On Sat, 02 Aug 2014 23:58:37 -0400, Fred McKenzie wrote:

In article ,
"Phil Allison" wrote:

"Fred McKenzie"

I have a couple of 20 Amp linear DC power supplies with hot transformers.

With no load, one gets up to 115 Degrees F after several hours. The
other gets up to 125 Degrees F. Measurements were made with an IR
thermometer looking at the top of the transformer.


** FYI there is a much simpler, low tech way to find the internal temp of a
transformer.

1. Measure the primary winding's resistance when at room temp.

2. Make the same measurement ( quick as you can) when the tranny's running
temp has stabilised.

3. Calculate the increase as a decimal factor and multiply that by 254 to
get the temp rise in degrees C.

Eg.

Room temp = 4 ohms

Hot temp = 5 ohms

Increase factor = 0.25

0.25 x 254 = 63.5 degrees C

Add on the actual room temp to get final result.

If the number is over 100C there is cause for concern.

Phil-

This is useful information, but my results did not agree with IR
measurements.

I think my problem was that it is hard to measure low values of
resistance accurately. For my "hot" transformer, cold resistance
measurement of the primary was around one Ohm on a Fluke digital meter.
However shorting the meter leads produced 0.4 Ohms.

After several hours, primary resistance measurement was 1.2 Ohms.
Subtracting 0.4 from both measurements, the increase factor was
(0.8-0.6)/0.6 = 1/3. Calculated rise was 84.7º C, plus original ambient
24.4º C gives 109.1º C.

The IR thermometer measured the top of the transformer laminations.
Initial temperature was 24.4º C. Final temperature was 53.5º C. I
could hold my hand on the transformer without screaming, so I am fairly
sure the temperature was nowhere near 109.1º C!

I think a 4-terminal resistance measurement would provide more realistic
readings.

Fred

Yes, you got the 4-terminal one absolutely right. A meter that can use
kelvin clips and the resulting measurements would be much better. This is
normal for resistances under 10 ohms and mandatory under 1 ohm (for 3
reliable digits). For lower resistance or more resolution it gets
progressively worse.

?-)