Ein lokal autokompensierender Bildsensor

This Ph.D. thesis describes the development and implementation of the image sensor
LACS ( L ocally A uto C ompensating I mage S ensor).

One of the most challenging problems in designing CMOS image sensors is the combination
of high sensitivity and wide dynamic range. Both are required for the detection
of weak optical signals that are surrounded by bright ambient light, which principally
leads to contradictory demands on the image sensor design. Various dynamic range enhancing
concepts are analysed. It can be shown, that the apparently achieved dynamic
range enhancement is largely limited by additional noise contributions.

The intelligent image sensor LACS suppresses ambient light by automatically compensating
the corresponding part of the photo current. Thus, a high sensitivity is provided
for the detection of weak signals without limiting the dynamic range of the captured
scene.

The characteristics of the LACS-pixel and the sensor’s periphery are described using
numerical simulations and analytical models. Emphasis is placed on the analysis of noise
and fixed pattern noise (FPN) as well as the deviation of the compensating current from
ideal behaviour. While photon shot noise turns out to be the dominant noise source, the
sensor’s performance is mainly determined by the FPN of the compensation current.

Sample images taken by the first autocompensating camera validate the functionality
of the locally autocompensating concept as well as the results from calculations and
simulations. The camera provides the control signals and transfers the achieved data
to a computer. The LACS-prototype was realized using a 0.35 µm-CMOS process and
consists of 32 × 32 pixels. Each pixel measures (38 µm) 2 and includes nine transistors
and two capacitances. Random access is provided as well as correlated double sampling
(CDS) to reduce fixed pattern noise.