I have an Arduino Duemilanove.
When powered from the USB the 5V regulator (IC4) gets very hot. Within seconds it is too hot to touch. Looking at the schematic I see no reason this part should be even getting warm.
Any thoughts? Am I wrong to think that the part should not be getting hot? Is there something I should test?
What voltage are you measuring on the 5V buss? Are there any loads connected
to the board?
That is very strange. With nothing connected to the Arduino except USB there should be no power going through the regulator. My best guess is that something is wrong with the regulator and the 5V on the output is getting shorted to ground through the regulator.
What else do you have connected to the board??? then disconnect EVERYTHING except power and test again, if nothing changes then does it work otherwise? and if it does work what else is getting hot?, there has to be something drawing the heat/power... typically it is an electrolytic capacitor, leaky or shorted. So feel the board if it works and try to find what else is getting hot. That might well be the clue required... IMO
Doc
It could be that the regulator has gone pop. Do you have anything else at all connected to the Arduino at the time?
Check what voltages WRT ground you have on the three pins of the IC4. They should be 0v, 5v and 0v, with the tab at 5v.
If that is all good, then disconnect the USB and power from an external power source at 5V, say a cut-off USB connector, connected to the jack, or the Vin pin, and measure the voltages again.
You should get 0V, ~3-to-4V and ~4.3V.
If that is all good (if it isn't, you may blow the board with this next step), then try a larger power source, say a 9V PP3 battery. You should get 0v, 5v and ~8.3v.
Docedison:
What else do you have connected to the board???
I don't have anything but "Blink" running.
However, I could have a stack of shields a foot tall and that should not make a difference. The only reason the regulator should get hot is power is flowing through it. The only power that should be going through the regulator should be coming from the power jack and there is nothing plugged in there. All the power is coming from the USB and that doesn't go through the regulator.
Someone correct me if I'm wrong in my assumptions.
majenko:
Check what voltages WRT ground you have on the three pins of the IC4. They should be 0v, 5v and 0v, with the tab at 5v.
I have 4.2 on the tab, 2.9 on the input and of course 0 on the ground.
I didn't see it release it's "magic blue smoke" but it suspect it did.
Does the board actually work, while your regulator gets hot?
You may have a short circuit somewhere in the board, or your regulator itself could have just given up the goose. Personally, at this point I'd replace the regulator; they're not too expensive and it's large enough to desolder and resolder by hand without much trouble.
I have 4.2 on the tab, 2.9 on the input and of course 0 on the ground.
I don't like that 2.9 on the input... I also don't have a Dualanimallove here though, only an UNO. Not sure what the reg is on an UNO off hand.
I don't like these regs... I have 5.02V on the output/tab, and 4.59 on the input. Personally I'd expect a decent regulator to block reverse current like that. Still, compared to mine, your 2.9 on the input is too low. You want to check the components on the input side of the regulator - the caps and diode.
If the regulator is in fact the component getting hot it IS MOST unlikely that it is a faulty input circuit or a defective regulator much more likely something 'down stream' from the regulator. That was the reason that I advised the person to find out if there was anything else getting hot because the 'hot' regulator is most likely to be a symptom of another issue like as I said before a leaky or shorted regulator. It is possible that an input by-pass is gone bad/away and the regulator is oscillating, I've seen that Many times, though usually by putting a slightly damp finger on the IC it will change or stop. A meter on the input/output will show that quickly. There is no possibility of circuit damage as the impedance's at that point are much too low to be affected/changed by that intrusion BUT it will slightly change the feedback that the regulator is experiencing and show in the meter readings as they Will change, perhaps slightly but still change... most of the time or at least that is the way it has always worked for me, I was and still am an "A" grade techie... I got into engineering when I became so bored fixing things that I decided to make things that didn't... I never got good enough to do the latter but I had a lot of fun trying... IMO
Doc
Docedison:
If the regulator is in fact the component getting hot it IS MOST unlikely that it is a faulty input circuit or a defective regulator much more likely something 'down stream' from the regulator. That was the reason that I advised the person to find out if there was anything else getting hot because the 'hot' regulator is most likely to be a symptom of another issue like as I said before a leaky or shorted regulator. It is possible that an input by-pass is gone bad/away and the regulator is oscillating, I've seen that Many times, though usually by putting a slightly damp finger on the IC it will change or stop. A meter on the input/output will show that quickly. There is no possibility of circuit damage as the impedance's at that point are much too low to be affected/changed by that intrusion BUT it will slightly change the feedback that the regulator is experiencing and show in the meter readings as they Will change, perhaps slightly but still change... most of the time or at least that is the way it has always worked for me, I was and still am an "A" grade techie... I got into engineering when I became so bored fixing things that I decided to make things that didn't... I never got good enough to do the latter but I had a lot of fun trying... IMODoc
What you say makes sense to me if I take the power out from the USB and apply it to the Vin. The power is coming in through USB, therefor it is entering the regulator via the "OUT" pin. Current flows through the regulator (backwards), and by the voltages read by the OP, is not what I would expect. Therefore my guess is that there is something awry at the input side of the regulator, like a shorting cap, that is causing excess current to be drawn backwards through the regulator, causing it to get hot. I wouldn't expect the regulator to be oscillating, as it isn't working as a regulator. It's working more as a diode.
More information.
Plugging the board into 12v (actually 13.2V) I have 13.2 and 12.5 on either side of the protective diode.
12.5V on the input to the regulator.
5.0V on the tab of the regulator
And the regulator runs cool.
If you have a DMM with capacitance measuring, measure the capacitance of the electrolytic cap closest to the Atmel chip.
majenko:
If you have a DMM with capacitance measuring, measure the capacitance of the electrolytic cap closest to the Atmel chip.
My multi-meter is supposed to be able to read capacitance but it can not read C6 or C7. I don't know if it is because it is in circuit or because it is out of range.
On my UNO, the upper (nearest the DC jack) won't read, but the lower one will, and reads 42.6µF.
If it is in range of your DMM, then it may well be that capacitor that has blown.
Actually, scratch that - they both read OK on mine.
Maybe it's your DMM that can't read them... 
or replace them both as from your new input the regulator is almost certainly oscillating due to a def input of output filter. Capacitors are cheap even SMT devices. A hint might be to plug it back into the USB power and jump each capacitor's positive lead (the one W/o the marking) Hint: use a cliplead to ground the negative lead of the 'test' capacitor and probe the input and output pins of the regulator. There are only three pins and one of them is ground... so You can hardly go wrong IF it is connected to the USB power, don't try it on an external power supply because you might inadvertently put a 12 v pulse on the 5V line (the USB is safe, fused and limited to 5 Volts) and it is the usb power that is an issue here, not the mains PSU... IMO
Doc
measuring a filter capacitor in circuit might not return useful information because of other devices connected to the same point. Some meters will and some won't. I use a Fluke 179 and sometimes I mistrust the information it gives me when trying to measure a component in circuit. If you can get them replace both of them... they Will fail at some time in their life and you just might have one that is "Dried" out. Literally the electrolyte will leak out and usually leave a whitish to brown film around the base of the capacitors. There is a German company called Wima and they make a series of capactors that will show spots on their cases... those are the ones that need immediate replacement, a wise tech replaces them all as most will fail before the other parts do... IMO
Doc
majenko:
On my UNO, the upper (nearest the DC jack) won't read, but the lower one will, and reads 42.6µF.If it is in range of your DMM, then it may well be that capacitor that has blown.
But if its the cap, shouldn't I see something when I plug it into 12 volts? The regulator is getting toasty dropping only a volt and a half. With 12+ volts across that cap would expect fireworks.
It makes me think that the power loss is coming in the output and going to ground with a little voltage going to the input. What makes me think that is P=I*E. The regulator is getting quite hot so P is rather high. The voltage drop is around 1.3 volts as some value for the current. Now if the same current is going out the input then somewhere else there is that same current at over twice the voltage drop so something should be getting twice as hot. So if the current was going through the cap, it should be ashes by now. And there just aren't that many components on that side of the regulator.
Have I missed something?
likely the linear regulator is oscillating and the 0 value is very possibly an indication of where the issue is. That class of regulators is really notorious just for that.
Doc