LM317 for opamps and HT tube regulation

[reinw33]

Hi everyone,

whilst the LM317 lwo voltage version is about to be used in one of my builds of an SSL preamp as the regulator for the positive +Vcc of the split supply of the THAT opamps, so low +18V.
Yesterday I t found out after researching a bit for the term "Maida", it turns out, it's a module probably named after someone with the same name.
That is somewhat mysterious.
You can actually use the LM317 high voltage version as a regulator for B+ or even the HT - Wow!
Don't know about the lower differential voltage version yet.
This Maida is supposed to have Vout range of +155V to +495V that are stepped in 10 steps IIRC and adjustable with a pot.


Then there was some talk about a Rev. 3 of Maida, that supposedly has superior noise or ripple rejection.
And is still mysterious?
Has anybody here used it or a rendition of it with success and wants to share a schematic?

I have never built a tube preamp before, but am in the process of building one.
The parts are arriving slowly.
Coming month more parts should arrive, like copper shielding foil, various caps., etc, transformers, etc.
I can wind my own transfromers but I have to wait on missing parts, and it sucks having to build bobbins as I don't have a 3D printer yet.
I'm trying to look for people in the city that offer 3D printing for material costs only.
Thanks for replies. You can shoot me a PM also if you like. Thanks.

 

[Jarno]

You can lift the regulator by stacking zeners on the regulator leg. There are a number of Elektor tube projects which use this trick (you can do this for the heater supply as well, to use a lm7812 and create a 12.6V supply).

 

[merlin]

There are lots of Maidas out there, it's quite popular.

 

[reinw33]

You can lift the regulator by stacking zeners on the regulator leg. There are a number of Elektor tube projects which use this trick (you can do this for the heater supply as well, to use a lm7812 and create a 12.6V supply).

Hi Jarno,
I do understand the theory of an elevated heater supply, so far I haven't made one yet.
But the sophistication of the LMxxxx type of ICs is still new to me.

If I understood you correctly, you mean connecting your LT e.g. 12,6V which is CTed and connect it to the LM7812, and stack the zeners on the ground pin o of the LM7812 between the ground pin and ground, so that ground is elevated by some Volts e.g. 12.6V - 0.6V-0.6V.......-0.6V= ~6,3V, perhaps use some bypassing also and the regulator LM7812 is then supposed to regulate to 6.3V on it's output.
Did I understand you correctly here Jarno?

In addition to that the LM7812 has to be capable of having the floated heater elevation Voltage connected to it's input by the CT of LT, floating on it's output.
Thanks.

 

[reinw33]

I forgot to ask, what kind of Wattage do those zeners have?

Are they like 5 Watts or sth. along this ballpark range? If they have to dissipate a certain amount of heater current.

 

[reinw33]

There are lots of Maidas out there, it's quite popular.

Yeah, kind of missed the trend.
Or am still in the process of hopping on the train.

 

[Cqwet Dbdfte]

The original 1980's LM317 data sheet has a schematic for floating it to any voltage with a HV NPN input, LT had it also with a schematic using a 3kV vacuum triode. A HV MOSFET works even better. I have submitted a diagram in one of my posts.

 

[merlin]

Here's my variant using the LM317L
https://www.diyaudio.com/community/...erlin-design-free-just-cover-shipping.406657/

See also Morgan Jones:
https://books.google.co.uk/books?id=QdGJlyCb7UsC&lpg=PR4&pg=PA389#v=onepage&q&f=false

 

[reinw33]

The original 1980's LM317 data sheet has a schematic for floating it to any voltage with a HV NPN input, LT had it also with a schematic using a 3kV vacuum triode. A HV MOSFET works even better. I have submitted a diagram in one of my posts.

Yes, just found the datasheet and looked into it briefly.
It uses two bipolars as input regulators. But the output is 160V@ 25mA, but I need 300V regulated @ 30mA.
So those are probably two weak in standalone.

 

[reinw33]

@merlin,
Thanks for sharing it so freely!

The question is in relation to regulated plate supplies only.
2N3906 seems to stabilize the LM317 voltage.
With an output requirement of 9 Watt 300V@30ma I would add a heatsink to LM317.
Would it be of any benefit to still buffer the LM317 with a mosfet that has an appropriate Gate-Source input voltage?

So that LM317 is just there for the voltage reference and the mosfet handles the load 30mA?
Would that make sense for a plate supply and protect the LM317 from any random weirdness? So that it doesn't go 'pufff'.
Does the 2N3906 and diodes make this circuit so robust that no random weirdness really matters to LM317?

 

[Cqwet Dbdfte]

the zener diode limits the dropout over the 317, which contains a "floating" reference. An input MOSFET, say 700V+ is regulated by the 317 between gate and source.
A decent cap between the gate and ground fed by a high value resistor adds capacitance multiplication and ramped startup if needed.
Linear Tech's 317 types have better specs, but vanilla types will do.
Current is limited by the 317, and any series resistance beteen FET source and 317 input, scaled against the zener value. With a good heatsink on the FET the reg is very robust.

 

[reinw33]

@Cqwet Dbdfte
Credits to @merlin for having shared his variation. Thank you!
Ok, it seems to be robust and proven to run stable, but up to what voltage output on the HT psu?

This schematic seems to minimize power dissipation in the LM317, that's what I was looking for.
I don't see a variation as a current booster for mic preamps ever necessary in the LT circuit of the psu. Am I wrong here?
How is it's PSRR on both psu vs 21st century versions using the LT3080?
If this variation is going to sound similar in quality to other topologies I plan to use, than it might stay longterm.
I ordered 10x LM317HVT. Don't know the vendor. I just took the stuff that was available domestically and didn't look fake.
But have no breadboard setup yet. So it is kind of slow going frankly.
I will do all of my tests with LT first ofcourse on a 24VAC transformer with a full bridge I think. Why? Because I'm sane.
And without elevation v., Seriously, I see no point in having it in there for getting to know the lm317.
And then use it as a breadboard supply or sth.
But if 10V headroom is better than the minimum of 3V, then I can't recall how low I could regulate it stabley.
See if I can get stable 9-12,6V without oscillation, considering the minimum headroom 3V iirc.
But am still so unfamiliar with lm317 so that LTspice will have to be first stop. I guess.
When you say

Current is limited by the 317, and any series resistance beteen FET source and 317 input, scaled against the zener value. With a good heatsink on the FET the reg is very robust.

I understand you still mean the MOSFET on LT psu.
Up to what high voltage Vout has this paticular regulator.pdf HT psu been proven to run stable?
Thanks.
EDIT: 24VAC, roughly ~32VDC. This would be a good exercise to make a regulated heater 12,6V or even 6,3V. Any objections?
I think about how a 21st century version compares in PSRR performance improvements.

 

[Matador]

Three terminal regulators in this floating configuration tend to go "poof" very easily when subjected to output shorts, be them intentional (e.g. inrush current charging up bulk capacitance) or unintentional (a shorted tube or slipped voltage probe). In general, they are extremely intolerant of exceeding the maximum input-to-output differential and destroy themselves very quickly. Remember all of those 47uF capacitors on the output look like dead shorts on startup, meaning that the input to the regulator can rise much faster than the output, which kills the regulators.

The MOSFET is there to soak up anything in excess of 12V above the output regulated voltage: like if your incoming supply is 350V, and the output is 300V, then you need to "waste" 50V. The LM317 will only "see" a 12V input-to-output differential, and the MOSFET will "see" the other 38V so that differential is always in spec (the MOSFET can tolerate 700V across it's terminals, however the LM317 can only do maybe 20V).

 

[Brian Roth]

EDIT: 24VAC, roughly ~32VDC. This would be a good exercise to make a regulated heater 12,6V or even 6,3V. Any objections?

Remember that the 317 will have to dissipate the voltage dropped as heat.

Example, dropping the raw 32 VDC down to 12 volts with 1 Amp current draw means 20 Watts of dissipation in the 317.

Raw voltage should be only just a smidge higher than the desired regulated voltage.

Bri

 

[Cqwet Dbdfte]

the 317 have survived hard shorts on output with the necessary diodes and current limiting resistor in place.
Max voltage is whatever the the MOSFET can take, 700V types are low cost.
I have made a few PCBs for thru hole parts.
The PSRR is very high, depending on frequency and first RC filter which helps a lot.
I don't recall exact numbers but -80dBc should be feasible.
I have used this setup since the early 1990's.
The FET needs a heatsink, a convenient chassis with requisite isolation is good
Fullpak TO-220 makes that easy. The 317 will not get hot.
I use w.w. resistors in series with rectifier diodes to emulate vacuum diode resistance, wich lowers peak current in the input filter. 10 - 20 uF should be sufficient.

 

[Cqwet Dbdfte]

Dropping out 20V in a heater regulator seems harsh, a chassis mounted resistor could absorb most of the heat. Tube heaters can also be put in series if the same type.

 

[merlin]

@Cqwet Dbdfte
Credits to @merlin for having shared his variation. Thank you!
Ok, it seems to be robust and proven to run stable, but up to what voltage output on the HT psu?

With a regular pot it's limited to about 350V, but with a fixed voltage there is basically no upper limit as long as the MOSFET can take it. The MOSFET is the part that shoulders the differential voltage and power in a Maida.

Personally I don't find much use for the Maida, when a Zener follower is so much simpler. Zeners are not as accurate of course, and you have to wait for them to settle at running temperature, but valve gear seldom needs that level of voltage accuracy.

 

[reinw33]

Dropping out 20V in a heater regulator seems harsh, a chassis mounted resistor could absorb most of the heat. Tube heaters can also be put in series if the same type.

In real audio gear dropping 20V for the haters, would be bad design and bad foresight.

Basically, I make a LM317 low voltage version test.
Why. Because the HV version is rare and I don't won't to zap one after the other. Only got 10 of'em.
I wouldn't push the lm317 beyond 10Watts without heatsink. Still too harsh? I have no LT experience with lm317LV yet, but, I think this transformer could shatter it, in terms of pushing it to far for short 1A transients.

 

[reinw33]

Three terminal regulators in this floating configuration tend to go "poof" very easily when subjected to output shorts, be them intentional (e.g. inrush current charging up bulk capacitance) or unintentional (a shorted tube or slipped voltage probe). In general, they are extremely intolerant of exceeding the maximum input-to-output differential and destroy themselves very quickly. Remember all of those 47uF capacitors on the output look like dead shorts on startup, meaning that the input to the regulator can rise much faster than the output, which kills the regulators.

The MOSFET is there to soak up anything in excess of 12V above the output regulated voltage: like if your incoming supply is 350V, and the output is 300V, then you need to "waste" 50V. The LM317 will only "see" a 12V input-to-output differential, and the MOSFET will "see" the other 38V so that differential is always in spec (the MOSFET can tolerate 700V across it's terminals, however the LM317 can only do maybe 20V).

Yeah, most of the current gets dissipated through the mosfet.
Thanks for the explanation @Matador.
That D2, is for the u47 capacitor to discharge. But why does it say R16 "DNP". For what case should it be populated?

They have arrived.
What's next. IRF710, IRF820?
In no way shape or form do I want to "puff" any of the HV versions @Matador.
Got to make sure my circuit works in LTSpice first.
Thanks.

 

[reinw33]

the 317 have survived hard shorts on output with the necessary diodes and current limiting resistor in place.
Max voltage is whatever the the MOSFET can take, 700V types are low cost.
I have made a few PCBs for thru hole parts.
The PSRR is very high, depending on frequency and first RC filter which helps a lot.
I don't recall exact numbers but -80dBc should be feasible.
I have used this setup since the early 1990's.
The FET needs a heatsink, a convenient chassis with requisite isolation is good
Fullpak TO-220 makes that easy. The 317 will not get hot.
I use w.w. resistors in series with rectifier diodes to emulate vacuum diode resistance, wich lowers peak current in the input filter. 10 - 20 uF should be sufficient.

View attachment 130839

This board is looking so nice @Cqwet Dbdfte
I'd be spending way more time in PCB design software if I had a laser printer and stuff. Is this by ExpressPCB?
Even if I had the .pcb or .sch, I wouldn't be able to produce it. I still have to do all that stuff on perf. board.
I got some 15 perf boards recently.