Mu Follower with a fixed bias upper triode

[Chuck H]

My point was about power supply ripple rejection, defined as plate supply ripple divided by corresponding ripple at circuit output.
It seems obvious that the bias divider adds 2/3 of the supply voltage ripple to the output.
The idea was to add a 120Hz ripple source (say 10mV) in series with B+ in simulation.
Dunno if that's possible with LTSpice or if all sources need to be grounded.

The PSRR result may not be a problem with a well filtered supply.

LTSpice can be finicky about that sort of thing, but I usually manage to work it out and see expected results. I'll ignore any series filtering and attempt to inject 20mV (just for an obvious reference) of 120Hz on the mu follower supply, compare the fixed and cathode bias circuits and report (at some time though maybe not immediately). Thank you.

[Chuck H]

Well I've tried injecting 120Hz. I had to go much higher than the prescribed 10mV and increase ESR on that filter node quite a lot to see any effect at all though so I question the accuracy of LTSpice for this evaluation. At any rate, I didn't see any difference in ripple effect on the signal between cathode and fixed bias. The only difference I noticed so far between the two is a small reduction in gain using the fixed bias circuit. Probably due to signal loading on the grid from the fixed bias voltage divider. I don't have anything like a 6111 tube in my CAD so I used circuit values more akin to a 12ax7 tube and the fixed bias divider is 820k/1000k for the same bias level as cathode bias. Essentially adding a 450k load as opposed to whatever signal loss there might be with the cathode bias arrangement and NFB through the 1000k bias resistor. So maybe it's not loading that's reducing gain. Though I don't know what else it might be.

[Helmholtz]

My proposal meant adding a 10mV/120Hz voltage source between (clean) supply voltage and circuit nodes A or B, so that the ripple source is in series with the supply.
This way there should be no influence of filter capacitance or ESR. You could even use an ideal 200V DCV source as supply.

[Chuck H]

My proposal meant adding a 10mV/120Hz voltage source between (clean) supply voltage and circuit nodes A or B, so that the ripple source is in series with the supply.
This way there should be no influence of filter capacitance or ESR. You could even use an ideal 200V DCV source as supply.

Ok. Let's try it that way then. I did inject the AC onto the supply voltage right at that filter node. And in any actual amp there WILL be a filter. This does mean any prior stages on that node are also affected though. I think I see where you're going. You want to see about any difference in ripple rejection from the circuits themselves. I'll run an isolated supply for the Mu follower with the AC injected report back.

[Chuck H]

Ok. Here's with the isolated supply, 10mVAC ripple and zero input signal. This is with the partial cathode bypass for the lower triode as shown in the first post (adjusted for frequency to the circuit I'm using). Cathode biased in purple and fixed bias in green.



And here is is with a full bypassed cathode on the lower triode. It looks like the cathode biased circuit has about ten times the ripple rejection in both cases. Is this what you were looking for?

[Helmholtz]

Is this what you were looking for?

Yeah, that's about what I expected. So the fixed biasing spoils the excellent PSRR of the cathode biased version.

Can I see your schematic/wiring of the ripple source?

[SoulFetish]

Huh??? Obviously, we would expect to see the cathode biased upper triode with lower supply ripple at the output.
But your not going to get more power supply noise at the output than is on the supply. How are you getting roughly 28mV RMS at the output of the fixed bias version, with 10mV ripple on the supply?
The voltage at the output would be A gain of less than one of the voltage at the grid of the upper triode. Which would be the ripple voltage attenuated by the value of the biasing resistor divider
did I miss something?

[Chuck H]

Huh??? Obviously, we would expect to see the cathode biased upper triode with lower supply ripple at the output.
But your not going to get more power supply noise at the output than is on the supply. How are you getting roughly 28mV RMS at the output of the fixed bias version, with 10mV ripple on the supply?
The voltage at the output would be A gain of less than one of the voltage at the grid of the upper triode. Which would be the ripple voltage attenuated by the value of the biasing resistor divider
did I miss something?

The isolated supply with the AC component was fed to the "top" of the circuit. So it's essentially on both triodes. As it would be in the event of 10mV ripple at a supply node. Since the amplification comes from the lower triode (graphically in the schematic) there IS amplification of the ripple. Without the NFB provided by the cathode bias circuit there is less cancellation so the amplification of ripple at the lower triode is translated to the output of the cathode follower.

Still unclear to me is why the fixed bias circuit has lower gain than the cathode bias circuit. The cathode biased arrangement has roughly 10% more gain AS WELL AS better ripple rejection. No reason to fix what's not broken IMHO. Some clever guy cobbled that circuit together based on reality of all circumstances. Some of which we may not be considering now. I'm not an engineer but I do know the nerds that came up with these circuits were. Guitar amps weren't on their agenda.

[SoulFetish]

I want to see the LT spice circuit.
if It was amplified by the bottom triode it would be out of phase with the top triode and cancel, resulting even less ripple at the output

[SoulFetish]

By the way, not advocating for fixed bias. Just sayin is all

[loudthud]

The fixed bias circuit is not too different from a solid state circuit discussed here: https://www.ssguitar.com/index.php?topic=3373.0 (you probably need to join the forum to see the attachments) The advantage is, it works without a tweek if the JFETs are matched.

Will the 6111 work with only 50V across it ?

[Helmholtz]

How are you getting roughly 28mV RMS at the output of the fixed bias version, with 10mV ripple on the supply?

did I miss something?

I see only about 2.8mVrms ripple with fixed bias.

[Chuck H]

I see only about 2.8mVrms ripple with fixed bias.

Correct. That's with a fully bypassed cathode on the lower triode. With partial bypass it's worse.

[Helmholtz]

That's with a fully bypassed cathode on the lower triode. With partial bypass it's worse.

Upper picture ~3mV, lower pic ~2.1mV.

[Helmholtz]

....and NFB through the 1000k bias resistor.

What NFB?

How much does gain change and what frequency did you use to test the gain?