New opportunities for modern electro-optical instruments

January 15, 2015

As stated by Philip C. D. Hobbs on the back cover of his book:

Building a modern electro-optical instrument may be the most interdisciplinary job in all of engineering. Be it a DVD player or a laboratory one-off, it involves physics, electrical engineering, optical engineering, and computer science interacting in complex ways.

The figure below describes a basic architecture shared by many electro-optical instruments. A computer is used to control the laser modulation and to process the detected signal. This architecture presents some limitations when the computer is used in a feedback loop to compensate the random fluctuations of the laser. The detected signal must be stored in memory before being processed by the CPU. This results in latencies of tens to hundreds of microseconds which limit the feedback loop bandwidth to only a few kiloHertz.

Electro-optical instrument old architecture

Field-Programmable Gate Arrays (FPGA) can be used to reduce the loop delay to hundreds of nanoseconds, allowing feedback bandwidths up to a few MegaHertz. FPGAs are also critical in photonic applications demanding real-time processing of large amounts of data. In applications such as Optical Coherence Tomography and Doppler lidar, Fourier transforms must be computed in real-time on signals sampled at hundreds of MegaHertz.

FPGAs are a great tool for low-latency, high-speed, deterministic processing tasks. However, they are not suited to perform sequential, unpredictable, arbitrarily complex processing tasks, that are much more easily handled by a CPU. When the instrument requires internet connectivity or complex interaction with the user or the environment it becomes almost mandatory to use a CPU in the design.

Modern electro-optical instrument architecture