Lm317 constat current source. Caps on the in/output?

[george graves]

So - inspired by the intro of Dave's power supply video(where he started off with a lm317 on the "Dave cad")....I wanted to do some measurements of a LM317 in constant current mode.



I was impress by a jelly bean part.  Throwing a bunch of resistors at it and taking measurements, I saw better than 1% accuracy on it's current output for a given resistance with I(out)=1.25/R

But I was left wondering?  On most regulators (current or voltage) I often see a electrolytic cap on the input(10-470uF), and a ceramic (0.1 uF to 1uF) on the output.  Should a constant current source (0-1A) need caps on it's input and output?  The application note doesn't show it.

I guess I'm asking what are the pro's and cons of having a cap in there.  In the input, I guess none?  It's just like a de-coupling cap, there to provide some current in a pinch.  And the output, I would suspect a slower response time to changing currents/volts?  Is that about right?  If not.......What am I missing?

[fcb]

Don't bother with caps if using the LM317 in this configuration.

I built a Class A amplifier once with two LM317's - one was like this.  The response is pretty good.

[T3sl4co1l]

Do bother with caps.

Sorta.

If you tried hooking that up to a transformer, say to put 1.00A into it so you can measure the winding's miliohm range resistance, you'll end up with an AM transmitter rather than a CCS.

I would recommend bypassing the input to ground, and putting an R+C across the device, maybe 100 ohms + 0.01uF.  That won't wreck the CCS-ness too much (at AC) while keeping it more stable into questionable loads like that.

Best values are hard to guess at, and depend on the particular load.  Just checking that it isn't oscillating is probably fine.  To really evaluate it, you'd need to pulse it somehow and watch the ringing with a scope, which probably isn't worth the trouble.

Tim

[fcb]

The OP doesn't state how he's using the CCS, so in general I would start off without caps. If he finds he's made a power oscillator, then damp it out with C or RC or L or whatever.

[george graves]

Only using it with relatively stable DC source (generic wall wart, or better) - no AC.

I'm thinking of using it as both a constant current source, and a "dummy load" when shorted out to test power supplies.

If that makes any difference.

It seems as if the LM317 doesn't like to do less than 2-3 mA in CCS configuration.  But that's not a deal breaker.

[dannyf]

Parallel a small ohm high power resistor with R1 to set the minimum output current and use a small pot for R1.

A capacitor on the input is always helpful, as long as it doesn't upset the prior stage.

A capacitor on the output should be carefully evaluated.

[Zero999]

No, don't put any capacitors on the output or across the Vout to Adjust pins.

You could connect a capacitor between +Vin and 0V, if you're having problems with noise.