Author Topic: board gets very hot..  (Read 11231 times)

Offline neil

  • Sr. Member
  • ****
  • Posts: 438
    • View Profile
board gets very hot..
« on: June 02, 2008, 03:32:35 PM »
Hi,
  I have designed a simple board which uses the ethernet, serial and Atmel 1M serial chip with the M52233 chip (my first , simple project). I have a 1.5A 3.3v regulator with quite a large heat sink connected.  I looked at my power supply, and its drawing approx 300ma (which I expect as I am using the tcp 100MB stack). But the chip gets very hot along with the heat sink, has anyone else experienced this? My worry is that this small project is going all over the world, and in hotter countries, there might be thermal runnaway.

Thanks
Neil

MarcV

  • Guest
Re: board gets very hot..
« Reply #1 on: June 02, 2008, 05:09:20 PM »
The voltage regulator and the coldfire CPU on my demo board run very hot to the touch too. Must be a normal coldfire thingie I guess...

Offline mark

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 3236
    • View Profile
    • uTasker
Re: board gets very hot..
« Reply #2 on: June 02, 2008, 07:18:35 PM »
Hi Neil

300mA is typical consumption for the chip running 100M LAN. The Coldfire does get quite warm too - this is also normal (again due to the internal PHY). I am not absolutely sure whether you are worried about the Coldfire or your 3.3V regulator. You didn't write which input voltage you have - its heat dissipation depends on the difference between the input voltage and the 3.3V output voltage - this loss (Vin - 3.3V)*0.3A = 0.8W at 6V input, 1.7W at 9V, or 2.6W at 12V input is pure loss in form of heat. If this is an issue you may do well looking at a DC/DC converter (eg. SimpleSwitcher) which will improve power supply efficiency and thus thermal management.

Regards

Mark

Offline neil

  • Sr. Member
  • ****
  • Posts: 438
    • View Profile
Re: board gets very hot..
« Reply #3 on: June 02, 2008, 08:33:31 PM »
Hi mark,
  The input voltage is 12v, so we are getting 2.6W of power. If we use a 3.6 Rth   heatsink, then (if I am right), temperature increases 3.6 deg. with every watt. So  2.6 * 3.6= 9.36 deg. + room temprature 25 deg = 34.36 deg.  As this will not go above 70deg, hopefully be okay , but hell of a hot to touch (so I wont touch).

I noticed on the schematic of the demo board shows a heat pad, but noticed that the chip doesnt have one, or am I missing it?

Regards
neil

Offline mark

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 3236
    • View Profile
    • uTasker
Re: board gets very hot..
« Reply #4 on: June 02, 2008, 09:00:57 PM »
Neil

Do you need such a high voltage? Only 28% of the power supplied to the board is used by the processor, so efficiency is very low -why not a 5V input?

The MC9S12NE64 does have a heat pad on the bottom of its chip - but the Coldfire doesn't. I don't know why not since it gets about as hot as the NE64 when it runs...

Regards

Mark

Offline neil

  • Sr. Member
  • ****
  • Posts: 438
    • View Profile
Re: board gets very hot..
« Reply #5 on: June 02, 2008, 09:10:32 PM »
Hi Mark,
   I can use a 5v Input, and this would be better. I just worry in case clients use a 12v input, even though we recommend a 5v input, but something we can think of.

I see you can get small heatsink that can be placed onto the chip , and cheap. Have you ever used these?

Regards
Neil

Offline mark

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 3236
    • View Profile
    • uTasker
Re: board gets very hot..
« Reply #6 on: June 02, 2008, 10:29:22 PM »
Hi  Neil

No - I haven't used these type of heat sinks. They may be a good idea if MTBF is critical (seeing as the failure rate of chips does increase with operating temperate - proportionally(?)) however I would tend to believe that 'normal' applications should be adequately reliable even when the chip is running at what seems (to the touch) quite hot. Can you measure the actual operating temperature to see how much margin you have to the 85°C limit?

Regards

Mark