The goal is to design a machine capable of testing a kilometer of WLS fiber to determine its diameter, look for any surface anomalies, and search for light leaks from the fiber indicating a nick in the cladding, or other general defects.
The general design of the testing apparatus is shown in this picture,
or interactively (if you have a VRML97 capable
viewer). There will be two sets of guides that position the fiber
as it passes through the diameter testing section. The guides may
also serve as wipers to clean the fiber of debris. The fiber passes
through the center of a cavity with a focused
LED impinging on the fiber and a photodiode on the opposite wall.
(Again, an interactive version can be found here
if
you have a VRML capable browser). The fiber casts a shadow on the
photodiode. Measuring the amplitude of the light signal shows changes
in the fiber diameter. An increase in the diameter causes a decrease
in the amplitude and conversely, a reduction of the diameter causes an
increase in the amplitude as is shown in this figure.
The x-axis scales are not exactly the same, the bottom one is a bit wider,
I believe. The digitized diameters in the upper plot were made by
measuring the fiber diameter at 1 cm intervals with a micrometer.
The digitized amplitudes in the bottom plot were made by pulling the fiber
through the measuring apparatus at approximately 8cm/s by winding a wire
around a power screwdriver's shaft while tensioning the fiber with my fingers.
Therefore, any slight anomalies are more likely due to the drive mechanism's
accuracy. Nonetheless, the peaks and valleys show a nice anti correlation,
and good repeatability as is shown by the second offset wave form.
The circuits that are used to drive the LED, and amplify the signals from the photodiode are shown here, or in postscript format here. The driver circuit uses a 555 timer chip to produce a 50% duty cycle RC oscillator at 1 kHz. This wave form is buffered and smoothed a bit by an op amp and a low pass filter immediately before the LED and its current limiting resistor. The detector is a classic photodiode amplifier with an additional bit of filtering and buffering. The light signal is quite large from the focused LED, so the feedback resistor is only 200k, much smaller than one would usually use. The ADC that is used to digitize the output is a simple PC sound card. I use the line in jack to digitize the wave form. The amplitude is then calculated by fitting the shape of the peak by a polynomial and finding the maximum. This is a crucial step, as simply taking the maximum digitized peak is not accurate enough.
There are two other measurements that occur in the detector box, one
is simply a counter to mark the position of the fiber as it passes through
the machine. The second detector consists of two separate parts.
The first, is a light injector, a reflective cavity with several blue LEDs
to illuminate the WLS fiber. The blue light converts to green light,
some of which is trapped in the fiber. Several centimeters away a
second reflective cavity is instrumented with a large photodiode.
The photodiode will respond to any light that leaks from the fiber due
to surface anomalies, or opaque regions of the fiber.
Picture of Machine, as realized.
Comments questions? Send them to Leon Mualem
