Subject:                              Time:  2:11 PM
  OFFICE MEMO         High-Freq. Transmission Lines         Date:  3/1/96

Bob Holmes (Lucent), Dave Hoover (Hadco) and Harry Parkinson (DEC) have
discussed various use of TDR's for "true" printed circuit boards.

At frequencies, not much above 1Ghz, the surface finish "profile" of 
conductor surfaces starts to become a factor in propagation delay.  At high
frequencies, the signal's electrical current flows on the surface(s) of the
conductor.  The current tends to concentrate in the location of maximum
electrical fields.  In the case of a coaxial cable, along the outer surface of
the inner conductor and the inner surface of the outer conductor;  in the case
of striplines, the the current is concentrated on the two surfaces of the
inner conductor that are adjacent to and parallel with, the signal ground
planes and the signal current in the signal ground planes is adjacent to the
inner conductor; finally, in the case of microstrips, the maximum signal
current is in the area adjacent to and parallel with the signal ground plane.

At high frequencies, the signal path does not necessarily follow the
"straight-line" path parallel to the length of the conductor.  Instead, due to
the more limited skin-effected conductor thickness and the imposed electric
field, the signal current follows the profile of the conductor.  Some work was
done many years ago and it was determined that with a "normal" or standard
profile copper foil, the path of the electrical signal takes on a
"trapezodial" current path and actually increase the signal path by about 50%.
 This is why for "serious" high-frequency applications, the "low profile"
foils should be used.  Another problem whin we in the printed board industry
start to use "full additive" or foils with and more resistive later than
copper or other metals as "seed" laters for subsequent copper plating, the
signal current will flow in these seed layers due to high frequency the
skin-depth and if the "seed" layer happens to have a magnetic permeability
greater than 1 (unity), it's a subtle circuit performance killer.  Two of the
major "seed" layers that we have used as "adhesion promoters" for depositions
on polyimide film, ceramic, glass and silicon substrates are nickel and
chrome.  Both are serious high frequency performance killers.

These are food for thought as we get into "serious" high frequency designs.

Ralph Hersey
Lawrence Livermore National Laboratory
Phn:  510.422.7430
FAX.  510.424.6886
e-mail:  [log in to unmask]