Mysterious disintegrating 317s

[icedvolvo]

Hi People,

have a technical Q about 317s if someone can help!

I wanted to make a new "smooth" variable power supply for a welder drive motor, motor uses about 0.5A average and about 2.5A inrush and works form 3-24V. OK so I think OK a 317 with a 5K pot will work fine. BUT every 317 gets destroyed!! So I kept adding capacitors and protection diodes all around the thing to protect it but still gets destroyed after a few seconds! Even a 338 only lasted ~10 seconds.

They do not smoke or blow up they just die!!! So far I have a dozen dead 317s and a dead 338. Circuit is stock standard LM317 design as shown in attached gif.

Any and all suggestions appreciated

 

[(*steve*)]

Have you measured the input voltage?

How long do they take to die?

If not immediate, does it correspond to something like striking the arc on the welder?

My initial feeling is that they may be failing as a result of input voltage transients. I'd like to know a lot more about the source of the 24V.

 

[CocaCola]

motor uses about 0.5A average and about 2.5A inrush

I would like to know more about this, what voltage was this measured at? What are these numbers at 3v and 12v and 24v for example?

 

[davenn]

Hi People,

have a technical Q about 317s if someone can help!

I wanted to make a new "smooth" variable power supply for a welder drive motor, motor uses about 0.5A average and about 2.5A inrush and works form 3-24V.

before any voltage issues ... Isnt the LMN317 going to fail with 2.5 amps through it ? !!

Dave

 

[CocaCola]

before any voltage issues ... Isnt the LMN317 going to fail with 2.5 amps through it ? !!

Yeah it's only rated 1.5 max but I suspect it might be able to surge for a short period at some voltage levels, maybe? I honestly have never tested it, but I'm sure any surge over 1.5 isn't going to be healthy, especially if that surge value increases as the voltage goes down and the motor is still trying to start up, thus my initial question...

 

[(*steve*)]

Yeah it's only rated 1.5 max but I suspect it might be able to surge for a short period at some voltage levels, maybe?

I believe it will first current limit, then shut down to limit temperature rise.

I've never seen limits on the size of output capacitors (for example) that would exist if short term overloads (essentially shorts) would be detrimental.

But I shall look over the datasheet once more...

 

[icedvolvo]

Hey thanks for all the replies, there were a few Qs so here goes:

1: Have you measured the input voltage? 24V so 24-17 x 0.5 = 3.5W, 2: Should be OK with a heatsink. But in any case the 317 should shut down if it gets too hot not self destruct.

2: How long do they take to die? Varies, from 2 seconds to 10 seconds

3: If not immediate, does it correspond to something like striking the arc on the welder? Nope, the main arc of the welder is disconnected, only the drive circuit has power.

4: My initial feeling is that they may be failing as a result of input voltage transients. I'd like to know a lot more about the source of the 24V. Mine too but shouldn't the diodes/caps stop that

5: motor uses about 0.5A average and about 2.5A inrush I would like to know more about this, what voltage was this measured at? What are these numbers at 3v and 12v and 24v for example? Current stays about the same for all V and yes at low V, say 7V, the heat in the device is about 10W but again the device with a heat sink should be OK and shut down when it gets too hot.

6: before any voltage issues ... Isnt the LMN317 going to fail with 2.5 amps through it ? No not if is transient for a fraction of a second.


7: Yeah it's only rated 1.5 max but I suspect it might be able to surge for a short period at some voltage levels, maybe? Me too! BUT IT ALSO BLOWS 5A LM338

8: I honestly have never tested it, but I'm sure any surge over 1.5 isn't going to be healthy, especially if that surge value increases as the voltage goes down and the motor is still trying to start up, thus my initial question...Yep agree but this is a device specifically built to manage power supplies and this is a common occurrence in this situation? AND IT ALSO BLOWS 5A LM338

9: Yeah it's only rated 1.5 max but I suspect it might be able to surge for a short period at some voltage levels, maybe? I believe it will first current limit, then shut down to limit temperature rise. Yep thats what I though but alas they just die!!!

10: I've never seen limits on the size of output capacitors (for example) that would exist if short term overloads (essentially shorts) would be detrimental. Filter caps? : combo of 100+10 elctro + 470n greencap + 1n poly both sides! And diodes to stop back current through device.


Thanks to all who replied,

Like many I thought these 3 term voltage regulators indestructible but obviously they are not. It's sort of looking like the most reasonable answer is the 317 can't handle the inductive kickback in the motor and the diodes don't react quick enough to kill it off. If I had a DSO I could try to capture the point of failure but alas I don't :-(

 

[(*steve*)]

What diodes are you using? It is possible that they are too slow. Do you have any schottky diodes? The one across the 317 would be the most important.

You don't need a DSO. Do you have an analogue scope? What does the input voltage look like? What does the output look like?

Please try not to blame the 317. I'm sure it's not their fault. People have been using them for a long time and they are anything but fragile.

What happens after the 317's die? Do you get the full input voltage at the output, or none at all (or something else)?

 

[duke37]

Should there be a capacitor across the variable resistor? This will slow the feedback and perhaps put the circuit into oscillation.

 

[shrtrnd]

I don't know if I'm missing something here?
Are you using a LM317 TO-220 Pkg (rated at 15W with proper heatsink)
Or an LM317 in a TO-3 Pkg (rated at 20W)?
I'd be trying the TO-3 with the biggest heat sink I could find.

 

[gorgon]

If your diagram is correct, the one thing you do not protect against is a positive peak voltage from the motor into the output of the regulator. How about trying with a serial diode from the output capacitor to the motor/diode and add a separate capacitor or a capacitor/resistor over the motor.

I don't know how stiff the output is, but a train of spikes into the ouput transistor could nibble it away.

Another question is, how do you know the inrush current is only 2.5A?
Have you used a scope on this circuit to see how it works? Yeah, I know it's a short life but still.
Have you tested the circuit without the motor, and a resistive load simulating the motor?

TOK

 

[(*steve*)]

If your diagram is correct, the one thing you do not protect against is a positive peak voltage from the motor into the output of the regulator.

The reverse biased diode between the input and output of the 317 should take care of this unless:

• The energy in the spike is great enough to charge the capacitors to a voltage higher than the 317 can handle
• The diode switches too slowly

This was why I wanted to know what the input and output waveform looked like.


We essentially suspect the same problem.


I'm still curious as to the mode of failure of the 317 though.

 

[icedvolvo]

OK Short story is that the people who guessed a spike as responsible were correct! The back voltage was over 80V pp!!!!! And the second part of the story is the normal diodes are hopeless for killing this!

Long story is I went and bought a $30 PWM kit and in combo with a "fast" diode across the motor all is well!

Lesson 1: back EMF is a chip killer!
Lesson 2: dont be a cheapskate; do it properly!!!

As to mode of failure of the 317 I am not sure, diagnostically it just passed the full Vin through to Vout!



Again my thanks to all who assisted, I hope the discussion helps some others as well.