Friday, March 02, 2018

Noise in Charge Domain Sampling Readouts

MDPI Special Issue on the 2017 International Image Sensor Workshop publishes Delft University paper "Temporal Noise Analysis of Charge-Domain Sampling Readout Circuits for CMOS Image Sensors" by Xiaoliang Ge and Albert J. P. Theuwissen.

"In order to address the trade-off between the low input-referred noise and high dynamic range, a Gm-cell-based pixel together with a charge-domain correlated-double sampling (CDS) technique has been proposed to provide a way to efficiently embed a tunable conversion gain along the read-out path. Such readout topology, however, operates in a non-stationery large-signal behavior, and the statistical properties of its temporal noise are a function of time. Conventional noise analysis methods for CMOS image sensors are based on steady-state signal models, and therefore cannot be readily applied for Gm-cell-based pixels. In this paper, we develop analysis models for both thermal noise and flicker noise in Gm-cell-based pixels by employing the time-domain linear analysis approach and the non-stationary noise analysis theory, which help to quantitatively evaluate the temporal noise characteristic of Gm-cell-based pixels. Both models were numerically computed in MATLAB using design parameters of a prototype chip, and compared with both simulation and experimental results. The good agreement between the theoretical and measurement results verifies the effectiveness of the proposed noise analysis models."


  1. So this circuit looks like the readout method inside a micro-bolometer sensor.

  2. How about the linearity of this arrangement please ?

  3. Part of this paper looks like an exact repetition of Sepke's work: Noise Analysis for Comparator-Based Circuits. Nice extension and tailoring to APS, but those kind of studies are almost always overly convoluted to be practical in real-life design. In the end it all boils down to one thing: Sigma_V = K x Vn x sqrt(t)

    K is some coefficient (circuit-dependent and linked with drift and the central limit theorem under weak dependence), Vn the wide-sense-stationary noise magnitude (e.g. thermal,1/f,kTC), and you have the squareroot of time, so sigma increases with time ie. a drifting random walk. The additional expansion by adding concrete thermal, 1/f, and other noise models just complicates the picture. Add up boundary conditions, knee points, voltage ranges and such studies really becomes a problem for a superhumans... or, perhaps a SPICE engine?


All comments are moderated to avoid spam and personal attacks.