Closed loop stability help needed.

[Wavebourn]

Or remove C10 and shift 2 remaining poles further from each other.

But still, I don't see bias supply for NPN transistor base and output tube.

[jerluwoo]

Q1 is PNP and bias is supplied from the cathode of 6J9 being more negative than ground.

Maybe using a zener instead of R5 may help things as well.

[aeandm]

Hello, I recommend disconnected chain create loop,
and give signal 1000Hz input (P=.indifferently ). Chain creator instability is error diagram .
Bias 6P41C is -regulation, temporarily .

aeandm

[jan.didden]

Good ideas.
The oscillations are caused by the control loop - it's loop gain is set by the input cap, and the input source can be seen as a (low) impedance in series with the cap. The loop gain is directly proportional to that series impedance. Make C1 larger, that decreases the series impedance and that decreases the loop gain - no more oscillations. Classical.

The current position of C10 causes a lot of phase shift in the loop, exarbating the problem.

The above advice is a good start. Note that the control loop basically only needs to be a DC or very low speed loop.
You can also bang it with lots of cap at the C10 position, like 100uF or more. That'll teach it

jan

[Wavebourn]

Originally Posted by jerluwoo

Q1 is PNP and bias is supplied from the cathode of 6J9 being more negative than ground.

Sorry, my typo. Of course, I meant PNP. But on schematic I did not see any resistor from base to cathode, only to the anode of the preamp tube.

Anyway, 3 poles is a bit too much. I would use 2 of them with shifted from each other time constants.

[TheGimp]

The goal of the circuit is to allow me to direct couple the Mu stage output to the 6P41S.


My chain of thought was that since it is direct coupled, I can use the Mu stage to set the bias.

With the triode biased by an LED and it's anode feed from a current source it behaves like a voltage source, and it thus should be possible to control the voltage at the cathode to shift the bias to the 6P41S.

So, use the bias voltage at the 6P41S to close the loop and make it a servo.

It looks like my best approach might be to remove C10 and increase C4 to say 2200uF?

If I remove C4 and C10, what would be a good starting value for connecting from Q1b-c?

[aeandm]

Dear diyAudio

Sorry , I slowly translate english. For me is simpler
draw diagram, now I work out a problem. I think R8
to e. Q2, and b.Q1 through additional resistor e.Q2.
R8=60mA 6P41C,additional resistor =? (-U=?)
Remove VD1 and C9 is not line element.

aeandm

[Wavebourn]

I would simplify your servo a lot, you don't need so huge OL gain; on low frequencies servo adds own distortions and instability.

Try one NPN transistor, emitter on ground, collector to screen grid of dynamic load, resistive divider from cathode of an output tube shunted by a cap to it's base.

[TheGimp]

Something like this Anatoliy?

Certainly more elegant with fewer poles/zeros.

[Wavebourn]

Yes, something like this. But base current needs to be limited...