Design automation of a Digital to Analog Converter for automotive linear LED Drivers

Automotive LED drivers have a lot of requirements: they have to be accurate, compact in area and at the same time be optimized for performance. The chip has to offer outstanding reliability and cutting edge performances while being flexible in order to appeal to as many customers as possible. Flexibility is granted through the possibility of regulating more and more current levels, accurately and dynamically. This can be reached by increasing the resolution of the DAC that implements the output current, as well as giving different configurable full scale currents. In addition to this, to be more competitive a short time-to-market and low R&D costs are required and to do so the aid of automatic design is more and more pressing. This thesis presents a design automation workflow for the Digital to Analogue matrix: starting from the specifications, the tool allows the sizing of the circuit decreasing significantly the time it would take a designer to do it. First of all, a theoretical level optimal sizing of the DAC has been designed and verified, to ensure that the algorithm correctly computes the sizing parameters, according to the desired specifications. During a second step, implementation followed a generative AI supported development workflow. The algorithm has been written in Python to automate the DAC sizing process, which now is faster, more user-friendly and goes beyond the limits of handmade design. Gm/Id methodology has been used in the code thanks to access to the look up tables of the selected components: in this way all the data were available to be used as parameters by the algorithm.