I wonder how long it's going to be until fast GaN-on-Si processes are cheap enough to be able to stick 400MHz+ DC-DC converters inside addressable LEDs for super high efficiency conversion from higher supply voltages. the embedded inductance can be just a few nH at that switching frequency so it's not even that hard from a packaging perspective.
@azonenberg possibly. I seem to recall there being something different about the blue LED GaN process versus the GaN FET process that might make them incompatible, but that's a half-remembered detail from something I read (or watched?) a couple of years ago so I'm not sure.
The other thing I've wondered about for a long time is a monolithic single-wafer LED display, with various combinations of epitaxial layers and lithography to get R/G/B emitting LEDs on a single substrate (not OLED).
@azonenberg there'd be demand for that in the stage lighting space. right now the big players are using CoB LEDs with a very expensive optical train to focus and collimate the light into a point like we had with HID lamps, to avoid fringing issues and keep gobo and iris edges sharp. most use a single white LED with a subtractive filter, but some are using separate R/G/B/W CoBs with even more optical train tricks to combine and align them. costs a ton though and requires per-unit alignment.
@azonenberg having all the emitter colours distributed across the LED surface so they're pre-aligned would make things a lot easier.
right now I think the only competing technology is Ayrton's Cobra engine, which is based on a solid state RGB laser module. ludicrously bright but comes with a pricetag to match.
@djm@azonenberg not sure. the collimation system they use for CoB LEDs involves micromirror arrays (e.g. patent FR3094460A1) so they're no stranger to solving problems with fancy optical tricks. my guess would be some clever polarisation tricks but it's not something I've looked into in depth.
@gsuberland That is an excellent idea. And how big do the die have to be to make a DC-DC converter for a handful of LEDs? It can work at high defect density if the die size is small.
@trollball it would probably require a bit more die area for the feedback circuitry, but you could also save a bit of die area by deduplicating the analog parts of the internal clock gen since you could just divide the DC-DC clock down.
Add comment