Programmable High Powered LED Light
Self Reflected was a major undertaking. It can be read about in its own page. However, simply illuminating it was also non-trivial. Neither the piece nor the illumination system functions without the other, and while we hope that the illumination system isn't noticed by anyone, it was an integral part of the work.
The challenge of the illumination system is that it needed to illuminate the piece from a large range of angles in an arbitrary programmed sequence. The piece is large and so the illumination system needs to be even larger to wrap around it. We decided to do this with a sequence of 144 LEDs. In order to have each LED pointed at the center of the piece, they all need their own three-axis mounting system. However, the LEDs generate a lot of heat and so this mounting system also needs to have a low thermal resistance to structural rail, which is capable of dissipating it through convection. We also included 6 banks of colored LED spot lights, made with a similar thermal architecture, but with their own electrical requirements. Additionally, it needed to be transportable from Penn to The Franklin Institute. In short, the LED bar had structural, electrical, thermal, and logistical engineering problems that needed to be handled.
The Light Bar consists of a 24 foot extruded aluminum bar supporting 146 separate white LED modules and an additional 6 colored LED spot lights made up of six lights each. This bar was split into three 8 foot segments with inline connectors to make assembly quick. The white LEDs were powered using daisy-chained open-drain 24-channel DMX LED driver boxes. Through some luck in the electrical architecture of these drivers, we were also able to control the spot light drivers from these boxes as well by connecting the control lines of buck-boost current regulated power converter to the open-drain connectors of the DMX LED Driver box. At the beginning of a chain sits a DMX controller box with a preloaded program that Greg and I designed. This begins cycling the program at power-on. LEDs are constant current devices but our power sources were constant voltage. We simply used a current-limiting resistor inline on each white LED module. The spot lights each had their own current regulated buck-boost power converter.
The LED module mounts needed to (a) provide three axes of rotation (b) conduct the heat from the LEDs into the alumimun bar, and (c) be inexpensive. LEDs last a long time, but only if they're cool. After performing a thermal analysis, we decided to have flanges and disks water jet cut out of a particularly malluable alloy of aluminum. They were bendable with pliers. The light bar was assembled at Penn. All of the LEDs were aligned with the LED Bar inverted on the ground aimed at a bed sheet high on a wall in an auditorium.
After several stressful days, with the help of the engineering team at The Franklin, we managed to get it the ~30ft into the air.