Power inductors with temperature sensor

I've epoxied thermocouples to various devices as part of product testing. This is for development/testing, not production, right?

I've epoxied thermocouples to various devices as part of product testing. This is for development/testing, not production, right?

Yes, it is a research project. The problem is that I think the temperature of the coil will be much higher than that of the core, and we use closed-core inductors. How could I get a thermocouple inside the inductor? Of course, your solution is better than nothing, and we will probably do that.

Andreas

than that of the core, and we use closed-core inductors.

Measure the coil resistance. Copper has a positive TC about 3900 PPM/K.

we

nd

ve

g.

er than that of the core, and we use closed-core inductors.

that.

I'd think the core and the coil would quickly reach the same temperature can't be much thermal isolation between coil and core

but if you want t oget fancy something like this:

and the Tc of copper

-Lasse

On a sunny day (Thu, 24 Jan 2013 08:33:35 -0800 (PST)) it happened acd wrote in :

than that of the core, and we use closed-core

If you do that, maybe one of thsoe LM335 sensors, some output voltage as degrees C. and can be input directly to a micro.

John beat me to it.

If it's a research project, then put features on your board that let you shut the power supply down and measure the coil temperature before the heat bleeds off.

It shouldn't be hard to measure it live. Lowpass-filter the AC stuff out with a simple RC thing, and measure the winding's drop with a DVM. Measure the DC current independently somehow, with a shunt or something.

like to measure the temperature in the various components.

sensor.

gets hot, adding a temperature sensor to an existing product does not really make sense.

Actually the core of the inductor gets hot due to losses in the core. The best thing to do is to calculate core and copper losses and see if the core doesn't get too hot if you design your own inductor.

If you buy a readily made inductor the datasheet should tell you up to which temperature an inductor can be used at its rated current and how it should be derated at higher ambient temperatures.

In both cases the temperature of the core should give you a rough estimate whether your calculations where right.

than that of the core, and we use closed-core

degrees C.

Just be careful about putting it close to some nasty switching node; the LM35 series is tricky. Safer would be a thermocouple or RTD. There are really, really tiny thermocouples.

Manufacturers of "closed core inductors" have a bad habit of providing no information whatsoever with regards to losses. At least some will provide information on request (in at least one series, Bourns uses Kool-Mu cores, from which I was able to establish the pertinent information).

Tim

As others have said, one can use the copper winding resistance to measure temperature.

Alternately, one can also drill a hole in the core and install a thermocouple in the resulting well with silicon caulk (which has pretty good thermal conductivity.

Making the well: If the core is iron, an ordinary cobalt drill bit will do. If ferrite, use a diamond drill intended for glass, running in water. Drill deep enough to be near the center.

Joe Gwinn

For heavy machinery, it is not uncommon for an extra winding or two to be employed as a 'copper resistor'. If your coils are custom-wound, just add some fine wire (bifilar wound) in non-inductive winding (i.e. two matched windings in anti-series connection).

It's also possible to get optic fiber thermometers so the pickup of electrical noise won't confuse the issue. I haven't used these, but know that some properties (time delay of ruby fluorescence) have well-known temperature effects.

The old Heathkit H14 printer read the head temperature at the end of each line by measuring the resistance. As it heated up, the print rate slowed down to protect the printhead. Then all you had to worry about was a dry ribbon, since the oil based ink was the lubricant for the print wires.