The backstory is that I recently had to replace an old Cree bulb, like the ones discussed at viewtopic.php?t=22018 and viewtopic.php?t=22033 . I went to the hardware store and selected FEIT brand in their "classic glass" style. As soon as I took one out of the box, I was struck by how totally different it felt from the old Cree bulbs. The Crees were heavy, with lots of formed metal for heatsink and internal supports. But the new FEITs looked and felt exactly like old incandescents. How odd!
Of course I had to investigate, and the first tools I reached for were a plastic bag, a centerpunch, and a hammer. Breaking the glass envelope revealed the inner structure shown above.
Inside, these "classic glass" bulbs have the architecture of Edison bulb look-alikes, with long skinny yellow things that light up when you apply power. Of course the long skinny yellow things are actually a bunch of LEDs, blue-emitting chips mounted on a ceramic strip and surrounded by fluorescent material that turns some of the blue into other colors so as to end up with "white".
Removing and dissecting one of the filaments is what produced the other photos shown above. After an absurd amount of time, effort, and assorted chemicals and heat treatment, I settled on physically carving away most of the fluorescent covering with a #10 scalpel, then scrubbing off what remained by using a toothpick and xylene. That gave me a couple of clean chips, unfortunately with all their electrical connections removed because those were exquisitely thin wires embedded in the fluorescent covering. The distance between contacts on each chip is only 0.5 mm, so they're small enough to be troublesome, though not nearly as fiddly as mounting diatoms. After a couple of days' consideration it was not terribly difficult to make a jig using size 000 insect pins, blunted with 15000-grit sandpaper, that would maintain stable contacts for electrical testing.
In the bottom of the bulb, not shown here, is a simple power supply that consists of little more than a full wave bridge rectifier, one large capacitor, and a couple of current regulator chips that appear to operate purely in analog mode, no evidence of digital switching. The bulbs have quite a bit of 120 Hz ripple at full brightness, but dimmed to about half brightness essentially all the ripple goes away. I'll write about that separately in the coming days. (Now done, see HERE.)
From a photographic standpoint, these chips trigger the behavior I've seen before, in which the color as seen by the camera looks nothing like the color seen by eye. By eye, these things are bright blue; by canon R7, they're purple; by Google Pixel 6a cell phone, they're bright blue but surrounded by a halo of darkness. I think I understand the color difference; the halo of darkness still seems a bit odd. I suppose that's some sort of HDR processing in the cell phone, toning bright blue down to something that fits better with other stuff in the photo.
Here's a cell phone shot, through the eyepiece of a stereo microscope:
The single high mag image cropped for the first two images shown here was shot with Mitutoyo M Plan Apo 10X on Canon EF 100mm f/2.8L Macro IS USM plus 1.4X teleconverter, EF-RF adapter to Canon R7 camera, in-camera focus bracketing. LED drive current about 5 µA.
--Rik
