Current on a PCB
Dan Michaels email (remove spam text)
Don Burdette wrote:
>Dan Michaels wrote:
Don, hopefully my designs are a little more precise than my
Yes, I always calculate everything I can - general and "worst
case" - during the design phase, add in a substantial safety
margin on things like power dissipation, always use the widest
possible traces for power/gnd as a matter of course, cook it up,
and then test/measure the resulting hardware. There are always a
few things that take a 2nd cut to get right, but most things are
pretty good on the 1st go.
After I sent that last memo, I realized it sounded a lot like I
just totally wing everything. Not quite. <:-)).
>The topic of 'fusing' the traces is actually very different from the
>topic of 'reasonable' currents. The fusing (melting or vaporizing)
>currents are at least an order of magnitude higher than what you want to
>use. A trace that will melt at 10 amps will discolor the solder mask and
>the substrate, even melt the glue and peel off the board at currents of 1
>amp or less. Fusing normally occurs only if there's a major fault
As I pointed out last time, I was especially struck by the fact
that the nomographs show that it takes "only" a 3x increase in current
to go from the 5degC to the 100degC temp rise curve. That's an
enormous effect for not so enourmous a change in current. Essentially
no effect at all to blistering your finger/whatever.
I would not want my pcbs to be designed anywhere close to the point
where this could occur. I would calculate the worst case currents as
best as possible, find the trace width that produces say a conservative
10degC rise at that current, and then try to allow a substantial
safety margin over **that**.
10degC rise for a 100mil 1oz cooper trace is ~4A, so I would think
that operation at currents <= 1A will "never" be a problem here,
since it would take ~12A to get 100degC rise. At currents much beyond
1A, I think I would go with heavier copper, 2-3oz, from the start.
Looking at the nomographs, and thinking about the physical size of
relays and switches required to handle large currents, it seems you
pretty much have enough room to widen/thicken the traces so heating
should rarely be an issue. Eg, for a 10A design, a 300mil 1oz trace
gives about 10degC rise - there should be plenty of room for parallel
150mil traces top and bottom - and it would handle 30A before you get
to 100deg rise. Plus, there is probably room here to widen the traces
further. 3oz copper would be much better.
>The vast majority of PCB's are made on either phenolic or FR4 fiberglass,
>with one of several mask materials. Does anyone know the max temp of
>this stuff? If nobody knows offhand, I'll call my PCB vendor on Friday
I could not find this in Johnson's book, "High-Speed Digital Design",
but as indicated above, I think one can generally design so that you
shouldn't have to worry about it.
- Dan Michaels
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