"Floating" DC HV regulator

Of course we all are entitled to our opinion! But if those regs all fail (which you mentioned), that points to incompetent design, no?
Or did you break 'm all ;-)

T-reg article etc here: https://diyaudiostore.com/collections/power-supply-kits/products/linear-audio-t-reg

https://cdn.shopify.com/s/files/1/1006/5046/files/T-reg_5_article_diyaudio.pdf?v=1612255814

Jan
The owners of said amps broke them often as the result of tube rolling inappropriate tubes into the output stages, I warned them but they did not listen. ARC recommended replacements be purchased directly from them and at a pretty exorbitant mark up. The amps were designed around Sylvania made 6550A which were particularly rugged and had a very hard vacuum. (Amps using them never failed IMLE. The problems started when other brands were substituted - just ran them too hard for them to survive. (This was the early 1990s) Also had one in a DC servo circuit that had failed. I serviced them, it got so bad I finally decided I was not going to continue. The M100 had 2.3mm thick PCB with at least 2oz copper and the parts which were not socketed were really hard to get out of the PCB without damage to the traces. (I did no damage, but it was a slow and painstaking process to remove with the tools I had available.)

Those amps were far outside the range of gear I could afford at the time, and after listening to and working on multiple pairs it was not a sound signature I found very appealing. Upkeep costs would have been high.

Thanks for the links, I will check out the T-Reg. 🙂
 
FWIW, the authors of "Art of Electronics" have put Chapter 9 online. https://artofelectronics.net/wp-content/uploads/2016/02/AoE3_chapter9.pdf

The HV regulator I describe is on page 126 (696 in the book).
Nothing but respect for H&H, but that reg illustrates what's wrong with that topology.
The feedback divider is 4M/26k. That's over 40dB of loop gain down the drain.
That's 154 times worse regulation and 154 times larger Zout than an equivalent floating regulator.

jan
 
  • Like
Reactions: Bonsai
Thanks for the suggestion Elvee. This was a starting point for a shunt reg circuit.

Jackinnj, I have had a look...at what I have read using CCS devices optimizes the floating circuit. At this point perhaps I should say that before diving deeper into these things I should at least know more about the theory behind these circuits, in this sense the linked book seems to me a good reference without too much maths.
 
A few days ago I came across a circuit on tubelab, interesting in my opinion, dating back to about twenty years ago. It relies on an opamp as a voltage reference using a certain feedback (error opamp). The resulting circuit is claimed to have low noise and output Z, and stability regarding output voltage drift.
The opamp in question then is fed directly from the AC voltage of the filaments as a voltage doubler (DC floating voltage). There's to say that, according to the author, it must have certain requirements: OP27 is mentioned (unity gain stable at operating conditions) but other cheaper ones I think can be used (NE5534?).
I tried a simulation and gave excellent results, but you know there is half a sea between saying and doing... so I was wondering if it could respond to real needs for powering a pre/phono.
A few days ago I came across a circuit on tubelab, interesting in my opinion, dating back to about twenty years ago. It relies on an opamp as a voltage reference using a certain feedback (error opamp). The resulting circuit is claimed to have low noise and output Z, and stability regarding output voltage drift.
The opamp in question then is fed directly from the AC voltage of the filaments as a voltage doubler (DC floating voltage). There's to say that, according to the author, it must have certain requirements: OP27 is mentioned (unity gain stable at operating conditions) but other cheaper ones I think can be used (NE5534?).
I tried a simulation and gave excellent results, but you know there is half a sea between saying and doing... so I was wondering if it could respond to real needs for powering a pre/
A few days ago I came across a circuit on tubelab, interesting in my opinion, dating back to about twenty years ago. It relies on an opamp as a voltage reference using a certain feedback (error opamp). The resulting circuit is claimed to have low noise and output Z, and stability regarding output voltage drift.
The opamp in question then is fed directly from the AC voltage of the filaments as a voltage doubler (DC floating voltage). There's to say that, according to the author, it must have certain requirements: OP27 is mentioned (unity gain stable at operating conditions) but other cheaper ones I think can be used (NE5534?).
I tried a simulation and gave excellent results, but you know there is half a sea between saying and doing... so I was wondering if it could respond to real needs for powering a pre/phono.
If Your going to do a floating HV regulator do a variation on Roger Modjeski's RM-4 power supply using the UA723, much more predictable results, than using a dual supply opamp, and lower noise floor, 2uV ripple if done correctly. Ive used a similar setup but with a 250V out.
 

Attachments

  • 3A07EF7A-706F-42F0-95CF-E8C67BA3EA1E.jpeg
    3A07EF7A-706F-42F0-95CF-E8C67BA3EA1E.jpeg
    163.9 KB · Views: 107
Nothing but respect for H&H, but that reg illustrates what's wrong with that topology.
The feedback divider is 4M/26k. That's over 40dB of loop gain down the drain.
That's 154 times worse regulation and 154 times larger Zout than an equivalent floating regulator.

jan
Simulated, the Z out is very low, but the PSRR is abysmal. Will test in real life to see if empiricism confirms simulation.
 
There’s always the maida regulator as an example however finding a good SOA mosfet is a pain. So many modern designs are now targeted towards switching that they have a small short time period SOA but longer term DC results in popping candy (exploding mosfets).
 
Last edited:
  • Like
Reactions: Bonsai
Thanks for the proposed circuit Danny, as far as I can see it seems to use a regulator from the TL783 family. In any case I think to abandon the idea of a "floating" in favor of a more traditional one, like the one among the examples proposed by Elvee.

I've actually used the Maida reg before. and obviously it's an excellent circuit, only that I needed to try to change it to notice differences compared to what I usually use. That's it, with pros and cons. The only thing I didn't find particularly good at are the Zener regulated Mos. but for the rest I'm fine with more or less everything, obviously if I can understand how it works on something not too complex it's better.