Temperature rise (circuit potted in epoxy)

[Jester]

I have a small circuit, PCB is 1" x 1" that will placed in a small plastic enclosure 1"^3 and then filled with thermal epoxy.

Under normal conditions, power dissipated will be ~200mW.

Under fault condition (end user connects incorrectly), power dissipated could rise to 700mW

Can anyone help me determine temperature rise?

 

[BradtheRad]

700 mW is enough to make a little TO-92 transistor so hot that I get burnt if I hold it between my fingers.

My uneducated guess is, that if that same amount of heat were drawn away and dispersed throughout a cubic inch of heat-conducting epoxy, it would remain cool enough that it will not burn.

 

[FvM]

It's the surface that matters. A standard number for convective heat dissipation is 10 W/(m²*K). Applied to your cubus, you get 38 mW/K.

 

[D.A.(Tony)Stewart]

I am not sure when to apply Convection Loss or Conduction Loss as the dominant factor.

When the material is highly conductive such as Aluminium, we know Convection resistance rules for the exposed surface area, which might be only 5 of the 6 cubic faces. This would give the surface temperature rise above ambient.

But if the thermal conductive surface is filled with a very low thermal conductive material such as epoxy, then I think linear Conduction loss may be a major factor for calculating junction temperature rise with the thickness of insulation being dominant to compute the centre junction rise.

I know a class of LEDs used encapsulated in epoxy where the junction is 5% of the case diameter are often rated at RjA of 500 deg C/W, which might closer represent this cube model of epoxy. This would project a 100 ' C rise of a junction with 0.2W. if there was only one (1) source. If this were true, I would incorporated external power dissipation with a Polyfuse for a user fault condition for increased power dissipation.

In any case , no pun intended, using any LED , with -2mV/degC electro-thermal response, this can easily be tested in a 25mm cube of epoxy vs 5mm cylinder dome of epoxy to see the difference using 100mW and measuring the voltage drop intermittently.

 

[andre_luis]

Once the source of the heating is confined into a plastic enclosure, this probably could be another significant problem, regardless of the capacity of the internal filled material conduct the heat with greater or lesser efficiency.

 

[Jester]

I will embed a thermocouple inside and pop it in the temperature chamber and see what happens, I'm guessing very hot. There will be several copper wires coming out as well and they should help pull some heat out. I will post results in a couple of weeks, chamber is busy for now.

 

[KlausST]

Hi,

If the most heat dissipating device is a BJT, then reading Vbe might give you the best indicator of chip temperature.


Klaus

 

[chuckey]

I don't think there is a problem providing the thermal conductance of the epoxy is low enough so there is not a big temperature gradient between the centre and the outside of the cube. Think about a one watt resistor with the same dimensions, it would not get even warm.
Frank