G2 stabilized supply from wavebourns pyramid - modifications?

[FLT]

Hello,

i try to find a good solution for my GU-50 project G2 Supply.
First i wanted to use a simple zener/mosfet circuit. Then i
decided to try Wavebourns B+ Power Supply Circuit.

Had to replace some devices:

Russian SG-4 for 0A2
2SK2760 for IRF730
R1 10k for 15k
R5/6 100k for 200k

Just look at the left picture...

With no load i have:
340V Input from Laboratory Supply
149V at the SG-4S
303.6V Output

22k Load Resistor:
333V Input
148.9V at the SG-4S
303.3V Output

2x22k || Load(=11k Load):
328V Input
148.9V at the SG-4S
303.1V Output

So far so good...
0.5V Drop at 27mA Load.
My Problem is the big Stabilizator Tube. I want to replace it with a zener diode.
First try: direct replace of the SG-4S, no mods to the circuit.

No Load:
340V Input
155V Zener
319V Output

22k Load:
334V Input
152V Zener
316V Output

11k Load:
326V Input
152V Zener
313V Output

Not so good...6V Drop from Idle to full-load
And the Zener is realy hot...So i changed some Resistors to decrease the current
and the Heat on the Zener.

Look at the right picture.
Icreased R1 to 22kohm and R7 to 100kohm.

No Load:
340V Input
148V Zener
307V Output

22k Load:
334V Input
148.6V Zener
306.1V Output

11k Load:
329V Input
148V Zener
305V Output

2V Drop between no-load an full-load...and the Zener still hot...
IIRC the main Zener current is fed by R7 from the stabilized Output, not
from R1/Q2. I played with discrete stabilization circuits years ago, but not
above 100V.
Could somebody please give me a hint in the right direction?

Thanks

Frank

 

[PRR]

BZX85C150 = 1.3W, 2mA test current. 8mA is about the maximum safe, and that will be very hot.

Design for 2mA.

Assume 1mA from R1 and 1mA from R7.

R1 is nominally 340V-329V dropped to 310V. But the raw supply may vary? Assume R1 may be asked to drop 50V. Then R1 is 47K.

R7 is nominally dropping 150V. For 1ma, R7 can be 150K.

R1 15K change to 47K
R7 47K change to 150K

A teeny Zener is not a match for a fat gas bottle. The bottle supports 30mA, though here run at 8mA. The Zener you picked will run 8mA but VERY hot and dubious reliability. The R1=15K and R7=47K resistors should be around 3 times bigger.

I am not too worried about the higher impedances. Lower current is less gain, but higher R1 is more gain, so loop-gain will not change much.

I am NOT worried about a 2V drop on a 300V screen supply. Tubes are not that fussy.

 

[FLT]

Thank you for your reply.

I changed R1/R7 to 47k/150k. No-load zener voltage is now 146V, output is 298V.
Zener temperature dropped from 62°C to 45°C...
11k load: Zener 145.3V, Output 296.3V. 1.7V Drop/~27mA=62.9 Ohm output resistance.
I know that the tube had 8mA current, thats why i already increased R1/R7.

The absolute voltage is not the problem for tube circuits, but the drop under load.
A varying g2 supply has influence on the plate current. 0.5V Drop/27mA=18.5 Ohm
output resistance --> lower is better.

regards

Frank

 

[PRR]

> varying g2 ... 0.5V Drop/27mA=18.5 Ohm

And 296V at 27mA is 10,900 Ohms, the approximate G2 dynamic impedance.

19r feeding 10,900r is about one part in 500, a VERY small change.

Further, for speech/music audio, the 10uFd or 470uFd will only allow a "slow" change of voltage on G2 current peaks. If you test no-signal and FULL signal, you will see the G2 drop a volt or so but over a period of 0.1 Seconds (longer than musical peak) or 5 seconds (with 470uFd).

You can make the output impedance dead zero or even negative if you wish. But I think you need a more complicated plan. And I don't think 50r, 20r, or 0.0r has any effect on the audio.

 

[FLT]

Yes, everything you wrote is right.
But it must be possible to replace the voltage reference without
making the circuit worse.

I did a quick hack:

Resistors, 2x150V Zener, Mosfet and some capacitors...
Simple Zener/Mosfet stabilisation circuit and tested it(no-load/11k-load).

Voltage drop was 3V...
It makes no sense to build a regulated supply when it is only slightly
better than a simple Zener/Mosfet circuit.

And thank you for your replies to my problem!

regards

Frank

 

[PRR]

I do NOT think 1% regulation is any issue for this application.

As an Abstract Problem: why does it sag, and why does it sag worse with a Zener?

A flip answer is to say the gas-tube is better than the Zener. It has much higher power capacity, and the bent-metal construction may allow "mechanical" tricks not easily done inside a Zener. Other issues are isolating the reference from source and load variations, and the loop-gain to accurately compare reference to output and drive the pass-tube (MOSFET).

The most obvious "flaw" is that R1 current flows in gas/Zener but varies a large amount. The reference voltage is liable to vary. I don't have a good grasp of gas tube flatness. I do know that Zeners have internal resistance, and for this diode apparently 1K to 6K. You show a 10V change in input voltage no-load to full-load, so that is (assume R1 is round=number 20K) about 0.5mA change. So in 1K-6K we expect 0.5V-3V change in Zener's "150V". This is doubled to the output, 1V-6V. So all your observed 3V change is explained by R1 flowing to Zener.

I see two un-tested ways to stabilize Zener current. A second small MOSFET as buffer, or a BJT "amplified Zener".