I replaced a dead 75 W halogen PAR30 bulb over the kitchen sink with a Satco S9415 LED bulb that was, at the time, advertised as “75 W equivalent”:
It’s noticeably less bright than the surviving halogen bulb, which is what you’d expect when 950 lm goes head-to-head with 1100+ lm (based on casual searching), but with a similar color temperature and beam pattern, so it’s Good Enough. I should have bought two and converted the halogens into glass sculptures.
The difference between the 22.8 year Life and the 3 Year Warranty always seems amusing. The warranty requires returning the bulb, so that’s about useless …
The Smell of Molten Projects in the Morning 
Recently we have begun replacing our existing exterior HID fixtures with LEDs. So far I have been very pleased with the results. The light output is comparable, the color spectrum is not an exact match, but fairly close and easy on the eyes. The energy used is a lot less (as I recall a 175w metal halide bulb is replaced with a LED bulb of around 30 w) and best of all–no ballast.
Even if the bulbs themselves don’t last any longer than the HID bulbs they replace, if I never have to swap out an HID ballast from the basket of a lift I will be a happy man.
As nearly as I can tell, the reliability of industrial-grade LED lamps leaves a lot to be desired. Traffic signals, street lights, (new!) car headlights / taillights / markers, and outdoor illumination in general suchlike seem to have far more than the number of failures I’d expect from the touted lifetime (50,000 hours!) and my small sample size (I don’t get out much).
But, yeah, the power saving alone might make up for everything else!
What sort of failure mode(s) do you suspect? What I’ve seen sort of suggests broken traces, which in turn makes me think “thermal cycling”. Kind of hard to avoid outdoos. Maybe using very thick boards and/or traces would help? I have no idea.
Traffic signals have one LED string go dark; your suggestion of PCB trace cracks makes more sense than a semiconductor failure.
Automotive LEDs seem to be a complete failure, although the fixture may have only three series LEDs that let a single PCB trace / driver failure kill the whole thing. Most of them are new in 2016 / 2017 model year cars: that shouldn’t be possible.
I saw a shiny new LED street light flickering, just like an old sodium lamp, which I wouldn’t have believed if I hadn’t see it with my own eyes. Gotta be an ugly electronics problem.
Consumer Reports has, err, reported (hmm) that after decades of improvement, new car quality probably hit its peak a few years ago has started downhill again. This seems to be across manufacturers, price ranges, and import/domestic status.
So, we’re once again entering the decontenting phase of the cycle, where they make cost reductions that won’t have any effect on quality. Didn’t work last time, but this time will be different, because Big Data!
In the stone ages of LEDs, a big issue was light output degradation. OTOH, my HP45 (circa 1973) had usable, though never bright LEDs until the processor electronics went wonky a few years back. We have some early LED “white” (more a lavender color) night lights that have degraded over the 8ish years we’ve used them, but the current crop of screw in bulbs has been fine.
I’ve seen some bulbs listed as OK for totally enclosed fixtures. Not sure what I have in the overhead bedroom fixtures, but they were designed for two 60W incandescents. I’d hope that 15W worth of LEDs wouldn’t overstress anything.
Had to dump my first two overhead sink lights. Number 1 was 5000K color–way too blue, and #2 was a 3000K spot. Lovely color, too tight of a beam. I ended up with a 2700K flood and replacements for the outside incandescent lights.
I think they still define the end-of-life point at 50% output reduction: a mere 3 dB down!
The “beam angle” for this one is 40°, with the other choice being 60°. I waved my hands over the sink to figure a tighter flood was better, without becoming a spot. Haven’t the foggiest what the halogen beam might have been, though.
I see the box says “Not for use in totally enclosed luminaires.” I wonder if a ceiling canister with a tight-fitting shroud surrounding the bulb allows enough air circulation to count as not totally enclosed…if so, then I may queue this up for our own over-sink light, since PAR30 bulbs are getting hard to source.
Though a few more lumens from the LED would be nice. I’ve been eyeing the current generation of chip-on-board LED “work lights”. 1100 lumens / fifty bucks including rechargeable LiIon battery and charger. The wife-acceptance factor for one of those stuck to the ceiling above the sink would be pretty low, though :-)
Maybe “totally enclosed” means a waterproof can with a glass lid? This one lives in an open-face can with vent holes on the top, so it’s probably all good. I hope, anyway.
Nah, they don’t want anything around the bulb IMO. I’ve tried 4 different brands of LED bulb – Cree, GE, Philips and forget the other – in my 2 bathroom fixtures that formerly held a pair of 100W incandescents one on either side of the medicine cabinet mirror. These have a glass flower shaped surround that the bulb sits in about 5″ diameter and are open on top (so most heat can easily escape). The bulb sits vertically with the threads down and bulb up – these burned out just about never with incandescent bulbs. They burn out reliably with LED bulbs – every brand I’ve tried so far is dead within a couple months max. A fully enclosed fixture would probably kill them in weeks.
I suspect the “lifetime” is strictly the expected life of the individual components and does not account for bad design, poor assembly and manufacturing and environmental exposure. I think on “How It’s Made” there is a video of LED traffic lights being made and all they did was twist the LED tails together. No solder, no PCB, just a mechanical fixture and twisted leads. And, of course, for even more savings, they are connected in series to limit the amount (if any) of voltage dropping or current control circuitry. Which is why when you see these things fail, it is always sets of LEDs failing.
boggle
It does minimize exposure to deadly toxic solder, though …
This may not be the best place to ask, but I’ve seen several times a traffic light with one string of say 8 to 15 LEDs flickering. There seemed to be somewhat randomness to it, but fairly periodic. I can’t figure out how one string could be doing that while all others fine… could they really have a separate solid state current controller for each string? Any thoughts or guesses?
I wonder if corrosion could cause intermittent connectivity in one string of LEDs causing them to flicker. Alternatively thermal characteristics might be causing intermittent behavior in one failing LED in one string.
I’ve had a single LED, usually a white one, flicker like mad, with nothing more complex than a dropping resistor in series, so it must be an on-chip failure. Alas, I didn’t have the Tek current probes, so I couldn’t measure the current. Next time, I’m ready!