Maida regulator

[billr]

Hi

anybody had much success with the maida regulator running about 600v?

I am intrigued as to what pass device was used. The stuff I have seen in morgan jones' book are all at 300v, using a MJE 340, however, they are no good at 600v.

Look forward to hearing from anyone.

kind regards

bill

[Geek]

Hi,

I don't know your application, but the MJE340 will be fine as long as the difference between input and output voltage does not exceed 300V.

Cheers!

[danFrank]

Hi,
Any of the horizontal output transistors (without internal damper diode) used in CRT TV sets work well. I like using the 2SD1403 as it's easy to mount and has IIRC 150 watt dissipation rating. The problem with high voltage transistors is that they have low gain, so you will have to "stand it up with another transistor" as a darlington pair. For the lower transistor you can use an MPS-A42 to drive the bigger power transistor.
Take a look at Glass Audio issue 1/89 for an article Joe Curcio wrote about the Dynaco ST70 with regulation. His implementation of the Maida regulator is very good and I have only blown this set up once in all the times I have used it, by turning the mains power of an amplifier off and on several times rapidly (Just to see what would happen). The main item that has to be protected from overvoltage is the LM 317 regulator and you can do this by puting a zener across the input and output of this chip.
I would not recommend using series LM317's like that one guy did on the other forum as that looks like blown chips waiting to happen.
Good luck... Daniel

[SY]

Use a 1kV transistor or MOSFET and it will work fine and forever.

[Brian Beck]

Yes, a 600 volt Maida regulator is possible. Years ago, I made a pair of 1000 volt regulators (22mA) for a driver circuit (don't ask) using the Maida concept. Worked great. They never once failed on me, despite many cycles of turn-on and turn-off during development testing. If I were to do this again today, I would use far, far less capacitance on the output of the LM317, and less in the raw supply, and probably something more modern than the BU208/MJ12005 Darlington combo. There's enough energy storage here to turn a screw driver into vapor, or worse, send a human to the morgue! Be careful!

There's something magical about a delicate solid state part floating on top of a 1000-volt mountain without a care in the world...

[SY]

It looks like that regulator is running just on the hairy edge of dropout. Or maybe beyond. A higher voltage Zener would do better, I think.

[EC8010]

I see you used the "high voltage" variant of the 317.

[Brian Beck]

Originally posted by SY
It looks like that regulator is running just on the hairy edge of dropout. Or maybe beyond. A higher voltage Zener would do better, I think.

Yes; you're observant. The 317 is running with about 2.6 volts across it (6.2 volts, minus two diode drops, minus voltage dropped across current-limiting 100 ohm R). That's actually quite enough, but it is close. At these currents, drop-out is at about 1.5 volts, typically. I seem to remember changing the 100 ohm current limiting resistor to 47 ohms, which would have given the 317 another volt, but I'd have to check.

I don't propose that anyone should actually build this 1000 volt Maida, but I just wanted to show that it can be done, and that the 600 volt supply that was in question is feasible.

Maybe we ought to have a contest for the highest voltage Maida that can be built! Tesla would love it.

[Brian Beck]

Originally posted by EC8010
I see you used the "high voltage" variant of the 317.

Ah yes, the LM317"H". Of course that's the old TO-5 style can, which I think is now long obsolete. With a mere 60mW of dissipation, it needed no heatsink.

[Brian Beck]

Back to the original question: When the Maida is used to regulate any high voltage (say 100 V or more) the output impedance increases because the resistor from the ADJ pin to ground becomes far larger than the resistor from OUT to ADJ. This changes the loop feedback ratio. The data sheet refers to putting a 10uF cap from ADJ to ground across the shunt resistor for ripple reduction. While it does help with ripple, it also helps to keep the output Z lower at audio frequencies by improving loop feedback performance. See picture. If you did go to 600 volts with a Maida, I would suggest using a two 22uF/450V caps in series across the shunt resistance, which can be split into two parts to keep each cap "biased" to 300 volts. You can also put zeners into the ADJ path in lieu of a resistor, but that may be overcomplicating things. Use low-tempco resistors for the feedback network so that resistance-value drift won't change the voltage setting much. Even so expect some output voltage creep.