Maida Regulators Redux

[Eli Duttman]

Guys,

Maida's design calls for a 20 K./5 W. pot. I'm having difficulty sourcing the part. My application of the regulator is in refurbing OLD Fisher mono phono preamps (schematic) attached. The filtered B+ will be about 150 V. The regulated rail will be 110 V. per the schematic.

I'm a NOOB, when it comes to the "care and feeding" of regulated PSUs. The ability to adjust down to 1.2 V. is unnecessary in this case. My thinking is to put a 15 KOhm fixed resistor in series with a 5 KOhm pot. Is my thinking correct?

TIA for any info. provided.

[SY]

If it were me, I'd tell myself, "Self, it really doesn't matter if the voltage there is 148 or 152. And the voltage spread on LM317 tends to be pretty low- the equations for the output voltage get me within a percent or so every time. So just calculate the proper value of fixed resistors, and go with that, avoiding running DC through a pot wiper."

That's what I'd tell myself.

[Giaime]

SY is right. And also, why should you keep those low resistance values listed in the application note?

See schematic attached.

[kevinkr]

Hi Eli,
I've had pretty good luck with the LM317, for your application the resistor between ref and out could be about 1.21K (~1mA) which makes selecting the resistor for setting the output voltage really easy.

A 143K resistor would give you about 149V depending on regulator and resistor tolerances. The resistor dissipation is about 150mW, but due to the voltage involved should at least be a 1/2W RN60 type.

Using a minimum of 1mA in the resistor string I have not observed strong temperature dependence in the output voltage, and relatively consistent voltage setting over a spread of parts. The reference current is typically about 100uA, good design practice dictates a string current at least 10X this to minimize errors due to the reference current.

[Brian Beck]

I've had great luck with resistances in that higher range, too (drawing about 1mA). Giaime, the reason the resistances are so low in the app note is that he's trying to ensure that the specified minimum LM317 current is being drawn at all times. If you use larger resistors for lower dissipation, you must have the Maida loaded at all times or it will go out of regulation.

[kevinkr]

Originally posted by Brian Beck
I've had great luck with resistances in that higher range, too (drawing about 1mA). Giaime, the reason the resistances are so low in the app note is that he's trying to ensure that the specified minimum LM317 current is being drawn at all times. If you use larger resistors for lower dissipation, you must have the Maida loaded at all times or it will go out of regulation.

I've read this too, but have not seen issues even with load currents of 1mA with any LM317 I've purchased since the early 1990's.. Pays to check though, because there are quite a few variants of the LM317 and not all of them are likely to behave as well when lightly loaded.

Recent low drop out versions of the LM317 apparently have much worse ripple and noise rejection than the original version which fortunately is still readily available.

[Giaime]

Hello Kevin!

Originally posted by kevinkr
Using a minimum of 1mA in the resistor string I have not observed strong temperature dependence in the output voltage, and relatively consistent voltage setting over a spread of parts. The reference current is typically about 100uA, good design practice dictates a string current at least 10X this to minimize errors due to the reference current.

Yes, that's absolutely true. As a matter of fact, my 317s (yes, there are many versions out there! All different) draw 50uA from the adjust pin, and I never observed great change in voltage with temperature with a string current of 2x-3x the adjust current.

Most of the work is finding the right combination of standard value resistors: most of the time 1W ones aren't available in 1% precision.

Of course if you don't load it (and you have chosen high valued resistors), the regulation fails, but mr. Duttman asked about a regulator for a very simple valve circuit, not a full-blown computer controlled HV bench supply

Edit: remember to put the mosfet on an heatsink! Its voltage drop is almost the full drop of the regulator (leaving 5-8V for the LM317), and the whole current must pass there, so make calculations and choose an appropriate heatsink. Note that the mosfet can have the metal tab at a high voltage: insulate it, or do like me, dismantle old power supplies to find those lovely japan full-insulated mosfets

[Eli Duttman]

Guys,

Thanks a HEAP! The light bulb over my head is slowly starting to glow.

The high mu triodes in that phono preamp draw very little current. I came up with 604 Ohms for the current set value and 53.6 KOhms for the voltage set value. That way, the LM317 will stay in regulation, when the triodes are conducting.

Please advise if my math is incorrect.

[Brian Beck]

Eli,

It will PROBABLY work, but it is not guaranteed to do so. If I read your schematic correctly, your preamp (both channels combined) will draw a scant 1.6mA of plate current from the 110V supply. You have the rare case where the current draw of the amplifier is not enough to guarantee keeping the LM317 awake. Your Maida divider resistor values add another 2mA. So with fingers crossed we'd hope the LM317 would stay awake with 3.6mA drawn. The LM317 is spec'd as needing a minimum current of 3.5mA TYPICAL, but as much as 5mA or 10mA WORST case (depending on whose data sheet you read). As Kevin says, and as my experience has also shown, it will usually stay in regulation at current levels well below 10mA. Were it my project however, I'd probably tweak your resistor values to give the LM317 at least 5mA total to chew on. If it's any consolation, output impedance will drop a bit the more current you yank out of a series pass device. That's what we call rationalizing, because the improvement will be tiny.

[SY]

FWIW, I use 5mA routinely and haven't ever seen a problem. Well, not a problem from THAT; me doing other stupid things, yeah, that's been a problem.