TO220 metal tab temperature compared to plastic body front?

[cupoftea]

Hi,
Been doing thermal testing on the secondary diode of a 20V, 0.8A (non interleave wound) Flyback SMPS (90-265VAC).
This TO220 diode is not heatsunk, but stands with the back of its body about 1mm
away from the earthed metal enclosure wall.
There are no cooling fans in the enclosure.
The enclosure is all metal.

The external ambient was 21degC.
VAC input was 265VAC.
Three different TO220 diodes were tested and compared....

BYVF32-200 (plastic tab TO220, 3 pin)

https://www.vishay.com/docs/88558/byv32.pdf

BYW80-200 (Metal tab TO220, 2 pin)

https://eu.mouser.com/datasheet/2/308/1/BYW80_D-2310586.pdf

STTH1502 (plastic tab TO220, 2 pin)

https://www.st.com/en/diodes-and-rectifiers/stth1502.html

All three diodes had a thermocouple glued to the front of their plastic body.
However, the metal tabbed one also had a thermocouple superglued to its metal tab, and the temperature of this was quickly taken by quickly turning off the power (as the Picologger would have broken otherwise)

Results:
The BYVF32-200 ran coolest at 70degC
The STTH1502 was next coolest at 74degC
The metal tabbed BYW80-200 was hottest at 80degC. (Its metal tab was only a degree less at 79degC).

It seemed a surprise that the metal tabbed one was hottest.
Also, would you say it is unusual that the metal tab temperature was so close to the temperature recorded
at the front on the plastic body of the same metal tabbed diode?

(All diodes were a few degrees hotter at 265VAC compared to 90VAC input)

 

[cupoftea]

..ie the metal tab is 2 degc/W from the junction, and the plastic body front is 31degc/W from the junction...so howcome they are virtually the same temperature in our test? (was repeated 5 times).

 

[dick_freebird]

Low or no airflow would explain it neatly.

 

[cupoftea]

Thanks, and would you agree, the Metal tab temperature will be just 1 degrees Centigrade hotter than the junction?.....the Rth(j_tab) is 2 degc / watt, and the power dissipated is 0.5w (0.8A * 0.7V)

 

[KlausST]

Hi,

I´m not surprised.
* The air flow is unpredictable.
* Mounting and soldering also may introduce some variation.
* And the power dissipation between the different diodes may also differ.

I wouldn´t be surprised if you do the same test again ... there will be different results.

Klaus

 

[cupoftea]

Thanks, i am doing the test often, and the results are coming out pretty well the same though...(The thermocouples were made with high voltage discharge and they look a little kind of "burned" and i am wondering if i have good seebeck junctions though).

Either way, i am sure all would agree, that if the metal tab is 80degc, then the junction could be no more than 81degC? (since its 2degc/w and the diode dissipates just 0.5W.

 

[D.A.(Tony)Stewart]

If there is no cooling on the plastic case it will reach the same temperature as the metal tab. This indicates your cooling design is inadequate.

 

[cupoftea]

If there is no cooling on the plastic case it will reach the same temperature as the metal tab. This indicates your cooling design is inadequate.

Thanks, but the temperature of the metal tab is just 1 degc hotter than junction, and so junction is just about 81degc at full load. This seems good to us?

 

[FvM]

temperature of the metal tab is just 1 degc hotter than junction

Physically impossible, as long as junction is the heat source.

 

[cupoftea]

Thanks though its 2degc/W

 

[KlausST]

The problem is not the "2°C/W"..
the problem is which one is hotter.

If the junction is the item that causes the heat .. then the junction is expected to be hotter, not the metal tab.

Kaus

 

[D.A.(Tony)Stewart]

With low Rth of metal and high Rth plastic , it means almost no heat flow thru the plastic, There is no convection cooling effect, and ignoring measurement error, small external ambient rise from other source for Tcase. If you are happy with conduction cooling only, you're good.

But now are all your caps getting hot too with 50% drop in MTBF every 10'C rise?