[sdiy] Making toast with chips

[Magnus Danielson]

From: "Lincoln Fong" <Linc at christeld.freeserve.co.uk>
Subject: [sdiy] Making toast with chips
Date: Wed, 27 Nov 2002 21:01:01 -0000

Hi Lincoln,

> Can anyone tell me if Opamps should ever run hot (to the touch) and what
> it implies if they are? I notice on some audio boards I'm looking at
> that the NE5532s (and 5534s) are hot to the touch. Some hotter than
> others. None of them is under much of an output load and the supply is a
> good quality +/- 15V fed in groups through fusible 4.7 ohm resistors
> that periodically blow. Each chip has a pair of 47nF ceramic PS
> decoupling caps right by it though they're 12 years old now. Am I right
> in thinking that the ones that are hottest to touch are the most likely
> to cause excessive current draw or is it a red herring? After all the
> boards don't give trouble 95% of the time- not a very good average I
> know. The odd thing is that usually replacing the resistor cures it at
> least for a while.

NE5532s and NE5534s should not run very hot on a +/- 15V diet and relatively
mild load. They should be comfortably warm on your fingers, but notthing should
make you in the neighborhood of wanting to remove the fingers for the sake of
the heat. If you got that hot chips, you've got trouble my friend!

There are only a few cases where one should expect very hot op-amps for the
normal case, but now where into obscurities like the Synthi HiFli phasercore
and such, and they are not the kind of stuff people usually fix, now is it?

Naturally, one of the earliest things to check is a simple short... to ground
or any of the power-lines.

Then, check if they are self-oscillating, and they can do that up in the MHz
range, been there... done that... had my NE5534s bleeding hot by a 1.3 MHz
oscillation once... and yes... it DID sound like crap!

> I notice the hotter ones generally (though not exclusively) have a P
> suffix and are dual types while the cool ones are N suffix.

Do they have different packages also? Different packages can toss of more or
less heat energy per timeunit, or watts if you preferr... thermic resistance
differs.

> Finally do some of you use some kind of (infrared?) thermometer to test
> ICs for symptoms of trouble? I would get one if this were the case. What
> would be a 'dangerous' temperature?

Say you have a component graded up to 70 degrees... but degrees of what?
You may sometimes see them talk about ambient temperature. That's really the
temperature just around the component. If you stick a temperature probe in
there between the components, that's where it is supposed to be 70 degrees at
the highest, that's how ambient temperature works. You may also see a different
temperature, the junciton temperature... this is the maximum temperature that
the NP-junction up on the chip is allowed to have. It's typically 105 degrees
or something. Now, 35 degrees, that's piece of cake... naw... you generate heat
on a rather small volume, the heat has to radiate away through the body of the
component, and the surface of the component has to radiate away to its
surroundings. We sometimes but heatsinks on there to help reducing the
thermical resistance from the hot junction to a cold surrounding. The less
temperature difference (potential) the less efficient radiation (heat current).
In a similar sense, telco equipment may only experience on the top 40 degrees,
and that's ambient temperature for a whole rack of equipment... 1 meter from
the floor, 3 dm from the front-panel. It's 30 degree raise of temperature
calculated from that point to in between the components. They've thought some
on this... those component ranges isn't chosen on random you see...

Cheers,
Magnus - if physics is my friend, why isn't my physics friendly back?