[chris719]

Their own website admits it should not be placed near component pins despite the fact they have gone to great lengths to try to make it as nonconductive as possible. Again, anything made with silver particles is a danger when put on pins, mostly because the particles naturally separate from the grease.

 

[Kippei]

I got AS2 and 5. Basically i use 2 coz it is just leftovers

 

[Voodoochile]

While the reg might be dying of heat exhaustion with nothing attached to the output, it should be pretty reasonable with a load attached.

Unloaded, going from 38v to 24v is asking a lot of the regulator. Still in spec, but you start to need a pretty serious heatsink whaen you get anywhere near 10v, even. A 24v 400mA wart with regulator and amp drawing 105-135mA should swing that voltage way down. I'd take a WAG and say to maybe 28-29 volts? Possibly lower. Unloaded, the regulator will not be happy without a large heatsink.

What is the voltage at the input to the regulator with the load attached and running?
...oh I see... it looks like 27.7. You should be fine, even with a modest sink and ordinary silicone grease. That's about as low a differential as you would want to go anyway. Meaning that if your amp was to draw more current, the differential gap would narrow, and you will likely fall out of regulation at times, depending on the line voltage to the wart.

 

[tangent]

Quote:

Originally Posted by Kippei Results: unregulated voltage ≈ 38 V, drop across regulator ≈ 14 V, regulator temp ≈ 40 °C over ambient

Sorry, but you've misapplied my Power Supply Parameter Estimator. It's partly my fault, since I link to it from the TREAD pages. The estimator currently assumes you're using a plain transformer, which is not always the case with the TREAD. The 38V number is the result of rectifying AC, which doesn't happen with a DC wall wart.

I need to add a mode to it that lets you get useful results with unregulated DC wall warts.

 

[Kippei]

Thanks a lot

 

[mono]

If the load is not connected to the power supply, we can ignore the unloaded input voltage to the regulator as it is not passing much, if any current.

If the load is attached and it were a 14V drop, that's not even close to exceeding the limits of an LM317 with any reasonable heatsink. HOwever, that does not mean one can expect good life running one up to 125C.

14V drop with 105mA current, 1.5W is fairly easy to manage with passive cooling. It only seems really hot relative to parts that aren't (linear regulators, aren't dissipating over 1W in about 1cm²). The primary issue was lack of proper regulator-sink interface as rectified with the thermal paste.

 

[Kippei]

I think i isolated the problem. It is the lm317 itself. I was using a off clone branded JRC. So i suspect something was wrong. So off to the shop to look for 1 with a N symbol. (National Semiconductors). True enough, it wasnt feeling any heat at all...