Troubleshooting a variable power supply (LM317 no ref Voltage).

Thread Starter

Yami

Joined Jan 18, 2016
262
Hi guys,
I've been working on some variable power supplies and I posted a discussion trying to figure out how the circuit works here:
https://forum.allaboutcircuits.com/threads/help-me-understand-this-variable-power-supply-circuit.167704/#post-1484380

I must admit I am not 100% sure with some parts of the circuit. Anyway now 'shit has hit the fan', I have got a couple of power supplies fail on me. Its used at the local college for teaching purposes. The variable output should go from 0V - 30V but what happens with the faulty units are the range is from around 6V - 11V. I have changed the LM317 regulator. The regulator I soldered were pre checked in a working circuit and it works fine. I changed the JFET as well. But no luck! There are no change to the result. Voltage measurements against a good working unit agrees with the faulty units. eg. across the JFET and some resistors. I have checked all the diodes everything checks out ok. One important thing I noticed was that across LM317's Adjust pin and output pin it does not read 1.25V its around 8V!.
I haven't de-soldered the regulators to check whether they are damaged yet. If something in circuit was damaging the new regulators what could it be.

I was hoping that you guys could shed some light into what could be the issue and what to check for. Thanks so much.

p.s I should add that these units were working well for years before they started to fail. Some of the damage might have been caused by the students, for instance shorting out the outputs.
 

Attachments

danadak

Joined Mar 10, 2018
4,057
A typical application has the diode across reg to handle discharge of output
cap when power removed from input, for protection.

1583747721684.png

The part has short circuit protection, but shorting output on your circuit, without a simulation,
could have unknown effects.

The lack of 1.25 Vref across pins says part is toast. Are these chinese origin parts ?


Regards, Dana.
 

Alec_t

Joined Sep 17, 2013
11,169
Simulation shows that the output capacitor does indeed discharge via the LM317 output stage. Not good.
Apart from R6 the resistor values are surprisingly low, causing unnecessary power dissipation IMO. Hot parts are not ideal.
What are the input voltage and Vss values?
On the faulty units, what is the voltage at the Jfet source?
 
Last edited:

KeithWalker

Joined Jul 10, 2017
822
Did you check the rectifier and smoothing of the supply voltage? What is the voltage and ripple level? A problem before the regulator could cause problems like that.
Keith
 

Thread Starter

Yami

Joined Jan 18, 2016
262
Sorry I forgot to mention Vss = -15V. The voltage at the input is 48V.
Across the JFet's drain and source its around 11V, with the faulty one as well as the working ones. Source to ground reads something around -1.8V
My suspicion is also that the regulator is getting fried (yet to confirm this). I should add that I'm not powering anything from this power supply during my testing. I have very few spare regulators left so don't want to toast anymore with out knowing what is happening. What could be causing this. I shall check the smoothing of the supply voltage. Could there be an issue with NTC theremistor at the mains input?
Anymore things to check for? Could the diodes be doing something odd when power is applied?
 

sagor

Joined Mar 10, 2019
192
Remember that maximum input to output differential is 40V for those devices. If you feed 48V, and something shorts out the output, you may have close to 48V differential, blowing the 317. You have to make sure the output never drops below 8V.
If you start up and output has a large capacitor at zero charge, you may also be exceeding the maximum voltage differential for a split second.

EDIT: Alec_t beat me to it by a few minutes...
 

Thread Starter

Yami

Joined Jan 18, 2016
262
Oh yeah! It is quite high. But that's from the last board I checked, the functioning units also measured above 40V at the input pin of the regulator.
What might have changed for this to happen as these units have been working for a lot of years without any trouble. Yes there were times when the fuse blew. Even when the outputs were shorted the voltage is dropped across the PTC and the indicator LED would turn off, anyone who is using it would power it off immediately and troubleshoot their circuit. That's how we have instructed the students.
Its really baffling as to what might be causing this.
 

bertus

Joined Apr 5, 2008
20,465
Hello,

If the input voltage is to high for one regulator, you can use a second regulator as "pre-regulator":

LM317_tracking_regulator.png

Bertus
 

Alec_t

Joined Sep 17, 2013
11,169
The regulators that are 'misbehaving' may just be going into self-protect mode.
Why do you have such a high input voltage for a 0-30V regulator? It's unnecessarily wasting energy as undesirable heat.
 

Thread Starter

Yami

Joined Jan 18, 2016
262
Hi everyone thanks so much for the insight and help. @Alec_t is this the common symptom of the LM317 misbehaving? I didn't choose the input voltage its just how the manufacture made it.

Just wanted to update and give some ideas I had, your thoughts are really appreciated :
1. I have changed the JFet, Potentiometer, and voltage regulator and the diode 'V6' of one of the faulty unit. Powered it on not working 100% but was able to vary the output voltage but the range was limited. 5V - 12V. The voltage across adj and Vout still remained at 8V (very odd as all the faulty units show this value). I noticed mainly the voltage drop happens across the 'polyswitch' PTC....:S.
2. Shall I add a diode across Out and In - as @danadak suggested. Maybe an IN4007?
3.Do you have any idea as to what the function of R7 'adj' is used for and how to set it. Its one of those small blue variable resistors with a screw on the top. The manufacture has set it and the screw is glued. Could this be what I have to change for this to be working again?

I'll also add, that one of the units which got fixed also behaved like this having odd voltage across adj and Vout and not being able to vary the output. I kept changing the regulators until it worked. I went through 4 regulators. The unit is working fine now I use it to check all the rest of the new regulators I had. All of the new one checks out good on this board. But not on the other faulty boards. :(

Thanks so much for the help really do appreciate it.
 

Alec_t

Joined Sep 17, 2013
11,169
@Alec_t is this the common symptom of the LM317 misbehaving?
The datasheet says the IC "includes current limiting, thermal overload protection, and safe operating area [SOA] protection. Overload protection remains functional even if the ADJUST terminal is disconnected." AFAIK the protection measures cease as soon as any overload condition ceases. I don't know how the SOA protection handles the input/output voltage differential.
 

Alec_t

Joined Sep 17, 2013
11,169
I'll also add, that one of the units which got fixed also behaved like this having odd voltage across adj and Vout and not being able to vary the output. I kept changing the regulators until it worked. I went through 4 regulators. The unit is working fine now
Perhaps you got a batch of dud/fake regulators. The voltage between Vout and Vadj should be ~1.25V.
Shall I add a diode across Out and In - as @danadak suggested. Maybe an IN4007?
Yes.
 
Last edited:

Alec_t

Joined Sep 17, 2013
11,169
Simulation shows that the 0-30V output should be obtainable if you simply remove all of V12, R6, V8, V5 and V34. No significant current flows through any of those components and I don't see what purpose they or the Polyswitch serve.
It's not unknown for proprietory circuits to include redundant components (1) as an aid to prove cloning by competitors or (2) to make the circuit appear more complex to impress customers. :).
Btw, R7 sets the upper voltage limit.
 
Last edited:

Thread Starter

Yami

Joined Jan 18, 2016
262
Thanks very much,
@Alec_t - yes I had a bad batch of LM317 before, but with those it never worked even once. But I have checked all the current regulators in a working unit (I soldered three wires to the solder pad of the working unit and 'bread boarded' all the regulators before soldering them into the faulty units).
I was under the impression that V8, V34 was needed as the LM317 can't got down to 0V. I thought that the polyswitch was needed as safety if the output were to be shorted. Could the polyswtich be faulty? How would I check this?
Thanks again.
 

Alec_t

Joined Sep 17, 2013
11,169
I was under the impression that V8, V34 was needed as the LM317 can't got down to 0V.
Vss being -15V is enough to pull the Adj pin down below 0V, when the pot is set for minimum voltage. This allows the output to go to 0V.
I thought that the polyswitch was needed as safety if the output were to be shorted.
Output current doesn't flow through the Polyswitch at all. Polyswitch current is ~15mA maximum, dropping to ~5mA when the IC goes into protect mode with its output shorted. I can't find the spec for the Polyswitch you have, but I'd be surprised if it ever sees enough current to cause it to trip, regardless of the output setting and load. Is it perhaps in close thermal contact with the IC to act as an over-temperature cut-out?
Here's the simplified circuit and its sim result with the SetVout pot being cranked from minimum to maximum :-
LM317RegMod.png

Here's the sim result with the load resistance reduced to 1mΩ (i.e. a short circuit) :-
LM317RegModShort.png
 
Last edited:

crutschow

Joined Mar 14, 2008
24,750
Simulation shows that the output capacitor does indeed discharge via the LM317 output stage. Not good.
That's if you suddenly drop the input voltage to zero.
With a typical rectifier-capacitor supply that won't happen.
The input voltage will decay from the LM317 load so won't go below the output voltage while it drops.
 

danadak

Joined Mar 10, 2018
4,057
That's if you suddenly drop the input voltage to zero.
With a typical rectifier-capacitor supply that won't happen.
The input voltage will decay from the LM317 load so won't go below the output voltage while it drops.
I would agree "With a typical rectifier-capacitor supply that won't happen."

But atypical, or maybe more often than one wants, input cap shorts to ground. So one can argue
it takes out LM317., not a big deal. But if supply recycled, who has ever heard of restarting stuff because
of "phantom aperiodic faults", I have. And if short clears, I have had that happen as well, and LM317 not
functioning, will loads get a taste of unintended HV.

I would posit the National guys put that diode in because they had a customer problem for which that
occurred. Or designers hammer testing part on bench prior to release ran into just such an issue. Guesses
on my part.

Soooo I would recommend going with the factory recommendations on this, although I would bet 98%
of LM317 circuits implemented did not use the diode, at least thats what I saw on most customer prints
when I was a NSC FAE.


Regards, Dana.
 
Top