diyAudio (
-   Tubes / Valves (
-   -   Trouble w/ Maida regulator bias supply (

wicked1 20th January 2013 05:16 PM

Trouble w/ Maida regulator bias supply
I'm trying to use a maida regulator for a bias supply, and sometimes it works, but I am having some trouble. I must have something wrong... I may be using it completely incorrectly.
I'm taking the regulated negative voltage and feeding it to a balance and then level pot (like a standard un-regulated bias supply) The maida is set up to deliver -60v, but the tubes I'm using only need about -30, so I am dropping the extra voltage to ground.
That worked great for a while, then the sound faded in my amp.. I traced the problem to the bias supply being far too negative.. I'm getting close to the full -60v on the grids of the tubes.
I literally tapped on the regulator heatsinks and it started working again.
When it is working, I actually measure about -45v at the output of the regulator (and again, It's set for about -60v.. 10k and 200ohm resistors for the voltage set divider.)

I built a new regulator today, and set it for -45v, and it is working right now.. Still through the balance and bias level pots.
(the original regulator set at -60v ran for hours a day, several days before I had the first trouble, so I don' t know if I've actually made any difference)

Any ideas? First guess (from reading other maida trouble threads) is that it needs more of a load.. but wouldn't bleeding the extra voltage to ground cause some current to be pulled through it? In case someone is wondering, the input to the regulator is about -130v.

Thanks for any help.

--Thought I should add.. About tapping on the heatsink making it work... The regulator and transistor are each on their own heatsink, so those making contact wasn't the issue.

SY 20th January 2013 06:00 PM


Originally Posted by wicked1 (
I literally tapped on the regulator heatsinks and it started working again.

Bad connection? Bad chip/pass transistor?

You want to draw 10mA minimum out of the reg for it to work reliably (the spec is 5mA, but it doesn't always work, 10mA does). If your resistor from out to sense is 120R and you scale the set resistor for the right output, the 10mA minimum is guaranteed.

edit: I should have mentioned that it's POSSIBLE that you have an oscillation. Not common with those regs, but it does happen. Have you put a scope on the reg output?

wicked1 20th January 2013 06:32 PM

I have not put a scope on it. I need to step up my game and start testing things....... (I do have a scope and signal gen, etc, but have been lazy)

Am I doing anything wrong by bleeding the extra voltage to ground? Don't the regulators do everything they can to keep the voltage at their set level? So, since I have it set at 60, but am pulling it down to 45v through the resistor(pot) to ground, is that an issue? (it seemed to work well for a few days :))

And when you talk about the draw on the regulator, the current it's pulling from its set resistors, etc, is all you're talking about? Not the demand of the circuit after the regulator?

(I could have damaged a leg of my transistor when I built the thing.. I contorted it a good deal to get it to fit... I'll keep the one I built this morning in there and look for trouble)

SY 20th January 2013 06:46 PM

Maybe if you can sketch a schematic of what you did and post it, we can run through an analysis?

tomchr 20th January 2013 10:32 PM

I'm not sure what you mean by "bleeding the extra voltage to ground". A schematic would be rather helpful.

As SY's mentioned already, the biggest issues with Mike Maida's original circuit is that the LM317/337 requires 10 mA load current to regulate properly. This is normally accomplished by using low resistance resistors in the feedback network (that's how the 120 ohm from OUT to ADJ comes about). For low output voltages (you mention 45 V), this is not an issue, but in B+ regulator applications, significant power ends up being dissipated in the feedback network. That was the motivation for my 21st Century Maida Regulator, actually. Modern regulators don't require as much current to run.

Rather than "bleeding the extra voltage to ground" to get the output voltage you need, I suggest changing the feedback resistors such that the regulator delivers the required output voltage. But maybe that's what you're doing.

Anyway. Let's see as schematic of your circuit. I'm sure we can get this figured out.


wicked1 21st January 2013 01:13 AM

1 Attachment(s)
Ok, here's what I had..
W/ the regulator disconnected, I read about 60v at its output. The voltages here are while it's functioning in the circuit.
When it stops working, I read about -60v where I have -45 on the schematic.. Which is what the regulator is set at.

Now I have a 300 ohm resistor at r2 (on the maida part of the schematic), which should put out about -43v.. I get closer to -45. It has been working all day today. It does still occasionally fluctuate by a few volts for a while.. (10, 30 seconds, maybe), but then stabilizes again.

Thank you both for your help!

tomchr 21st January 2013 04:39 AM

So here is my suggestion. Scrap the 10 kOhm pot to ground and implement the bias adjustment on the LM337 directly.

R2 (between ADJ and GND on the LM337) should be 120 ohm for reasons mentioned earlier.

R4 should be a resistor in series with a trimpot to ground. The trimpot will be your bias adjust pot. You'll have to use the equation for the output voltage that's given in the LM337 datasheet to figure the value of R4.

Remove R35, R36 and connect the output voltage from the LM337 to the right side of R37.

Note that the cascode device (FQP3P50) will need a heat sink. The LM337 might need a small heat sink as well. Don't forget to isolate the packages from the heat sink with mica or Silpads and shoulder washers.


wicked1 21st January 2013 03:12 PM

Thanks, and am I correct that I'm using so little power w/ this application (bias supply) that I don't need a big pot, or to do a parallel resistor with the pot, like Sy did for his RLD screen regulator?

Also, I built Sy's screen regulator for my screens, w/ the PCB he's posted on his site, and here somewhere. For my amp I need about 200v for the screens, so changed a few things..
He's got a 56k in parallel w/ a 180k and 100k pot, for about 300v. I changed the 56k to a 39k, and the 180k to 150k.. The pot is turned all the way to ground, so it is out of the equation.
that makes my set resistors 200ohm and 31k. That should give me 195v, but I get about 215v.
It is stable... Drifts a little over time,but I think that's temperature... Should these things be exact, or is that standard variation? that's 20v more than I calculate I should be getting.

tomchr 21st January 2013 05:18 PM

I don't know what you mean by "big pot". Any pot will need to be able to handle the power dissipated in it.

For R4 (between ADJ and GND on the LM337), I'd use a pot in series with a resistor. Set it up so you can adjust the voltage on the output of the Maida regulator 10~20 V. That'll make it easier to adjust your bias.

The balance pot is already there and shouldn't require resizing.

Get the Maida reg to work reliably before hooking it into the circuit.

One question: If you're using this regulator as a screen regulator, why is the resistor between the cascode and the input of the LM337 (R8) only 5.1 ohm? That would give you a 1 A soft current limiter. In fact, it probably wouldn't work much at all. This resistor needs to be sized such that,

V(D9) - (Vgs(Q1) + V(R8))

is less than the worst case drop-out across the LM337. In your case with the low currents you're dealing with, I'd size it for 2x the expected load current. In higher current applications, I tend to not leave more than 20~30 % margin.

If the soft current limiter isn't effective, you fry the LM337 or the cascode on start-up.


wicked1 22nd January 2013 02:06 PM

Ok, bias supply was not regulating due to not enough dissipation. When the music was cutting out was actually when the regulator started working properly. Thanks for your help there.

then I confused my last question w/ one about the screen reg..
The screen reg does have 47 ohm resistors there, where I had the 5.1 on the bias supply.. (the bias supply has been corrected there, too. I stuck the wrong part in.. used the one off the example schematics, and not mine)

So, the screen reg problem. I built it exactly as per Sy's, except for the divider changes I wrote above.. Basically made it so the divider is 200ohm and 31k. That should give me 195v, and it does w/ no load. But, with a load it increases to about 215. I've built two now, and they both do the exact same thing.
I can't grab the schematic off Sy's site, but it's mid way down this page.. The Red Light District, p2

Thank you again.

All times are GMT. The time now is 04:45 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2