Howard Johnson ([email protected])
Wed, 18 Feb 1998 14:49:25 -0800

Electrically, for digital applications, right angle
bends work just fine.

T.C.Edwards, in his book "Foundations for
Microstrip Circuit Design", John Wiley and Sons, 1983,
reports that the primary electrical effect of a right-angle
bend is the appearance of some extra parasitic capacitance to
ground (or whatever is the nearest reference plane) in the
vicinity of the right-angle bend. He provides a useful
approximation for this excess parasitic capacitance:

C = 61*w*sqrt(er) / z0

where w is the trace width, in inches,
where er is the relative permittivity of the substrate,
where Z0 is the line impedance, and
where the answer, C, comes out in units of pF.

Example: Assume Z0=61 ohms (to make the numbers easy), and that
er=4 and w = 0.010 inches. In this case, C = 0.020 pF. That's
not enough to cause much concern.

Compared to the typical digital application, microwave people typically
use much wider traces, on dielectrics with higher relative permittivity,
and at higher frequencies (like 10 GHz). They are also more
susceptible to tiny perturbations in the signal quality of each stage
of processing, since they chain together many stages
and expect good gain flatness across the passband from the whole system.
Overall, their sensitivity to tiny effects like right angle bends
makes them about 1000 times more concerned about the right-angle
bend than we are. Electrically, we can just ignore the right-angle

Another topic related to this that I sometimes hear discussed is
the question of whether, at a right-angle corner, the signal
electrons will simply "fly off the trace" as they attempt to round
the corner at high speeds. This mental image is very misleading.
While it is true that the electrons are wildly spinning, bumping,
and pushing on each other in the metallic lattice, the net
effective drift velocity of a group of electrons in measured
in millimeters per inch. It's very, very slow.

The same phenomenon happens in air. Soundwaves
travel at 1100 ft. per second in air, but the average drift velocity
of the air molecules (meteorologists call this the wind speed) is
very slow, slow enough in a closed room that you can barely detect it.
Believe me, if the wind was blowing at 1100 ft. per second (Mach 1)
you would really feel it.

Back to your question about right-angle bends: I should point out
that some manufacturing engineers will complain about the use of
right-angle bends when using wave-soldering equipment. Mostly they
seem worried that solder balls will get trapped in the right-angle
bend traces, causing shorts. With reflow soldering, this is not
a problem. I have not heard any other credible negative comments
about the use of right-angle bends, but, of course I am always
happy to hear from others whoose experince may differ.

Anybody else got some advice about right-angle bends?

Best regards,
Dr. Howard Johnson

At 09:47 AM 1/26/98 -0800, you wrote:
>Has anyone found any problems electrically in using 90 degree turn PCB signal
>Eric B. Lewis
>Raytheon TI Systems
>6600 Chase Oaks Blvd.
>Plano, Texas 75023
Dr. Howard Johnson, Signal Consulting, Inc.
tel 425.556.0800 // fax 425.881.6149 // email [email protected] -- High-Speed Digital Design books, tools, and workshops