William Donzelli wrote:
Exactly. And the spacing between the tracks
matters a lot. And having
sharp corners, and ...
Be careful here...when using very fast logic (10G GaAs), you do NOT want
90 degree sharp bends. The best bet is to truncate the corners in a
specific way so that the edges of the pulses reflect the direction you
want them to go, and not back at the source.
Ah yes, I found out about that when I looked at trace widths of the boards I
had contracted to build, and found that the actual trace width at each of
the corners you describe got down to the several mil area. Of course I found
this out *after* I had taken on the job. And yes, the boards came out just
fine although the reject rate was much higher than I would have preferred.