Winning LM317 and unexpected voltage.

[davelectronic]

Recently I've been putting together a linear power supply schematic is below. Only difference is mine is fixed at 13.80 Volts. I am getting the expected voltage with no load, but once I load the power supply the regulator starts whining and the voltage climbs to 15.80 volts. The load shows ok on the panel amp meter, but I'm at a loss as to why it's whining. I have used an led on the regulator board ( not in view as the main board is above it ) What might be causing this strange behaviour from the regulator ? I've used the right capacitance, and protection diodes on the regulator board. As the regulator is just over 6 inches from the main board, I've tried adding some capacitance 470uf and the whining stops. But then the voltage climbs to over 16.00 volts. Any help appreciated and thanks for reading. Picture of circuit and schematic below. Only my version is fixed voltage.

[Benta]

Lower the 240 ohms resistor to 120 ohms. This should be common knowledge after the 40+ years existence of the LM317.
What's the 1 A fuse supposed to do? It makes no sense at all to me.
The 100 ohms resistor is far too high. 10 ohms would be more to the point, perhaps even less (I'm not going to do your calculations).
The 470 uF capacitor on the base connection will do no good, but can easily provoke instability.

Finally, learn how to draw schematics. This looks like a Playmobil or a Kinder-egg assembly sheet. KiCAD is a good starting point.

[T3sl4co1l]

Seconded, nix the 470u.  Poor wiring may oscillate anyway, for which perhaps a ferrite bead on each base lead might help.

Tim

[Benta]

Seconded, nix the 470u.  Poor wiring may oscillate anyway, for which perhaps a ferrite bead on each base lead might help.

Tim

I doubt ferrite beads would do any good here, both LM317 and 2955 are not really high-frequency parts.

One more point: replace the 10 kohms potentiometer with a 1.0 kohm resistor in series with a 470 ohms potentiometer. That way, you'll limit the output voltage to a sensible range.

[T3sl4co1l]

High frequency, no.  All that's needed is a few ohms of loss at some MHz -- well within the bandwidth of "modern" (since ~80s) epitaxial 2955s.

Needed, that is, after trying the above.  Preferably an oscilloscope would be used to identify the nature of the oscillation, which will make this possibility easier to confirm.

Tim

[Benta]

I think that just the modifications already listed will solve the problem. It's not a complicated circuit.

[davelectronic]

Thanks for the reply's. Everything electronic i learn i try to implement to the best of my ability, its no secret my maths is not that good, although i can get there in the end. I wasn't asking for the math to be worked out for me. My opening question was about a linear regulator whining and climbing in voltage under load. These are phenomenons that i have not experienced until now, i don't have a fancy lab to do this stuff in, just one end of a front room and kitchen worktop. Really i didn't explain well enough in my first post. the circuit is the same lay out as the above drawing, but iv'e altered some components. My R1 is 120R my R2 is 1210 ohms using 1K2 in series with 10 ohms resistor, all 0.5 watts. the transistors i used where TIP 147's as i thought it would help prevent voltage sag under load. The capacitors 2 x 470uf electrolytic type where just to prevent transients as the regulator is on a small board below the main board in the photo, some 6 inches away from the main board. I was tired of trying to get higher current from copper stripboard, so purchased a bench shear to make some heavy gauge FR4 stripboard of the home made variety. the wires to the transistors as as reasonably short as i could get them, as the heat sink for them is below the main board, and so internal. the transistor wires come through an aperture slot i cut along the circuit board. I added a single status led to the LM317 small board, just so i could see if its on then the regulator is functioning. I don't expect to get it right every time, but a ringing winning regulator is something new to me. Yes i will try your suggestions, as your clearly more experienced than i am as a mere novice hobbyist.         

[Benta]

Hey, no problem, we all started out once.
TIP147 are Darlingtons, though.
No big issue, I think, but the resistor values might need to be recalculated.

But dropping capacitors all around the place won't help. Keep the circuit clean as we've said above.

[NiHaoMike]

Move the 470uF to the output and the 10uF to across the potentiometer.

[gbaddeley]

Given your construction, it will be difficult to get more than 10-15A out of this design, and at high voltages only. The losses will be too high and heat sinking will probably be inadequate.

[davelectronic]

I will remove the two 470uf capacitors (Red) one at a time, the input first. This power supply is only going to be 10 to 12 Amps with its 200 VA 18 Volt chassis mount transformer. The heatsinks are substantial under that homebrew PCB. And there are two 80mm fans to pull air through the case when it's under load. But I will start by removing those capacitors first and see what happens. Thanks for the help.

[davelectronic]

I did remove the two 470uf capacitors, the whining was less noticeable. But still there, the voltage also climbed to over 16 Volts. So I will start over, I do have a hameg 60mhz old skool scope. But I don't see the point in breaking it out for a dodgy regulator. But thanks for the help, better luck second time round.

[Gyro]

It actually sounds like the perfect time to break out your old scope. As much for establishing that your final implementation is stable as debugging. Not all (very few) instabilities show up in the audible range.

[davelectronic]

Maybe I should have done, it's to late now. That circuit for the LM317 is no more, but for future reference, yes, if there's a next time I will set up that old scope.

[dustooff]

Hi Dave,
  Are the base and emitter connections the right way around?
  Emitter should be connected to bulk cap, and base is between bias resistor and 317 input.

[davelectronic]

Yes the transistors are connected the right way round, although in the past I have made that mistake. And the collector and emitter where the wrong way round. This worked but the current under load was bad. Once I had rectified my mistake, the power supply worked fine. But yes, this time the transistor connections where the right way round. The only thing causing the whining must be my capacitors selection for the LM317, they where maybe a bit high. Instead of 0.1 uf and 1.0 uf, I used what I had, 47 uf and 1.0 uf. But I didn't think that would hurt. And I added an led to the LM317 remote board ( 6 inches) from the main board as a status led just letting me know the regulator was working ok, maybe that caused the whining noise and voltage climb, I'm not sure.

[Terry Bites]

You need base current resistors as well to share the current properly. You also want some kind of current limiting scheme to stop your lab catching fire.

[mycroft]

I think the top of R1 should be connected after the transistors, to the top of C1.

[T3sl4co1l]

Base resistors aren't important with emitter resistors, though like I said, they can be important for AC stability (in which case the DC resistance isn't important, hence ferrite beads being an option).  Base resistors are necessary for switching of course, where you don't use / can't afford emitter resistance.

Tim

[Terry Bites]

Is AC stabilty not OPs issue?

[xavier60]

I would have tried dropping the 100R to 10R as well as removing the 470uF.

[Benta]

You need base current resistors as well to share the current properly.

No, you don't need base resistors when emitter resistors are there for current sharing.

And why do you post a completely different design using NPN transistors?
Not really helpful.

[davelectronic]

I've looked over many PSU design schematics, I preferred the PNP transistors variant as it had a fewer parts count. And it seemed as capable as an NPN design. Although the linear retail market preffers the NPN variants. I'm starting to learn fast about " an after thought " I wish I'd mounted the regulator above the circuit board and closer to it electrically, for me a lot more thinking needs to go into it before deciding on parts components location. I don't have access to PCB production, for a conventional circuit board with proper traces. So the clunky Manhattan style PCB design goes on for now. The circuit can not be rescued now, but maybe I've learnt a few more things for the future. I thank you for all the help offed.

[Zero999]

You need base current resistors as well to share the current properly. You also want some kind of current limiting scheme to stop your lab catching fire.
(Attachment Link)

The problem with that circuit is poor voltage regulation, because the voltage across the emitter followers and associated resistors, will increase under load, causing the output voltage to fall.

A PNP transistor could be used to turn on an NPN transistor, when the current exceeds a certain threshold. Multiple NPN transistors can be connected in parallel, with low value emitter resistors for current sharing.

[davelectronic]

Yes I've seen this very circuit on a data sheet, but it shows a single PNP transistor driving an NPN transistor. I'm puzzled as to why there are quite a few schematics as the LM317 with PNP power transistors, but none mention the existence of poor voltage regulation. That's not to say your wrong, I just haven't come across the negative reports of the circuit at the top of this thread, my opening post. I would try the PNP transistor driving the NPN to see if I can build this layout. Thanks for posting it.