LM317, basic voltage regulator, why does it get hot when not connected?

I have built this little voltage regulator circuit: circuit diagram

I am a hobbyist, so these may well be screamingly obvious questions. I ask because I haven't found the answer to my first question.

1. Why does the LM317 get hot even when current is not being drawn, ie when the ends are not connected? (To my intuition it would only get as hot as current that it is 'choking' to reduce the voltage, but if there is no current there is nothing bring down. Where does this evergy come from?)

2. Why are the two capacitors used in this circuit, it works fine without them? (I know this is a simple question so the technical term for it or alink would totally suffice)

Much appreciated!

• Maybe you already know, but in the diagram you linked, the formula for Vout is not correct. It should be, Vout = 1.25V*(1+R1/R2) + Iadj*R1. Swapped resistor names?
– Cerv
Oct 18, 2014 at 13:42
• @Cerv, thanks, I didn't actually see that, thanks for pointing it out! Oct 18, 2014 at 13:45

Not exactly 'hot', but if you fed (say) 30V into that circuit and set the output to 1.25V, the regulator would be dissipating about 150mW with the output disconnected. That's because the 240$\Omega$ resistor draws about 5.2mA from the output, and the regulator must dissipate (Vin - Vout) * 5.2mA even with the output disconnected.

A TO-220 has a thermal resistance of about 80°C/W so it would get noticeably warm.

It's not optional, BTW, to draw that much current. The LM317 needs some current internally to work, and if you don't draw a minimum current of around 5mA the output could go out of regulation (rise above the desired voltage).

P.S. If you use a 24.0 Ohm resistor (don't laugh, I've seen it happen, the markings or the color code is 2 4 0) the regulator would get hot.

Obviously it's not working fine, so you can't say it's working fine without the caps. The missing caps may be exactly the problem. The regulator is probably oscillating at a high frequency without the caps there.

The values on that schematic are also not a good idea. Put a 1-10 µF ceramic cap immediately on the input and output of the regulator. Keep those connections as short as possible. Then you can add a larger bulk electrolytic to the input if you wish. The 100 nF this schematic shows on the input is very skimpy. 10 µF on the output may be OK, but it needs to be a low ESR type, not a electrolytic.

• Hi, sorry I didn't explain that fully, there is no noticeable difference with and without the capacitors. But I will try larger capacitors and see if that stops it from getting quite as hot. I will let you know if it helps. Thanks! Oct 18, 2014 at 13:18

Olin is right about the caps. Self oscillation will dramatically increase power consumption! Sometimes a regulator will oscillate even though you have added the recommended capacitance to the input and output. This can happen if the wires leading to the input and ground are long. In such a case it is like connecting large inductors between the 28 volt source and the regulator. Keep your leads short and add more capacitance than the book says. A 47 ufd aluminum electrolytic cap on the input and output is not going to hurt anything, but may stop long lead oscillation. Always use caps rated for about 2x the expected voltage to account for spikes and voltage doubling due to reflections. So that is 50 volts on the input and output. (These are 32 cents at Digikey) Lastly, regulators do need some standby current. In normal operation its standby dissipation could be as high as 28 volts times this current. A TO220 317 without a heat sink might get warm but I doubt hot.

The main purpose of the capacitors is to keep the voltage from changing during changes in the load or supply conditions. If you start drawing twice as much load current, the supply must also provide twice as much current. But the supply might be on the other end of long wires or PCB traces, which have parasitic inductance. This prevents the supply current from changing instantaneously. During a transient event like this, the capacitors supply current to keep the output voltage constant.

You may have seen this same idea when using integrated circuits like op amps. It's common to put a 0.1uF capacitor close to the chip to provide a low-inductance power supply. In this role, the capacitors are called "bypass capacitors".