Finally got a Cree TW series LED to play with. I waited several months for my local Home Depot to start carrying them, but apparently they’re limiting retail sales to online in quantity, or retail in California, where there’s apparently a law requiring high CRI lamps for certain applications. Since I have no intention of converting my house to LEDs until when and if they get better, I bought a single unit on eBay.
Overall, the frosting is thinner and more delicate then the non-TW version. I damaged it attacking the clamshell packaging trying to get it out. It’s thin enough so you can see the light engine inside when you hold it against a strong light. I suspected the light engine is the same one that is used in the regular version, just driven harder with a bigger heatsink for the extra wattage. The basic idea is that since LEDs produce a lot of yellow light since they’re basically blue LEDs with a yellow phosphor, you can increase the CRI by increasing the wattage and then filtering out some of the yellow, so the glass enclosure has a bluish tint.
I plan to directly compare it with an incandescent lamp and regular Cree LED. Some of the things I thought of doing:
Dimming- make a 30 second video of them being dimmed from zero to 100% and see if the LEDs can dim to 0%, and then up to 100% over the same range as an incandescent. I expect it will fail compared to an incandescent using the same light engine as a standard unit.
Filters- place a red, green, and yellow lens over them and see how much of these colors each unit emits. I tried it with the standard unit, yellow passed as expected but red and green failed. The red was very dim but pure and the green shifted towards the incorrect hue.
Photograph- print out color bars and a standard photograph and take a picture of each being illuminated.
Fruit bowl- Since Cree uses this demonstration- take a picture of a colorful fruit bowl under each light.
Off Axis light- see how much light is emitted directly downward, a typical failure of “snow cone” LED bulbs.
The A/B test- I have two identical nightstand lamps- bring my sister (who’s as fussy about light quality and hates CFLs as much as I do) “which has the funny bulb”.
Any ideas? I don’t have a lumen meter or any fancy equipment.
IMHO, the TW lamps sacrifice way too much efficiency for marginally better CRI. I guess if you have some sort of CRI hyper-sensitivity they might be nice but I don’t see myself buying one anytime soon.
For example, this light has a plot of its light intensity vs angle, which describe your item 5, though in a jargony format.
Otherwise, without lab equipment, you can do a pretty good assessment of any light with a digital camera and the RGB ‘histogram’ function available in many prosumer cameras or Photoshop. Shoot and analyze in .raw format though, so that the camera doesn’t manipulate the image with auto-white balancing or other effects.
I love gadgets, but hate non-incandescent light, so I buy LEDs as gadgets to play with (once I’m done playing with them I generally just stash them in the cupboard, or use it outdoors where the lights are on a long time and I don’ care about quality.
I also noticed they have dimmers designed for LEDs now. I don’t see what’s so hard about being able to make an LED perform as well as an incandescent on a standard dimmer (I don’t care about the lack of color shift, I just want it to dim down to 0% and use the full range of the dimmer). I’ve spent over $2000 on computer controlled incandescent light dimmers, so I won’t be equipping my house with LEDs on a mass scale until when an if they work properly.
The LED dimmers don’t fix the “threshold problem”, but they actually start out at about 10% so the LED jumps to that brightness as soon as you nudge the dimmer, and the LED does use the entire range of the LED dimmer so that problem is solved. Although if they start making computer controlled LED dimmers at $60.00 a pop (plus $10.00 for the LED), converting them would be a looooong payback period. If they ever ban halogen bulbs I think I’ll just save up a lifetime supply.
The “traffic signal” test is completed. For some reason the camera had difficulty capturing the pure colors of the light, so the different tints shouldn’t be counted. Also, there are really two variables, how much light is emitted in the red, yellow, and green spectrums, and how much light is emitted facing the sides and back (most of the light you see in a traffic signal is reflected back). I thought this would be a good test since some LED bulbs have serious problems with red light and emitting light facing down and to the sides.
The bulbs were
Clear 69 watt, 130 volt “traffic signal” bulb as used in all modern 8" incandescent signals (although that particular lens is an extremely old design, so it may have been meant for 60 watt 120 volt bulbs).
60 watt pre-ban bulb, soft white
43 watt halogen replacement.
Very early Sylvania LED bulb
The Philips “UFO” bulb that many people here like
Cree standard bulb
Cree TW bulb
Just for fun, red traffic signal LED module.
I set the camera to manual exposure, adjusted it so the 69 watt lamp showed “0” on the expsosure display, and left it at that setting. The red LED module appears white because it’s brighter than the rest and overexposed, IRL being monochrome LED it is an extremely pure red.
Overall, the Cree LEDs performed the best, with very slightly better red and greens from the TW. The regular actually put out more yellow than the TW, which makes sense since excess yellow light relative to the other colors is what they’re trying to correct with the blue globe on the TWs.
I didn’t own any CFL bulbs at the time, but I found one for $1.00 at Walmart, so I figured I could buy that to test out just for fun sometime in the future, although it’s obviously not dimmable so I can’t do that test. I’m never going to use it in my house so at that price if it lasts just an hour or to to test out I’ll be happy.
One thing I noticed- there’s two Cree integrated recessed light trims- the TW (90CRI) and FD (80CRI). The Home Depot ones sold cheaper as EcoSmart rather than Cree branded have been switched from TS to FD.
Much as I don’t like LED dimming performance I’m using the TW series outdoors on my deck- I like that they’re water and wasp proof. I’ll probably wind up using them in the basement too, they seem to produce just the right amount of usable light (halfway between a 65 watt BR type and a 75 watt Par.
If the camera has automatic white balance (and if you haven’t overridden it), it will try to adjust the image so that the brightest object in frame appears white - obviously, if you point it at a very bright, non-white source, it will skew your results.
I read that auto white balance was kind of like the color equivalent of a camera metering the entire scene to 18% gray. Basically the camera takes the color info of the entire frame and averages it out, compares it to some kind of standard “gray”, and then applies an adjustment factor across the board to bring the average to that standard.
This usually works pretty well for uniform color casts like say… incandescent light, or outdoors at various times of day. It chokes pretty hard in mixed lighting- like a room is lit with fluorescent lights, but there’s an incandescent lamp on. You end up with the lamp’s light having a pronounced orange-red cast. (or vice-versa, fluorescents balanced for incandescent light have a green cast)
Anything else would mean that something like a really bright green light as the brightest thing in a photo would make the entire photo have a pronounced red cast, and that doesn’t/hasn’t happened in my experience.
I think as with focusing, there are different schemes for white balance metering. I’ve certainly seen whole-frame colour casts caused by a bright coloured object/light in frame, but that would happen regardless of whether it’s reacting to brightest or average anyway.
I have a 75 year old light fixture that has a globe that actually hangs from the bulb on a springy wire contraption. I haven’t been able to find a CFL or LED that would work with it until I found the 60W Cree at HD this weekend. It doesn’t have the rubbery covering like the bigger bulbs and I believe the bulb size is actually a bit smaller than a filament bulb. The globe stays put and I like the color (went with the daylight version). So far, so good.