripple, the screen node voltage will be half way between the positive and negative peaks minus the resistive losses. In a class AB amp, the screen node follows big changes in ripple on the main B+ tracking the average and filtereing out nearly all the ripple.

If you put a diode in series with the choke, don't be surprised if the kick from the choke kills the diode with excessive voltage. An inductor will generate any voltage (limited by parasitics) necessary to maintain current flow. Simullations of choke input filters easily generate warning errors that solid state diodes are exceeding breakdown voltage.

[Alan0354]

I don't see it will burn the diode. If the reservoir got drawn below the screen voltage, obviously the diode want to turn off. Yes, the choke being an inductor wants to keep the current flowing and the terminal connects to the diode jumps up to try to keep the current going. But then the diode will turn on again. As the choke/diode junction rise above 0.7V over the screen voltage and current starts to flow and the choke is happy!!!! This will go on until the choke is demagnetized and return back to normal. No diode will be burned.

The Choke/diode junction is not going to go negative when current cut off, it goes more positive.