Re:RE: [SI-LIST] : GND plane heat dissipation

davidj@chips.com
Fri, 24 Jul 98 10:04:41 -0800

This problem reminds me of a painful experience I had some
decades ago when I encountered a 3W rated wire-wound resistor
dissipating about 2W of power. I found that the data sheet
for the resistor only specifies what's needed to avoid damage
to the resistor, not what's needed to avoid discoloration of
the PCB due to heat. When I unwittingly touched this particular
resistor, I found that it was more than hot enough to burn a
human finger (mine) faster than the automatic human reflex
pulling the finger away. Obviously it's the kind of problem
that arouses the attention of safety agencies concerned with
safety hazards, not to mention issues of general product
quality. A similar issue applies to voltage regulators and
other discretes and IC's of all kinds that dissipate lots of
power in small areas. The device data sheets may, if you're
lucky, mention the thermal coefficients for the device, but
it's up to the designers to calculate what those spec's mean
for device case temperatures. Trying to get a lot of heat
removed from a small area is not an easy or pleasant problem
to solve. A large enough copper square directly under the
component in question and in direct thermal contact with
it (via thermal putty perhaps) might help a little. In
recent years, Intel has been doing a lot of promising work
with heat-pipe technology in notebook PCs.

---- Dave W. Johnson, davidj@chips.com

____________________Reply Separator____________________
Subject: RE: [SI-LIST] : GND plane heat dissipation
Author: Tim Parks <tparks@ferrotec.ie>
Date: 7/24/98 9:09 AM

The transistor used in the Max712 switching circuit is a FZT949 in an
SOT-223 package (high current transistor). It is functioning correctly in
circuit but the body gets so hot in contact with the PCB that the board
temperature in places is 50-60 degrees celsius and higher. I'm concerned
about the effect the high surface board temperature on our digital circuit
and a rubber keymat that lays atop the PCB.

We have a six layer board. Two layers are dedicated to GND and Power plane
for the entire board. It may be possible to a allocate quite a large copper
area in one of the 2 remaining internal layers to an Analog GND plane for
the switching section. As I imagine this is a 'noisey' GND return that
should be seperate from the Digital plane.

This is to be a hand-held consumer product in a very small case, so large
open space for air flow, or fan ventilation is'nt an option open to us.

thanks for you support

Tim

At 04:08 PM 7/23/98 -0400, Andrew Ingraham wrote:
>Is this high a temperature consistent with how much power you expected
>the pass transistor to dissipate? Is it functioning correctly?
>
>If you were to isolate the ground plane, the area around the transistor
>would get even hotter. Should it be heat-sinked? How is the airflow?
>
>Regards,
>Andy Ingraham
>
>

_____________________________________________________
Tim Parks
Technical Manager
Ferrotec Ltd
Unit T9
Maple Avenue
Stillorgan Industrial Park
Stillorgan Tel: 00-353-1-2952529
Co. Dublin Fax: 00-353-1-2953625
______________________________________________________