Thermal Flow into / around a PCB as result of SMD power resistors

[TwinScroll]

Hi,

First post and I just wanted to gauge the Forum members views/opinions/experience on the general amount of power dissipation that you feel comfortable with taking place on a PCB.
I am not sure if there is a general Wattage/area rule that anyone uses and has found works well.

To explain further, I have a small PCB that will contain (besides some control devices) - 12 X Power resistors (2512 package devices) that are rated at up to 2 Watts each at 70 degrees ambient.
http://www.te.com/catalog/pn/en/4-1879501-1
These resistors will each be dissipating a worst case max of 0.55W each, so about 6.6 Watts of dissipation/heat in total on the board.
The area that the 12 loads will be positioned in is about 65mm x 50mm.
I am able to use decent sized copper pours around/under each load, and the board is a 4 layer item too – but I am unsure if the amount of ‘heat’ / thermal flow that I will occurring will cause the whole board to reach excessive temps.
The total board is 65mm (high) x 85mm long. FR4 construction, 1.6mm - and as I say above 4 layers (assume all layers will be 1 ounce copper).
I am confident that the ambient temps will not get much above ~40 degrees, but my concern is more about if there is a way to estimate flow of heat into the rest of the board. Generally the components used are rated upto about 85 degrees C, some slightly higher.

Or, will I just need to design the PCB with the best practise - build a prototype and see how it goes! < Which is what I am heading towards at the moment.

Any thoughts or comments would be great.
Thanks.
TS

[free_electron]

that is way too much power for such a small board.
the core material is a very bad thermal conductor. inner layers don't radiate anything so you only have the exposed copper to deal with. unless you start playing with thick copper and really large surfaces in respect to the resistors can you get some cooling.

what sits on the other side of the board ?

IPC hs a document that shows thermal spread in boards and what you can get away with. it is full if images shot with IR cameras and shows thermograms for different approaches.  for your board i'd say you can't burn off more than 3 watt.

you will also need to take care of having enough width in the conductors or those will get toasty too. you may want to stitch layers vertically to up the power rating. but keep in mind an inner layer can only handle half the current of an outer layer for the same increase in temperature.

[Niklas]

We have a PCB with the size of 50*30 mm with 6 layers of 17 um or 35 um copper on FR4 where roughly half the size of 3 inner layers are used as heat spread. A lot of via stitching around a TO252/DPAK NPN transistor in the power supply that dissipates roughly 1 W. The PCB gets hot but you can still hold it without any problems at room temperature. During a test two weeks ago we increased the dissipation to 2 W and then it became almost impossible to hold it for more than a second or two with getting burnt. The outer 2 mm around the board's edge is not covered with solder mask and it was a huge difference is heat transferred from that frame compared to the masked areas if you ask my poor fingers.

[jmole]

IPC hs a document that shows thermal spread in boards and what you can get away with. it is full if images shot with IR cameras and shows thermograms for different approaches.  for your board i'd say you can't burn off more than 3 watt.

Do you have a link for this? I've been doing some new LED stuff lately and wondering how much I can get away with before heatsinking.

Also, wishing I could afford (or knew how to use) thermal analysis software.

[Alex Eisenhut]

A good first approximation is Saturn PCB tool.
http://www.saturnpcb.com/pcb_toolkit.htm

There's also AppCad from Agilent.

http://appcad.software.informer.com/3.0/

As for thermal analysis, I guess you mean some sort of FEM tool?

[TwinScroll]

Hi, Thanks for the feedback.
The load dissipation numbers were starting to stack up in front of me, and my gut feel was this may be a bit of a push - hence asking the community here.

As a very crude test to get a feeling of power / thermals, I just connected up 2 x WH5 package panel mount resistors on the bench, and had them dissipating ~5Watts (in this case it was dissipating into free air - not into an FR4 board).
My line of thought being that the surface area on these packages would be more than 12 x 2512 devices.
After a few minutes the temperature on the devices bodies becomes way hot, I had not measured it but it was just way to hot to touch and I cant imagine this level of heat/thermal discharge working out to well in a finished design!

I have a Plan B and think I will go down that route (External loads connected by wire:board connections).

[jmole]

A good first approximation is Saturn PCB tool.
http://www.saturnpcb.com/pcb_toolkit.htm

There's also AppCad from Agilent.

http://appcad.software.informer.com/3.0/

As for thermal analysis, I guess you mean some sort of FEM tool?

Yeah FEM would be ideal. I suppose I could try to slog through learning solidworks flow analysis, but it seems like getting the PCB data (copper, stackup, etc.) into solidworks would be a PITA.