Issue
152 March 2003
2-D
Optical Position Sensor
by
Roger Johnson & Chris Lentz
DUOLATERAL
TYPE
The duolateral
PSD shown in Figure 1a consists of N-type silicon substrate
with two resistive layers separated by a PN junction.
The front side has an ion-implanted P-type resistive
layer with ohmic contacts on two sides. The backside
has an ion-implanted N-type resistive layer with two
contacts at opposite ends placed orthogonally to the
contacts on the front side. (On a single-axis PSD, the
electrodes are placed at opposite ends of one P-type
resistive layer.) The equivalent circuit shows how each
position signal is divided into two parts by the two
resistive layers (see Figure 1b).
|
(Click
here to enlarge)
|
Figure
1a—The duolateral 2-D PSD has a resistive layer
on both sides of a substrate that acts as a PN junction.
This type of PSD is the most accurate, has the highest
resolution, and is the most expensive—the two resistive
layers being the main reason. This type of PSD has
only four leads; biasing it is more complicated
than the other types of PSDs. b—The interelectrode
shunt resistance, RSH, affects frequency response;
usually it’s in the neighborhood of 5 to 20 kohms.
In addition, the larger the PSD’s area, the larger
the junction capacitance (CJ) and the slower the
frequency response. |
Because the
position signal is divided only into two parts, the
duolateral PSD has the highest position-detecting ability
of all the sensor types. The resistivity of the ion-implanted
layers is extremely uniform, so the photocurrent for
each electrode pair is inversely proportional to the
distance between the incident spot of light and electrodes.
But, this PSD’s complex structure also makes it the
most expensive.
TETRALATERAL
TYPE
The PSD shown
in Figure 2a has a single resistive layer and four electrodes
on the front surface of a photodiode and a fifth lead
that provides a bias. The signal photocurrent is divided
into four parts that are used to generate the position
signal; because of this, it has only half the theoretical
resolution of a duolateral type.
|
(Click
here to enlarge)
|
Figure 2a and b—The
2-D tetralateral type of PSD has a single resistive
layer on only one surface. The bias electrode is
a dedicated lead on the rear of the substrate; it
makes biasing simple. A higher reverse bias causes
a reduction in the junction capacitance and higher
frequency response, but also causes higher dark
current. This type of PSD has the worst accuracy—typically
a 3 to 6% position error near the perimeter of the
device. |
The equivalent
circuit in Figure 2b shows how the four signals interact
with each another on one surface. This PSD also has
distortion that’s greater on the perimeter. Nevertheless,
it’s less expensive, features a simple bias scheme,
smaller dark current, and faster response time than
a duolateral type. Its position formulas are different
than the duolateral type’s formulas, too.
PINCUSHION
TYPE
The PSD shown
in Figure 3a is an improved version of the basic tetralateral
type. It gets its name from the surface contacts that
have a large radius rather than straight sides. Viewed
from above, the four contacts look like a pincushion.
This subtle change greatly improves the extent of the
high-linearity region over the tetralateral type while
still retaining simple signal processing and biasing.
The position equations are the same as the ones for
the plain tetralateral type. Finally, the equivalent
circuit in Figure 3b shows how the surface electrodes
are placed at the four corners instead of the four sides.
|
(Click
here to enlarge)
|
Figure
3a and b—The 2-D pincushion type of PSD is really
a tetralateral type that uses shaped electrodes
to increase its linearity near the perimeter. It
has the same simple biasing requirements as the
tetralateral PSD, but only suffers from positioning
errors of approximately 1%. |
The PSD will
operate at a low frequency in this application. The
shunt and positioning resistance and the junction capacitance
give a definite limit on how fast they can respond to
modulated light. Generally, larger devices are slower
than smaller ones; applied bias voltage increases speed,
but does so at the expense of dark current. Typical
upper-frequency limits are well in excess of 20 MHz.
And, if signal-integration schemes are used, PSDs can
respond to 100-ps pulses.
The position
resolution of a PSD is the minimum detectable displacement
of a spot of light on the detector’s surface; it is
dependent on detector area, light intensity, bandwidth,
and temperature. This application will use a low bandwidth,
relatively high intensity, and low noise to give good
resolution.
Position nonlinearity
is defined as the geometric position error divided by
the detector length; it is measured within 80% of the
detector length. In addition, position nonlinearity
is typically better than 0.05% for a single-axis PSD,
approximately 0.3% for a duolateral type, 1% for a pincushion
PSD, and 2 to 3% for a tetralateral type of PSD.
It should
be emphasized that the PSD is not an imaging sensor.
Unlike a CCD sensor, the PSD cannot detect any structure
of the pattern of light falling on it. Instead, it senses
only where the centroid, or center of "mass,"
of the light pattern is falling on it. By design, this
is usually a laser beam or small point of light imaged
by a lens; it’s not a limitation. Alternatively, if
the PSD is flooded in light except for a small dark
spot or stripe, it can detect the position of that feature.
PSDs are easy to interface—requiring only a few op-amps
to produce signals—so it’s no wonder this sensor is
used in a variety of applications.
