Re: [SI-LIST] : Design Note on Right Angle Bends

[email protected]
Mon, 13 Jul 1998 10:55:48 EDT

Folks,

lots of responces from SI types, here's a thought from an EMC type ( concern
is mainly about what radiates....):

First, only worry about the traces that carry repetitive level changes. eg
ALE, W/R, CS, LSB, etc. These signals rarely Quasi-peak lower than their peak,
in the EMC test that measures readiated emissions. A board having many right
angle bends but infrequent fast edged transmissions, still appears relatively
quiet.....

Second, my own findings when evaluating boards with various trace routings is
that you can pick up more signal on a close field probe the tighter the trace
bends to 90 degrees. This is often picked up by the EMScan device I use that
shows "hot-spots".

Third, vias do not seem to radiate in the same manner as a trace bend: two
qualifiers though. I have not made this measurement on a board without a
ground plane, and whenever I allow a via to go thriugh the ground plane I seem
to see an increase of noise at the edge of the ground plane.

In a nutshell, don't use right angle bends on the signals mentioned above, and
try to keep them on the same layer.

Derek Walton.
Owner L F Research EMC Design and Test Facility