Perfect tube regulator

[CarlyBoy]

Hey Folks,

Flipping through old issues of Glass Audio, I came across Rickard Berglund's perfect regulator (4/96 p. 65):



Any possibility of substituting a low voltage regulator tube like the 0A3 (75V) for the zener? What component values would be altered? I'm afraid I have less knowledge than enthusiasm. Any suggestions would be appreciated.

[Zen Mod]

Originally posted by CarlyBoy
Hey Folks,

Flipping through old issues of Glass Audio, I came across Rickard Berglund's perfect regulator (4/96 p. 65):



Any possibility of substituting a low voltage regulator tube like the 0A3 (75V) for the zener? What component values would be altered? I'm afraid I have less knowledge than enthusiasm. Any suggestions would be appreciated.

leave it as is;
with gass toob instead of zener -you'll burn at least 10-15mA of adittionall current.
anyway-gass toob can be used as reference in better stab or even in stab constructed for known circuit,not universal one-like this .

this perfect reg can be much better ,with adition of output cap,small cap from output to wiper of lower pot......and even another zenner-included in output voltage divider upper "half".

in this way (with another zenner) ECC LTP is from both sides leaned on solid (in this case bridge) reference.........


bla bla -it's pretty late here .............zzzzzzzzzzzzzzzz

[Tubes4e4]

Hi Choky,

this perfect reg can be much better ,with adition of output cap,small cap from output to wiper of lower pot......

Yes, 100-220nF or something, so that the rest of the ripple is coulped for counter-action to the diff amp. Also, a very high ohm resistor from EL86 plate to diff amp input would implement forward regulation (further improving regulation of changing mains voltage).

Accordingly, here is a much better implemented (and performing) EL86 based regulated PSU on my homepage.

Tom

[kevinkr]

I'd watch out about modifying this design, if you read the article very carefully you will find that there are some interesting twists to this design in terms of the way feedback is applied. (Like the design has positive feedback implicit in its operation.) Don't bypass the pots with caps! Set up as described in article is not intuitive, follow to the letter to get it to work properly.

You could I think substitute a 5651 for the zener which would result in better drift and voltage stability than the typical zener offers. Do this only after you get the actual circuit to work.

I use pentode connected EL84/6CA7/6L6/6550 with 12AX7 cascode error amplifier (tube is not the optimum choice for a cascode, but the ~ 46dB loop gain is enough for my purposes, cheap, and stands off lots of voltage ok) and 5651 reference. I can furnish details at some point if there is interest. Note in lower voltage versions you can probably use the EF86 in place of the cascode to good effect, but I have not tried this yet. Ripple rejection >60dB typ, low drift, good line and load regulation <1% typ and low output z. It's simple too, reliable - variants of this circuit are in almost everything I ever built for sale.

[leadbelly]

Originally posted by kevinkr
I can furnish details at some point if there is interest.

Fer shure!

[CarlyBoy]

I'd certainly be interested in details, Kevinkr. No detail wasted on me... I'm still trying to figure out how designers decide on circuit values in tube regulators.

[kevinkr]

I'll try to get something up in the next week or so. This was quite a busy week-end with a meet at my place of our local tube diy circle. (NNETG)

[av-trouvaille]

Hello Kevin,

Would it be possible also to expand on your experience on how to reduce noise with this type of regulator?

Best, Arjen.

[kevinkr]

Here is my standard 300V regulator, note that it is zener based. Getting good performance with the 5651 gas reference is a little trickier. This design works right out of the box. Note I provide it for none commercial use only.

The voltage can be easily reset for different output values from about 200V to 350V with a few component value changes which I will happily provide to anyone who asks.

To make life easier for the 6BQ5 I recommend a tube rectifier, that said it works fine with ss rectification to output voltages of 300V, higher than that a tube rectifier is de riguer to prevent pass tube arcing. Currents higher than 40mA should use the 6L6 otherwise its the same.

Keeping the noise down on the output depends on how clean the screen supply is, and how much feedback margin is available. Typically I aim for a margin of >30dB if possible.

Open loop gain is in the range of 43dB +/-3dB and I recommend not scaling voltages by more than a factor of 3 to 4 in order to maintain sufficient loop gain for good regulation.

The cascode error amplifier works best with output voltages of 200V and above, the only limitation is to avoid exceeding the voltage ratings of the 12AX7A - in practice up to 500V output voltage is practical with a 6550 pass tube and a few other minor changes.

Note that you must provide separate filament supplies for the 6BQ5 and 12AX7A. Regulated dc on the 12AX7A filament is not necessary but does not hurt either. A 6CA4 may be operated on the same winding as the 6BQ5 if necessary.

Warm up drift is mostly dependent on the zener chosen, and I recommend 5W types for lower drift. Once stable the voltage should drift less than 1% in a 24 hr period.

Output ripple of <1mVrms is possible with careful layout and a 12AX7LPS in the error amplifier. Line and load regulation is better than 1% no load to full load as long as the raw supply is at least 100V greater than the output voltage at lowest line voltage encountered.

[SY]

Thanks for the circuit! Do you bias up the separate heater winding for the pass tube?