LM317 Power supply

[n0b]

So I found an LM317 power supply project on Instructables and am wondering if it is any good. I am a beginner and thought that if this is going to coast $20 on Jameco, it would be a good project. I have heard the unreliability in the LM317 voltage regulator. I was also thinking of doing a constant current supply with the power supply so I could have current and voltage.

Anyways thanks for any help you might have.

Mr.Clark

http://www.instructables.com/id/How-to-Build-a-Bench-Top-Power-Supply/

[Simon]

or just read the datasheet and you find it full of examples. LM317 is a generic name used by many manufacturers so checking the datasheet for the one you have is a good idea.

[Radio Tech]

Built one like that last year.





I used an old power supply case that was gutted.
Other than the circuit I used a transformer and bridge rectifier.
Was just something I threw together sort of quick.

[Seekonk]

I don't think I've used a 317 in 35 years.  As long as it is just a learning experience and you know it will be next to useless, go for it.  The learning experience will be when it burns up.  If you must, add a PNP power transistor and resistor to it for more power.  See app note.

[Simon]

I've just designed 2 LM317D's into a new design. SO8 chips, nice and cheap and easy to use do the job just fine. I am only using a few mA though.

[Yansi]

LM317 is not suitable for regulated (meaning with variable current limited output) supply design. Why?

a) doesn't go to 0V, only regulates from 1.25V which is usually not the biggest showstopper, that is
b) simple current limiter using external shunt and NPN transistor doesn't work when load voltage is less than 1.25V.

Hence I'd not recommend LM317 for simple PSU design, it is not robust enough. Such supply will be very easily destroyed by overcurrent: By overheating the IC, or blowing the mains transformer or whatever, as the shortcircuit current of the 317 is 1.5A+

Also these "simple instructables" mostly lack any other basic protection circuits, like antiaprallel output diode or out-to-filtercap diode which is there to protect against external supply voltage connected to the terminals of powered-off LM317 regulator.

[wblock]

I put one of these together a few days ago because I needed something to simulate the voltage of a discharging 12V battery.  It's powered from a 19.5V DC laptop power supply.  Only reading this thread made me realize that I wired the LM317 backwards, as if the pins were in, adj, out.

What is really odd is that it works.  No smoke, and the pot adjusts the output.  Even ran a 1M string of white LEDs for a little while.  They were not very bright, which is probably a clue.

I hurt my brain trying to figure out how it can work.  I can post pictures if anybody is interested.

Edit: a few pictures attached.  Under load, 8.10V was as high as it could go.  Still no smoke or anything obviously wrong other than not being able to supply adequate voltage under load.

[Simon]

Are you sure it's wrong ? remember many manufacturers make them. I have stock of PN2222A that are oposite to the datasheet I happen to have.

[wblock]

Didn't find any datasheets with alternate pinouts, but it's possible.  Might also be a wiring mistake.

The regulator was pulled from a scrap board and says:
(Motorola logo)   LM
317T
QSR928

Yesterday, I built a correct version on a different board.  The only other scrap regulator on hand was a 317M.  It really does not want to provide enough current for the LED string, but does so at 11.5V, dropping as the regulator heats up.  The next step is to swap the 317T into that board.

[wblock]

Desoldered the poor triply-recycled LM317T from the backwards circuit.  It voiced its disapproval by shedding the skinny part of one leg.  No other stick-through LM317Ts being found in the scrap pile, I soldered a resistor leg on it.  Jammed into the correct circuit, it works just fine.  15.89V is about the maximum with the LEDs attached (my eyes!), but it does still work.

[KL27x]

It really does not want to provide enough current for the LED string, but does so at 11.5V, dropping as the regulator heats up.

That's because with a 19.5V input, you're dropping 8V through the regulator. That's 6.5V more overhead than the regulator needs. That's 6.5W of extra heat in the little lm317 package per amp of current.

Since your input voltage is fixed, you can drop some of that voltage with a power resistor on the LM317 input. Calculate the maximum current you want. Use that to figure the series power resistor value at 6.5V.

Per instance, say you want 2A max, your power resistor = 6.5V/2A = 3.25 ohms. At the maximum current draw of 2A, the input voltage of the regulator has now dropped to 13V, which is enough to get an 11.5V regulated output, still. The regulator now dissipates only 3W at this current, rather than 16W. The series power resistor dissipates 13W, taking that load off of the regulator. 

At 3W, you will still probably need an efficient heatsink +- a fan.

At say 500mA, you could use a series resistor of 26 ohms. At 500mA, your regulator will max out at 750mW heat dissipation. The series resistor would dissipate 3.25W.

The series resistor caps the peak output current. Going above that chosen current value will cause the Vout to drop below your target.  But it will increase the effective max sustainable current. Obviously, this only works if your Vin is more than 1.5V greater than Vout.

 

[MrOmnos]

Hi, just a heads up for your project. If you are going to using a STMicroelectronics (you can google the logo) LM317T then make sure you do not exceed the input voltage more than 30 volts. It will fry the IC although the maximum input voltage is 40 volts. I recommend you get the IC from other manufacturers. Either China has been flooding the market with fake ones or the St has outsourced it manufacturing to china and doesn't care about old ICs. And if you get and ST Lm317t working with an Input of more than 30 volts, then please upload a picture. It would be really helpful for me. You can search JumperOnetv o youtube for better schematics.