LM317 Variable power supply

[audioguru2]

Yeah, the cheap little 1N400x rectifier diode is too small for much current. :'(

 

[Spagbol]

Don't forget the diodes in a bridge only conduct up to 50% of the time, halving the average current, so you might get away with 1n4000s, wouldn't use the 4001s though as they'd only be good for about an 18v supply!

 

[Um...Me123]

I have built this supply and was testing it before finishing up some wiring then assembling it in the box.
I was using a 25V @ 2A transformer and had it center tapped for the test.  As I adjusted the pot it would jump directly from around 10V down to 1V or lower.  What's causing this?
Thanks

 

[Alun]

The potentiometer is faulty it probably has a break in the track.

I've built a supply similar to this recently, I designed it for 1.25V to 16V and used the LM217 which is the same as the LM317 except it has better characteristics (ripple rejection, load regulation and wider temperature range). Using a 18V 22/3A torroidal transformer the output ranged form 1.2V to 16.4V and I could draw 2.5A 1.2V but the case got very hot as I was using it as a heatsink, I might include a 1.5A poly fuse to limit even more for safety reasons.

 

[Staigen1]

Hi Um...Me123

As I adjusted the pot it would jump directly from around 10V down to 1V or lower

Was it lower then 1 Volt on the output? Did you connect the right terminals? Did you use some protection diodes? Wich package did you use(TO3,TO220 etc)?

The potentiometer is faulty it probably has a break in the track

Probably not, the voltage should not go down below 1 volt then, maybee it is now.

//Staigen

 

[Alun]

You're right, if the track is broken then it would go up to the highest possible voltage.

 

[Um...Me123]

Weird things...
Yes I believe it did go below 1V, I think it went to .7V.
I tried it again not center tapped.  I got an output of 39V (how' s that happening?) no matter where the pot was.  Get this,  when the pot was at the one extreme it would make a slight buzzing noise and make a bad smell. 
Sothing is totally wrong!  Anyway when I un-plugged it, the voltage dropped very slowly and my main switch is before the transformer (same as un-plugging) so if I want to be able to cut the power off instantly do I need to use a SPDT to bypass the binding post and drain the caps?

Now that I think about it maybe it was 29V (sounds alot more logical) but I don't fell like plugging it back it when it seems to be burning.
Should I just get a new pot from radio shack?  I thought pot seemed to be working when I tested ohms on it but I'll check again.

 

[Staigen1]

Hi there Um...Me123

Check my last post and come with answers, and connect it only at the 12,5 voltage at the moment!

//Staigen

 

[audioguru2]

The output voltage would jump lower when the LM317 goes into thermal shut-down.
The TO-220 plastic package with a metal heatsink tab can't dissipate much power, about only 25W with an infinite heatsink and about 20W with a pretty big heatsink.

Alun, the chip in your LM217 must have been way over its rated max temperature. It was trying to dissipate 55W! I am surprised that it didn't go into thermal shut-down. Since it had more than 15V across it then I am also surprised that it didn't cut-back its current to about only 1A.

 

[Um...Me123]

No protection diodes.

 

[Staigen1]

Hi

Ok, have you checked the datasheet for LM117/217/317  that is right here on this site?

//Staigen

 

[Um...Me123]

O S**T!!
THE LEFT TAB IS THE ADJ.??
COULD HAVE SWORN I SAW THAT THE LEFT WAS IN, CENTER WAS ADJ, AND RIGHT WAS OUT!!!!
If the left really is the adj what have I ruined?

 

[Staigen1]

Hi

Dont worry yet! Connect it the right way and test, with the tranny connected to 12.5 Volts!

BTW! Is it a TO220?

//Staigen

 

[Um...Me123]

Any ways to de-solder when you dont have a pump or wick?

 

[Staigen1]

Hi

Printed board?

//Staigen

 

[audioguru2]

A blow-torch de-solders parts quickly!  ;D ;D

 

[Alun]

Alun, the chip in your LM217 must have been way over its rated max temperature. It was trying to dissipate 55W!

I measured the input voltage to the regulator calculated its power dissipation to be  36W.

I am surprised that it didn't go into thermal shut-down.

I don't know the LM217 has a maximum temperature of 150oC, that's 25o higher than the LM317 so it probably helps and my heat sink was experimental. I soldered its tab directly to a piece of 50mm bus bar then mounted this to the case with a piece of in between for insulation. I am aware that the device might not have survived this abuse but as I have so many of them it was worth the risk and I certainly wouldn't recommend this method of manufacture for expensive devices. This abuse may have paid off because the solder junction probebly had a very low thermal resistance because tab remained below boiling point.

Since it had more than 15V across it then I am also surprised that it didn't cut-back its current to about only 1A.   

Looking at the datasheet the minimum current limit is 1.5A, 2.2A typically which is 31.8% higher so statistically it could be as high 2.9A. I suppose I must've been lucky with this and as the device didn't overheat it had no reason to shut down the voltage limit is also a minimum specification which is also not very well controlled I reckon most devices exceed both these ratings.

 

[audioguru2]

Using a 18V 22/3A torroidal transformer the output ranged form 1.2V to 16.4V and I could draw 2.5A 1.2V

Hi Alun,
My math shows a 25.4V peak from your transformer. The rectifiers drop it to about 23.4V.
Subtracting the 1.2V output leaves 22.2V. At 2.5A then 22.2V produces 55.5W in the regulator.

The TO-220 plastic case has a 4 degrees C/W rating. Mounting the regulator to a very big heatsink creates a total of 5 degrees C/W.
55.5W times 5 degrees is 278 degrees C plus an ambient temp of 25 degrees equals 303 degrees C for the chip in your regulator. Way above its max temp.

 

[Alun]

Hi Alun,
My math shows a 25.4V peak from your transformer. The rectifiers drop it to about 23.4V.

I measured 15.something, your math must be wrong because it doesn't account for the internal resistance of the primary and secondary nor the ouput ripple or the ESR of the 10,000uf filter capacitor. Sure 23.4V might be the peak off load voltage but this'll soon drop down to <18V under the high peak current and the troffs in the ripple could be <12V giving an RMS of 15V or so, I could measure this on my oscillosope if you like.

The TO-220 plastic case has a 4 degrees C/W rating.

Here is says 3oC/W.

Mounting the regulator to a very big heatsink creates a total of 5 degrees C/W.

That's 4oC/W now so 36 * 4 = 144oC which is still pushing things a bit but this makes far more sense especially when you consider the practicle results.

Your figures (as usual) predict the worse case senario which excellent and is definitely the way to go if you're designing something.

I'm only planning to draw about 1.5A from my power supply which is why I 've selected the componants I have. I too also expected it to limit to 1A or so, in fact I wanted this as I didn't want the transformer to overheat when overloaded or short circuited. Now I don't know what to do about this and I've thought of the following possible solutions:

Swap the LM217 for an LM317 in the hope it'll over heat and current limit but the ripple rejection will be slightly poorer, also it'd be a pain to dismantle the unit and solder to the copper bus bar.

Add a 1.5A polyfuse, possibly placing this near the device on the bus bar in the hope that it'll trip more quickly when it gets hot, but this would mean another voltage drop on top of the huge 2.5V dropout voltage of the LM317.

Design a foldback current limiting system (possibly using a MOSFET) to limit the current to a low value like 200mA when 1.5A is exceeded, but this would also be a newsense to implement and I don't know if there's even enough space left in the case.

I'm very tempted not to bother and just be very carefull not to short circuit it for too long but there again I don't want it to catch fire.

What do you think I should do?

 

[Spagbol]

Could try putting a bulb in series between the rectifier/cap and the regulator. (eg a 12v/21w car brake bulb?). Shouldn't affect the cct at low current but if you get a short it'll light up and take some power away from the regulator!? Haven't tried it mind....