If you think this is going to be another column on color gamut, portability and battery power, you are wrong. I am talking about the driving circuits for the LEDs used in projection systems. These modest circuits normally play second fiddle to the LEDs themselves, or third fiddle behind thermal management. Like the thermal and optical design, they can be a key to getting peak performance from your LED projection system.
Insight Media Analyst
I am writing this from the LEDs 2008 conference in San Diego run by Intertech-Pira. The focus of the conference is on LEDs for general lighting although there have been a few talks oriented toward projection technology. For example, Professor Andreas Braeuer from the Fraunhofer Institute for Applied Optics and Precision Engineering in Jena gave a talk on collimating LED light and then using micro-optics to provide uniformity. As a climax to his presentation, he demonstrated a Sypro-made pico-projector based on the DLP pico chipset. It produces about 10 lumens, which is enough for maybe a 6" image in the well lit conference room. Most of the attendees to the conference had never seen an operating pico-projector and many had never heard of the whole product category. (Obviously not DD readers.) The image was good and the demonstration unit was small, although not small enough to build into a cell phone. Still, I think it could easily compete with the 3M or Microvision products for size and image quality.
Drivers came to my attention when speakers noted that image quality is optimized for a DLP projector by switching the LED on and off in synchronous with the DLP mirrors. To get maximum performance you need to do this in 1 microsecond. To get maximum light output, you need to do this with the Luminus PT120 chipset, which draws 30A. I am not an electrical engineer, but switching 30 amps in a microsecond sounds tough to me. Look at the heavy duty connectors in the photo of the PT120 LEDs from Luminus devices and the reality of the high current needs sinks in. I have written about drivers in LDR and MDR, but they were mostly multichannel drivers optimized for backlight units or LED signage, not brute-force, single channel, high-current drivers for projection systems.
I asked Christian Hoepfner from Luminus about this. He said ideally you should have 1 microsecond but you could still get pretty good performance with 5 or 6 microsecond switching. He said Luminus had worked closely with National Semiconductor on developing drivers for the Luminus chips and he took me over to their booth and introduced me to Dan Slupik, a Principal Product Marketing Manager at National.
The three of us chatted about drivers for a while. National has developed a special driver chip for Luminus Devices, the LM3433. This chip is designed to work with common anode LED systems. This somewhat unusual configuration is needed because Luminus, for thermal reasons, bonds the anode directly to the heat sink, which is typically at ground. Slupik showed me a small board with three LM3433s on it capable of driving all three LEDs in a projector.
When I asked if it would be possible to use a single driver chip and switch the current into the three LEDs, I was told it was theoretically possible and may be used in the future for cost reasons. However, that configuration would not allow LED illumination overlap in time. For example, if the red and green LEDs were on at the same time that would be the equivalent of a yellow segment in a color wheel DLP system and having all three LEDs on at one time would be the equivalent of a white segment. According to Slupik, the board was usable for up to 20A, although Hoepfner said he had seen the board used up to 30A. Both agreed, however, that the board was more of a reference design and any serious projector maker would design their own board if they wanted to use the LM3433.
These very high current drivers for LEDs can be designed as either switching or linear circuits. According to Hoepfner linear circuits can more easily achieve the 1 microsecond switching time with 30A current but switching circuits are more efficient, typically about 90%. Hoepfner added that linear driver circuits could be up to 85% efficient, so they were not actually too far behind.
In fairness, I asked who were the major players in the driver industry. Slupik, of course, said National Semiconductor was the leader and you didn’t need to go anywhere else. Since they have worked with Luminus to develop a special IC for their LEDs, if your design includes Luminus LEDs it may even be true. He did admit however, that Maxim, Supertex, Linear Technologies and ST Micro made drivers that could be used in projection systems. Maxim and Supertex had booths at LEDs 2008 and in fact I had already talked to them both. Supertex, for example, had supplied 200,000 driver chips to power 440,000 LEDs on the Water Cube in Beijing, but that is a different story.
Expanded coverage of LEDs 2008 will be in upcoming issues of LDR and MDR.