How to protect power supply from user fault?

Thread Starter

TheLaw

Joined Sep 2, 2010
228
Hi,

I'm working on a power supply in which I want to have an adjustable current limit via rotary switch. Like 10mA, 50mA, 100mA, 500mA, 1A, 1.5A. It'll use an LM317 (and then LM337 if I make an equivalent negative supply).

So the way the rotary switch works is it just swaps in a high power resistor which sets the current limit. The voltage regulator is hooked up to this and it just performs as a normal voltage regulator.

Just say I hooked up a load that draws 20A and this supply is only capable of 1.5A, other than a fuse, is there something that I could use that wouldn't be so...permanent? Resettable fuse? Something? Any other protections?

Also, I'm open to any suggestions about how to pull of a cheap and effective current limit. The way I would do it now would definitley work, but it's always going to be burning some power as it passes through the resistor regardless of whether it actually NEEDS to limit current. Maybe a transistor or something?

I don't want to make it too complicated because I have a tendency to do so on projects and it winds up being more than it's worth sometimes.

Thanks.
 

Wendy

Joined Mar 24, 2008
23,415
Well, the LM317 has built in over current protections, but they can fail (especially if the chip gets hot). You could make a constant current regulator in front of the voltage regulator, and turn it into a feature as well as a protection.
 

Thread Starter

TheLaw

Joined Sep 2, 2010
228
Thanks Bill.

I was scrounging around the LM350 datasheet...saw they had this schematic. Would this one allow me to use a potentiometer instead of switchable resistors? (as in would I not need to dissipate a lot of energy?).

I saw on another forum someone with experience say that there was a good way of doing this that involved two LM350s. But there was no schematic listed or anything. You don't suppose he was talking about the one in the datasheet...like this:




Anyway, I think I might just fall back to my original plan of using the standard National 5A constant current/constant voltage design. After giving it another look, it's actually much more simple than I originally thought.

Trickiest party would probably getting workable voltages for the op amp. Probably use a a combination of voltage regulators and maybe a 7660 to get a negative voltage. Shouldn't be all that hard.

1.2V is about as low as I'll ever need in the lab. Really have never worked on much below 5V, though some transistor applications might want lower. But never the less, 1.2V isn't all that bad. 0V would be ideal, but then again, I'm not paying for a $150 PSU.

Just out of curiosity, anyway you could sneak a negative voltage regulator out of the same design? I don't think I could just swap in the negative equivalent of LM350/317 in there...but if I did some modifying, could the same-ish schematic be used?

Thanks.
 

#12

Joined Nov 30, 2010
18,224
Look at a datasheet for an LM723. That chip is designed for "foldback" current limiting. When he current limit is exceeded, the voltage crashes. Very self protecting. There is also an explanation of how to do that with a negative supply in the datasheet.
 

SgtWookie

Joined Jul 17, 2007
22,230
Your image wasn't attached as you implied in your 1st paragraph.

The 2nd paragraph image didn't link properly.

Linear regulators dissipate a LOT of power and are very inefficient.

A 7660 is a very efficient regulator, but it's limited to around 10mA-20mA current, and it's basically just for doubling an input voltage or inverting it.

Have a look at the attached pair of switching regulator designs; the 1st one is a 1.2v-30v buck regulator, the 2nd is a -1.2v to -30v inverting switching regulator. Both operate off the same 34v-36v supply. They are both ~85% efficient or better.
 

Attachments

Thread Starter

TheLaw

Joined Sep 2, 2010
228
Your image wasn't attached as you implied in your 1st paragraph.

The 2nd paragraph image didn't link properly.

Linear regulators dissipate a LOT of power and are very inefficient.

A 7660 is a very efficient regulator, but it's limited to around 10mA-20mA current, and it's basically just for doubling an input voltage or inverting it.

Have a look at the attached pair of switching regulator designs; the 1st one is a 1.2v-30v buck regulator, the 2nd is a -1.2v to -30v inverting switching regulator. Both operate off the same 34v-36v supply. They are both ~85% efficient or better.
Wow awesome circuit. I'm not afraid of switchers, but I can't seem to see much regarding layout, component selection, etc. Also, it seems most adj switchers don't have a very large voltage range, but this does, by some magic.

I guess keeping it as tight together as possible would be good. Do you think I could do this on some decent double sided perf-board? Or would it be better to etch?...I don't have the supplies. That's the problem.

Not exactly sure how much I really require negative voltages at the moment. That might be fine with a cheap separate low power linear reg.

Also, anyway for current limiting on the lt1171?

Thank you everyone.
 

SgtWookie

Joined Jul 17, 2007
22,230
Wow awesome circuit.
They're based on a couple of the example circuits provided in the Linear Technology datasheet for the LT1170/LT1171/LT1172 regulator family.

I'm not afraid of switchers, but I can't seem to see much regarding layout, component selection, etc.
They are SPICE schematics/simulations, not PCB layouts. I didn't build these, but everything in the simulations appears to be well within limits.
There are several specific components in the BOM list, in the LTSpice .asc files.
Also, it seems most adj switchers don't have a very large voltage range, but this does, by some magic.
You're welcome to download LTSpice from Linear Technology's website and try it yourself if you'd like. I'll attach the .asc files.

I guess keeping it as tight together as possible would be good. Do you think I could do this on some decent double sided perf-board? Or would it be better to etch?...I don't have the supplies. That's the problem.
These regulators operate at 100kHz. While not terribly high in frequency, care is always needed in the layout. Short and wide traces are always a good bet. I have only used these regulators with custom-etched boards.

Not exactly sure how much I really require negative voltages at the moment. That might be fine with a cheap separate low power linear reg.
Well, these supplies can both operate off the same input supply, as long as there is sufficient current available to supply whatever your load demand is.

Also, anyway for current limiting on the LT1171?
I didn't specifically design that into the circuits. However, that family of switching regulators has internal limiting to prevent the regulators themselves from overtemp. The nature of switching regulators causes them to require rather large caps on the output; if a short develops in the circuit being powered by the supply, the energy stored in the cap will be dumped through the short. There isn't a good way to limit that current without affecting load regulation.

pot.zip contains a couple of files:
pot.asy - goes in \Program Files\LTC\SwitcherCad\lib\sym
pot.sub - goes in \Program Files\LTC\SwitcherCad\lib\sub
 

Attachments

Thread Starter

TheLaw

Joined Sep 2, 2010
228
Thanks a lot, really. I've been trying the National semi WUBench tools for a while but the interface just gets me POed sometimes. Never tried SPICE, because I really haven't used too many Linear products, I suppose.

In regards to negative, the only things that come to mind for negative voltages are for op-amps and that kind of stuff...which don't draw much. A.) An LT1171 costs like $10. Ouch. B.) I'm not sure if I want to get a 6-7A transformer while I could be fine with a small negative output and a smaller transformer.

Again, maybe I'm just being dumb. I'm still relatively new to electronics though I'm by no means a first-timer. Kind of at an in between stage.

Just meant to clear up "current limiting"; By that, I meant adjustable current limiting, but nevertheless...
 

SgtWookie

Joined Jul 17, 2007
22,230
Arrow Electronics sells the LT1171 for $5.75. Even if it were $10, you'd likely need more than that to pay for the heat sink for an LM317.
 

Thread Starter

TheLaw

Joined Sep 2, 2010
228
Alright that makes enough sense. Any idea on adjustable current limit?

If not, it wouldn't be the worst thing.
 

Thread Starter

TheLaw

Joined Sep 2, 2010
228
I think I'm going to use an LM2576-Adj...It's cheaper. There's more reference for it if I run into problems, and it switches at 52khz which I think would mean less problems with layout and component placement. It can do 80-85% efficiency which is pretty good for what it is.
 
Top