Andrew Seddon wrote:
> > Hard to imagine anyone would do this. Even though the nomograph
> > indicates you can run 750 mA thru such a trace, and only have a 10degC
> > temp rise, there are several "other" things to worry about. Trace series
> > R producing IR drops. Digital noise in this narrow channel. Inductive
> > effects of narrow trace width at hi speeds. Impedance issues. On and on.
>
> I always thought that it was best to use the least ammount of copper you can
> with digital logic, come to think of it I think I made this rule up myself
> so any input appreciated. BTW I just got a board back from the shop, it has
> 8 mil ground and power traces. Seems to be working OK. I dind`t mean to do
> it but it was a rush job and I fogot to manually route the power traces.

You're using enough despiking caps, I take it <G>  You can get away with
a lot on 2Oz boards that won't work on a 1/2 Oz copper clad board, too,
due to 4x the copper thickness.  If it gives you fits, try my wire wrap
wire trick <G>  Bet you didn't route the power/ground wires last, too
<G>

Not much is wrong with using 65 mil traces or so for CMOS logic traces,
IME, it doesn't sour the behavior - 8 mils works fine as well if your
board shop's better than my laser printer works, for short runs
anyways.  Capacitance increases with larger traces, but 1/16" spacing is
fairly large WRT making large caps out of 65 mil traces - Inductance has
a lot to do with clock speed, want to minimize lengths for high speed
circuits, as you know.  Resistance can affect things.

Have to watch out more on 0.020" thick boards (Boeing Surplus sells a
lot of that - Sorta neat stuff, if fragile) as interlayer capacitance
would be lots higher due to 1/3 the plane separation.

One advantage of "fatter" traces for prototyping boards is that it's
easier to break the trace then SMD solder a series resistor in place
when you find you need it on a 40 mil trace than an 8 mil trace <G>
Each of us does a different, Niche market to some extent - you make a
prototype once-off board a little different than you make a production
board (I tend to do things like leaving places where I can drill through
the PCB to solder additional jumpers / components on, WHEN they're
needed due to a design change, on some boards, due to knowing the
designer's habits.  You don't do that on production boards, "usually"
<G>)

I'm no RF PCB expert, most of my board work's been either on fixing
existing boards (re-work is a pain, it does tend to show you what design
flaws people made and what parts blow / break most often -
Electrolytics, power resistors, UJT's / JFets / MOSFET's, occasionally
transistors.  Often CPU's that're attached to a keyboard without
protection networks (diodes or zeners) to protect the I/O pins -
Fortunate we are to be working with PICs <G>) or on new prototypes and
on re-working those - I haven't done a lot of production, 1000+ off
boards, I'm sure for those you do more trying to shrink the board, try
to stay with 1-sided PCB's, or go to truly tiny boards that're
multi-layer, depending, and try to minimize through holes, so not huge
differences as I already do a lot of that but not all <G>  More SMD and
smaller SMD lately, from what I see.  And BGA's, my nemesis <G>

  Mark