Current and PCB
Peter L. Peres email (remove spam text)
In my experience I don't think that a 15 mil trace of any length will be
fine with 0.5A over any length of time. I would use a ~1mm (1/20 ~= 50
mils) trace for short lengths and as wide as possible for longer ones, at
The problem with this is, as others have pointed out already, that the
heat buildup from the traces and the parts sums up to heat the whole
board. Not enough, most analog circuit's power requirements increase with
the temperature, so you can contribute to thermal runaway. In my
experience exceeding 0.3W/cm^2 dissipation for small boards used w/o
forced ventilation in closed enclosures is asking for trouble. The evil
part is, that the failures will only crop up after a while with returns,
and not on the lab desk where the unit is open and rather well ventilated.
A 10 mm length of 15 mils passing 0.5A will be fine. 120 mm of it will
have a significant resistance, inductivity, and more, and may start to
contribute considerable heat to the board if you have a bus of these
running in parallel together. The PCB trace tables do not usually cover
bussing and bunching, so over-engineer copiously if using these.
The only proven way to find hotspots in a board is an infrared photo of
it, especially since it is very hard to simulate partial duty cycles and
other difficult to compute currents. If you can borrow an IR camera or a
35mm SLR loaded with IR film and expose 2-3 positions you will have the
data. By borrowing I mean find someone who uses IR photography all the
time and use his services. Thermal insulation installation firms in cold
weather countries come to my mind. The IR film is too expensive to use for
boards otherwise imho.
Just as an aside, my 233MHz K6 computer runs Linux (which saves a lot of
power at low computing load by HALTing the CPU between scheduler
interrupts), that emits a distinct CRACK due to thermal expansion near the
CPU socket when the load is high for 20-30minutes (CPU power trebles from
idle 5W to over 15W on full computing load), I seem to remember that
someone wrote a simulation program for thermal load in pcb's as a thesis,
this may be useful.
Of course you get to use this information at your own risk.
See also: www.piclist.com/techref/pcbs.htm?key=pcb
You must be a member of the
piclist mailing list
(not only a www.piclist.com member) to post to the