Page 191 - Photodetection and Measurement - Maximizing Performance in Optical Systems

Stability and Tempco Issues

184 Chapter Eight

gence of the beam. The resolution of this sensor clearly depends on the trans-
fer slope, and this can be made steeper through use of a smaller illuminating
spot. However, if the spot diameter becomes of the order of the dead-region
width between the two photodetectors, linearity will suffer. Two- and quadrant-
cell detectors are available from Centronics, Hamamatsu and others with dead-
zones typically 10mm to 100mm wide. As the total received power can be high,
very high shot-noise limited position resolution is possible. With 100mA pho-
tocurrents and a 1-mm wide linear region the shot noise limited resolution is
about 5.7¥10 -8 of 1mm in 1Hz bandwidth, or 57pm. Even in a 10kHz band-
width required for a fast mechanical position servo mechanism this is 5.7nm
rms. We will at least lose a factor 2 because of the uncorrelated noise contri-
butions of the two photodiodes, and amplifier noise and bandwidth limitations
must be evaluated as in all our designs. Nevertheless, this is a pretty impres-
sive resolution for such a simple device. It is the basis of many readout schemes
for the atomic force microscope. There, a laser beam is reflected off the probe
cantilever to impinge on a split detector. Resolution can be increased by increas-
ing the received laser power, although local heating can be a problem.
Position accuracy of the basic scheme is not as impressive as resolution. We
have mentioned geometrical gain-changes caused by beam-diameter variations,
and we have the old problem of LED and laser output power variation. This
can be corrected by using the optical referencing options in-built to the config-
uration. As long as the beam-shape is constant, the straightforward division of
(V R - V L) by (V R + V L) will correct for LED intensity. Accuracy will be no better
than the divider used; about 1 percent for an analog divider. Digital division
accuracy will be limited by the ADC resolution used, and if both signal and ref-
erence voltages are close to full scale, most of the resolution of a 12- or 16-bit
ADC could be used. Signal reductions will of course quickly reduce the number
of effective bits of resolution. Higher resolution ADCs are available, especially
the “SD” devices which extend out to 24-bits. However, increasing resolution
reduces sample rate, and achieving such resolutions for the whole measurement
is very difficult.
An alternative approach is to use light-regulation techniques such as in
Fig. 8.12, with the summed intensity stabilized. Again, we save one ADC, and
in most cases achieve higher resolution and accuracy by avoiding divi-
sion. The same approach can be used for two-dimensional positioning using a
quadrant-cell.

8.6 Multibeam Referencing
Simple power referencing relies on the precision tracking of detection sensitiv-
ity of the two (signal and reference) photodiodes. Over the life of a product, this
may be difficult to guarantee to the required degree of precision. In principle,
even this error can be eliminated with multibeam referencing.

Downloaded from Digital Engineering Library @ McGraw-Hill (www.digitalengineeringlibrary.com)
Copyright © 2004 The McGraw-Hill Companies. All rights reserved.
Any use is subject to the Terms of Use as given at the website.