Can an LM317 be made to regulate down to 0.5V?

[tylernt]

I'd like to build my own adjustable voltage bench power supply using the venerable LM317. However, I need to test circuits all the way down to 0.5V and the LM317's datasheet says 1.2V is as low as it will go. Can I just add a (big fat) diode onto Vout to drop the output voltage by the diode's Vf, or is there a better approach?

This will primarily be for testing LED driver circuits. I'd like to be able to handle up 1A Iout.

 

[ronsimpson]

I like the LT3080. It will go to zero volts.

 

[tylernt]

I like the LT3080. It will go to zero volts.

That is a sweet chip! I like that way better than working around LM317 limitations, plus it's only appropriate to use a Linear Technology power supply to test my Linear Technology LTC3490 driver circuits.

Thanks!!

 

[MrAl]

Hi,

That LT chip looks nice yes. I like the LM317 too though.

To make the LM317 go to zero you can use two diodes on the output, but that will sacrifice regulation quality. It wont regulate over temperature and load current as well.
If you use two diodes to make a negative reference voltage you can regulate down to 0v, with just some temperature dependence. If you use a reference diode you get around all these things.

 

[Warpspeed]

I'd like to build my own adjustable voltage bench power supply using the venerable LM317. However, I need to test circuits all the way down to 0.5V and the LM317's datasheet says 1.2V is as low as it will go. Can I just add a (big fat) diode onto Vout to drop the output voltage by the diode's Vf, or is there a better approach?

The voltage reference pin is 1.2 volts, which normally comes from a voltage divider off the output.

You need to figure a way to get a +1.2 volt feedback voltage for the control pin that tracks the +0.5 volt output you require.
Just dropping the output with a diode is only half the problem, you need to also close the feedback loop that controls the output of the LM317.
You may be able to do it by using a third resistor, and connecting it between the control pin and a fixed voltage source.
I naver tried this myself, but it may (?) work.

 

[ronsimpson]

If you can make a -2.5V supply and connect the pot to -2.5V then the output can go to 0 volts.

I used a TL431 to make -2.5volts.

 

[4pyros]

I like the LT3080. It will go to zero volts.

I dont Like the package layouts though. Why not have the output on the outside?

 

[ronsimpson]

If you want to add current limit, you can sense the output current by monitoring the current in pin (IN).

 

[alec_t]

If you can make a -2.5V supply and connect the pot to -2.5V then the output can go to 0 volts

....or you could use an AA cell as a -1.5V supply.

 

[MrAl]

Hi again,

Yes you can use a battery as long as you dont let it run down at all. For every 0.1v the battery runs DOWN, the output runs UP by 0.1v.
The current draw is roughly 5ma so that's something to keep in mind.

Using a larger battery and voltage reference diode could work too, as long as the battery never runs down too low where the reference cant regulate anymore because then the output voltage starts rising 1:1 again.

 

[dougy83]

This will primarily be for testing LED driver circuits. I'd like to be able to handle up 1A Iout.

So are you making a current source or a voltage source? If it's a current source (which would make sense for testing e.g. LED brightness or LED forward voltage), then the output will go down to 0V without any issue using the LM317.

 

[tylernt]

So are you making a current source or a voltage source? If it's a current source (which would make sense for testing e.g. LED brightness or LED forward voltage), then the output will go down to 0V without any issue using the LM317.

This power supply is going to be used for testing LED drivers, such as the Zetex ZXLD381, Linear Tech LTC3490, homebrew Joule Thief circuits, etc. I want to play with various inductors, smoothing caps, etc. and see how they perform at various voltages to simulate fresh/dead batteries. Particularly with the Joule Thief, it seems to be a badge of honor amongst builders to claim the lowest startup and operating voltages. I could just buy a nice commercial variable voltage bench power supply but building it seems like more fun.

I do plan to build a constant-current LM317 circuit as well, for testing LED emitters directly, apart from the driver. But that's a separate project...

 

[Warpspeed]

There is no real problem with using a battery, any degradation of the reachable minimum output voltage will be gradual, and easily fixed with a fresh battery.
Just make sure the battery is totally disconnected when the power supply is not in use.

Even if you don't plan to use a battery in an eventual final design, they are a great way of trying out experimental ideas in low power circuits, if you don't have the luxury of a whole bunch of expensive laboratory power supplies to play around with.

 

[dougy83]

This power supply is going to be used for testing LED drivers

Sorry, I missed the word 'driver' in your post. You can make your own regulator circuit using an opamp (e.g. LM358) and a transistor if you can't get the LT3080. I'll attach a schematic (you'll notice it's very similar to the LT3080 block diagram). The voltage on the + input of the opamp will set the output voltage.

 

[tylernt]

LT3080 circuit complete!

I finally got around to building this:

**broken link removed**

**broken link removed**

This is a junk-box random 5V wall wart, 500K pot, two 47uF caps (vastly oversized, but what I had laying around), and the LT3080 on a Radio Shack board in a Radio Shack enclosure. The + and - output hoops are made from a paper clip; I have leads with those little hook grabber things that will attach to feed whatever circuit I'm playing with.

(I also have a big heatsink for the LT3080, just haven't attached it yet.)

 

[Blueteeth]

Hi,

Although its a bit late... you can get the LM317 down to 0v by using a-5V bias supply from a 555, and a couple of diodes/caps. Its woefully ineffcient but the current draw the the circuit is <15mA so efficiency doesn't matter.

That said, always nice to use a fancy chip Linear Tech, though expensive, have some awesome stuff for power control.

 

[bountyhunter]

Hi,

That LT chip looks nice yes. I like the LM317 too though.

To make the LM317 go to zero you can use two diodes on the output, but that will sacrifice regulation quality. It wont regulate over temperature and load current as well.
If you use two diodes to make a negative reference voltage you can regulate down to 0v, with just some temperature dependence. If you use a reference diode you get around all these things.

That's the way, but use a decent 1.23V reference diode instead like LM4041 or anything similar. If you have no negative voltages available, use a simple 555 charge pump to generate a negative voltage. Connect a 1.23V reference diode from ground to the negative rail and then "ground" the 317 to the -1.23V point. Now it will regulate down to zero V.