e optical 
factor of 
2a of 463 
' using a 
etry, but 
  
odiode. It 
e resistor 
nitude. In 
he diode. 
, and the 
pends on 
| 
995 
49 
3. LOGARITHM OF THE PHOTOCURRENT 
The photocurrent covers a dynamic range of more than six decades. It is difficult or impossible to process this range of 
values using analog voltage amplifiers, but it is comparatively easy to work with this current directly. We built a current 
to frequency converter to get a digital signal which depends linearly on the optical power density. Another possibility is 
compression of the signal range. Figure 8 shows two circuits with logarithmic characteristic using first a MOS transistor 
applied as a diode and second a bipolar diode. We realized a line camera with 32 pixels. Each of the 32 channels 
consists of a phototransistor, an integrated MOS transistor as a logarithmic characteristic and an output amplifier. The 
sensitivity curve is plotted in figure 9. Below an incident optical power density of about 10 uW/mm° the relation 
between the optical power and the amplified output voltage is nearly logarithmic, above that point the output depends 
linearly on the power density. This curve is exactly the relation between the gate voltage and the drain current of a 
CMOS transistor. Using larger phototransistors moves the curve to lower intensities, using larger diodes as logarithmic 
device moves the curve to higher intensities. 
The relationship between the optical power and the output voltage using a pn-diode (p-source to n-well) as a 
logarithmic circuit is plotted in figure 10. The only disadvantage of a bipolar diode as logarithmic characteristic is its 
photonic sensitivity, therefore it is necessary to cover these diodes with black varnish. We covered those diodes with 
both metals of the semiconductor process, but the aluminium turned out optically transparent. The photonic sensitivity 
of the amplifier itself realised in CMOS was no problem. 
Reference traces 
  
  
       
Coded disc 
  
Cylinder lenses Beam splitter 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
i 
i 
< 
o 
9 Average Im 
> © calculation 
© 
© 
From 8 
laserdiode i Mirror 
Cylinder lenses 
To line of Eq 
phototransistors Reference traces 
Figure 12: Optical-electrical interface with 45 
Angular sensor 
Trace1<14:0> Trace Graycode Angle 
:0> : — — — «14:0» 
Traceg< 14.05 reconstruction F1 - dualcode 
[ [- Clock 
P P 
Difference SC [-——- Staius 
ISpiay Turn 
Alignment 
Figure 13: Top level diagram of the angular detection chip. This circuit has been realized as a standard cell design. 
4. MEASUREMENT OF ANGULAR POSITION WITH OPTICAL LINE SENSORS 
For the positioning of rotating machines an exact measurement of the angular position is a pre-condition. This must be 
possible without information about the history of rotation, therefore incremental sensors are not always useful. 
Solutions with interpolation principles are known [3,7,8]. We designed an angle sensor chip using a line of 49 
phototransistors combined with a graycode decoder in digital standard cell design. Figure 11 shows the diagram of the 
optical setup. This is similar to the principle of an audio compact disk. The light of a laser diode is expanded to a line by 
cylindrical lenses and transmitted to a reflective disk. The disk has a grid of channels 2/4 deep for coding a logical 0 
and a flat surface for coding a logical 1. The channels have a smaller width than the light beam. If the beam is reflected 
IAPRS, Vol. 30, Part 5W1, ISPRS Intercommission Workshop “From Pixels to Sequences”, Zurich, March 22-24 1995