Paper clip heatsinking for LM317?

[cetoole]

Originally Posted by Voodoochile

Have you considered some 16g solid wire to make your own copper clip? I suppose that would defeat the idea of using what you have on hand (eg, clip or penny).

That could be interesting to see. I have some 14awg solid copper, uninsulated and unenamled, so it would be pretty easy to use for heatsinking. I forget where or why I got it though. I would guess that it would do better than a penny sink, if done right, but I have no way to test it though. Some of those cheaper DMMs I have looked at while searching for a cheap hFE meter have temp probes, but I have no idea how accurate they are.

Tangent: What kind of temp. probes are those? Also, I cant tell from you writeup where you measured the temperature from. I wonder how much additional heatsinking effect you get from the contact with the probes. Very interesting results so far though.

[tangent]

Originally Posted by cetoole

What kind of temp. probes are those?

It's a 5W thermistor. This type is made more for measuring the temperature of liquids, but I like its ergonomics better than the bead type ones that are standard for this meter.

I cant tell from you writeup where you measured the temperature from.

I said above that one of the flaws in the original test is that I measured in several places, for a reason that seemed good at the time, but which I've rejected. My new test will measure from a single place.

I wonder how much additional heatsinking effect you get from the contact with the probes.

A significant amount, no question, but it's the same for every test, so it nulls out.

The only practical problem here is that the numbers I'm measuring aren't the actual values, but reduced somewhat. Between that and the fact that I'm not actually measuring junction temperature, you shouldn't use that to determine how close the regulator is to overheating. The junction temperature will certainly be higher.

[tangent]

I redid the measurements over a week ago, but just now got around to rewriting the results. I decided to make it an article on my site, if only because of the amount of work that went into it. There's more detail in the new article than my posts above, and I've also published the raw data and a partially-distilled version of same, so you can get a more nuanced picture than I gave earlier.

[Voodoochile]

That's great, Tangent! It was informative, if not practical. It does offer a very nice illustration of how different (albeit unorthodox) sinking methods can vary.

It would be interesting to see a similar test conducted with one heatsink: without any paste, with past, with epoxy, with superglue, with thermal epoxy, etc. Just to show that even having a proper heatsink does not guarantee a given level or performance, depending on how well you affix it.

That was not a suggestion, really, it was more a case of thinking out loud.

Nice job.

[ezkcdude]

Originally Posted by tangent

I redid the measurements over a week ago, but just now got around to rewriting the results. I decided to make it an article on my site, if only because of the amount of work that went into it. There's more detail in the new article than my posts above, and I've also published the raw data and a partially-distilled version of same, so you can get a more nuanced picture than I gave earlier.

Great work! Glad to be the "someone at head-fi" who instigated this thought-provoking, if esoteric, article. Tangent, I do have a question. I ended up putting heat sinks on the regulators that were getting hot on my DAC. Maybe I'm paranoid, but at a certain temperature, do those LM317's burst into flames or something? What happens when they get "too hot", and do you know what temperature that might be? My question is really about safety. I am just wondering how safe it is to leave my apartment with DIY stuff turned on (not playing music, though). So far, I've been turning the power (on the surge protector) to my DAC completely off when I leave it unattended.

[tangent]

Originally Posted by ezkcdude

What happens when they get "too hot"

Read the datasheet: they shut down.

If you're asking what happens when they shut down, it means the output voltage starts dropping rapidly. If you're riding right on the edge of the thermal limit, you can get a really ugly output as the thing turns rapidly on and off.

Don't do that.

I am just wondering how safe it is to leave my apartment with DIY stuff turned on (not playing music, though).

I wouldn't worry about any commercial IC catching fire. But, if you have things like discrete buffer stages, that probably doesn't have current or thermal limiting built in.

[ezkcdude]

Originally Posted by tangent

Read the datasheet: they shut down.

Yep. "Thermal overload protection" should have given me a clue .

[cobaltmute]

This thread was just indirectly quoted in Gizmodo:

DIY: The Penny as Heat Sink