0

A 20mil external trace with a copper weight of 0.5oz can carry ~2.5A with a temp rise of 30C (according to Saturn PCB Toolkit)

If my signal was a digital signal < 10khz with rise/fall time in the low ns with 2.5A flowing, what would happen if I narrow my trace to 7mils for a short distance ( < 5cm) ?

efox29
  • 11,827
  • 9
  • 56
  • 102
  • I would call 30 degrees of temperature rise poor design practice. – Matt Young Nov 14 '14 at 02:29
  • A 2.5A signal with rise and fall times in the low nanoseconds is rather an unusual thing - perhaps you can explain a bit more about what you're doing? The inductance of the trace might well lead to large voltage excursions, and you may also have to worry about radiated emissions. – pericynthion Nov 14 '14 at 02:51
  • @pericynthion the question is more hypothetical. It's nothing that is being designed right now. I will have high current signals, along with digital signals. I combined them all into one for this question to get an idea of what would happen if I actually did have a signal like that. – efox29 Nov 14 '14 at 03:39
  • @MattYoung http://electronics.stackexchange.com/questions/110972/what-is-a-suitable-temperature-rise-for-pcb-traces indicates that 20-30 is considered normal practice. Don't know what "normal" is – efox29 Nov 14 '14 at 03:41
  • I think you'd probably be making an unpredictable fuse on your PCB with that 7mil track ... – brhans Nov 14 '14 at 14:57

1 Answers1

1

The same thing that happens if you make a thick chain with one weak link:). I would feel it was bad design practice, and based on your last question I would say either choose a better connector or use the outside rows for power.

Thinking about it a little bit more hypothetically but without doing any math if you're going to keep the current constant then your power dissipation would rise as that portion of the trace will have a higher resistance. I guess if you held your voltage constant then you'd be able to deliver a little less current.

So I guess if your current is the same in all cases then a 20 mill trace will have the least power dissipation, a 7 mil the most and your solution will fall in between probably closer to the all 20mill than the all 7 mil.

Assuming you have a reference plane underneath you're also changing the impedance of the trace at this neck down point since the capacitance will change here. Although at the frequencies you mention I wouldn't consider this trace to be a transmission line. But that cap change could be import to you depending on your application.

Some Hardware Guy
  • 15,815
  • 1
  • 31
  • 44