Seminar on Dithering Techniques: New Results and Applications
In non-linear systems where some form of averaging is present (inertia) or can be introduced (low-pass filter), it can sometimes be beneficial to purposefully introduce an additional noise or disturbance signal. This signal is known as dither, and the combination of a dither signal and averaging tends to make a system smoother is some sense: In terms of the mean value, a discontinuous system can become continuous, or it can narrow a sector non-linearity (making it "more linear"). This smoothing property has been known in control engineering since the 1940's and at that time it was successfully used to reduce friction (mechanical lubrication) in cannon turrets and to quench limit cycles due to non-linear valves in hydroelectric power plants. In more recent years it has been extensively been applied to achieve "super-resolution" in digital signal processing: e.g. for image processing (most notable the Hubble space telescope), distortion-free re-quantisation between data types, or high-fidelity audio recording.
In this talk I will present the main theoretical and experimental results of my research into the applications of dithering: Using dithering techniques we have been able to build the highest resolution digital-to-analogue converter known in the literature, we have shown that it is possible to control and suppress the effect of timing glitches (in e.g. digital-to-analogue converters), and we have found a new method to maximise the resolution in a novel counting Fabry-Perot laser interferometer (measuring displacement with metrological accuracy).