Figure 6. A series of adjacent waveguides showing
repeated 60° crossing structures in a prototype
optical printed circuit board.
hole can vary between 50 and 200 μm.
Crosstalk also can be carried out on
more complex boards where radii and
crossover structures are incorporated
within the board design. With a waveguide
spacing of ~200 to 250 μm, the CCD can
accommodate the output signals from several guides. This allows isolation measurements of the first- and second-order guides
to be carried out. The isolation is derived
by the ratio of the total illuminated waveguide output power divided by the total
neighboring waveguide output power.
Figure 5 shows an example of the out-
put isolation levels between adjacent
guides. In this case, the level between
first- and second-order guides is 13 dB.
The corresponding CCD image shows the
virtual pinhole diameter that defines the
area in which the total power of a particu-
lar guide is measured.
Coping with increasing complexity
Larger and more complex OPCBs with
greater numbers of guides are envisaged
for the near future. The new measurement
system allows for a wide range of board
dimensions and configurations accommodating boards from approximately 50 cm
down to ~1 cm. It also can measure waveguide outputs up to 90° to the input, increasing the flexibility.
Besides planar waveguides, other
guides involving radii and crossovers have
been measured using a 50-μm launch spot
and a 70-μm virtual pinhole on the receive
CCD, and with an NA of 0.16. Preliminary results showed that the inclusion of
relatively shallow radii in a waveguide of
15 mm can produce a relative increase
in loss of between 3.1 and 4. 5 dB, depending upon the launch NA and an increase in the attenuation coefficient of
~0.32 dB/cm, indicating a notable increase
in propagation loss resulting from the
deviation of the waveguide path.
The inclusion of further similar bends
in the length of the guide increases the
total attenuation resulting from the increase in overall guide length but shows
little difference in the measured attenuation coefficient. More work is intended
to investigate a range of bend radii.
The same launch conditions were applied to measure boards containing a range
of crossing structures to simulate actual
waveguide crossing points (Figure 6).
Each board contained planar waveguides
as well as guides with structures possessing various crossing angles, grouped in
batches of 10. The loss per crossing structure was determined by comparison with
the respective planar guides on each
board. Greater attenuation was found for
low crossing angles, with the loss per
crossing structure for a 10° angle as high
as ~0.4 dB, reducing to about 0.2 dB for
a 30° crossing angle. Crossing angles
greater than 60° reduced the loss per
crossing structure to ~0.13 dB for 10
crossings to a minimum of about 0.05 dB
over 100 crossings as a modal power distribution steady state is reached.
This system is intended for systematic
study of the attenuation and isolation of
these and other optical waveguides as a
function of the optical launch. Further
measurements are planned on prototype
boards produced using a variety of
processes. As these boards become established as high-speed computer backplanes,
the system can provide the necessary
means to characterize them and enable
the manufacturers to improve their technology.
Further development of the system will
increase its operating dynamic range, and
an investigation into the coupling loss
component associated with the quality
of the waveguide end faces is expected.
Meet the author
Robert Ferguson is a higher research scientist at
National Physical Laboratory in Teddington,
UK; email: email@example.com.