Early Compact Fluorescent Bulb Failure

Early CFL Bulb Failure
Early CFL Bulb Failure

Being that sort of bear, I tend to make notations like this. Sometimes I’m delighted the next time the inscription sees the light of day and sometimes it ticks me right off…

Much of the energy-saving advantage of CFL bulbs comes from their touted long life. I’d say a year isn’t nearly long enough to reap any benefits…

There is certainly a warranty on the bulb, if only I’d:

  • saved the empty package and
  • had the original receipt and
  • be willing to call a presumably toll-free number and
  • go through whatever hassle they impose to swap the bulb

They know none of us will get very far down that checklist…

FWIW, the box of smaller CFL bulbs on the shelf says they have a two-year warranty “in normal residential service of 3 hours per day”. I’m sure the number of starts factors into it, too.

24 thoughts on “Early Compact Fluorescent Bulb Failure

    1. Haven’t a clue, but LED lamps remain dauntingly expensive… in a few years they should be affordable and I’ll give them a whirl.

    2. I haven’t used that one, but I’m running a (possibly?) Phillips-branded one. The one I use is sort of a “diffuse spotlight” (doesn’t try to radiate hemispherically, but doesn’t make a very tight beam, either). It’s lasted over a year, starts instantly (not surprised) and will light at temperatures that thwart a CFL. I suppose LEDs do have a minimum temperature, just to get enough free electrons to be a diode, but that must be wicked cold. On the other hand, the light is awfully blue-ish, has lousy color rendition, and aesthetically it just won’t hold a candle(power?) to a 2700K CFL. Oh, and a 5W LED “bulb” “feels” more like a 20W incandescent than a 40W, so don’t believe the packaging. Personally, I’m not buying any more except for outdoor use where it’s too cold for a CFL.

      1. the light is awfully blue-ish

        Our neighbor across the street recently replaced one of the two door lights with an LED light and it’s definitely not a good color match to the incandescent bulb. On the other paw, given the cost of the things, I understand why they’re waiting to replace the other bulb…

  1. I don’t have a reference but some years ago I read that one CFL start is equivalent to 10-15 minutes of operation.

    Have you done a post-mortem on the failed specimen? I seem to recall reports of fixable electronics in the base.

    BTW — Did I meet you last year at Cabin Fever or Names? You were set up near a music(??!) contraption.

    1. one CFL start is equivalent to 10-15 minutes of operation

      That’s about what I recall, too, so I try to leave them turned on when I’ll be trotting in and out of the Basement Laboratory.

      I haven’t torn the old ones apart, because the phosphors in the tube have obviously aged. I figure that I could probably tinker with the innards a bit, but they’re obviously designed to last just slightly longer than the warranty… so something else will blow out in short order.

      You were set up near a music(??!) contraption.

      That was the never-sufficiently-to-be-damned steam calliope, directly across the aisle. I didn’t attend this year, because I didn’t want to do that winter drive again, but I would definitely set up somewhere, anywhere else. He’s a good guy, it’s a nice piece of machining & woodworking, but …

  2. I read about one fellow who took a bunch of dead CFLs, wired ’em in series, and ran ’em with a neon sign transformer. But you hit the nail on the head when you pointed out that the cheap ones (that is, pretty much all of ’em) will lose output from phosphor degradation fairly soon, even when electronics and filament breakage are removed from the equation.

    1. ran ‘em with a neon sign transformer.

      That sounds appropriate for the Burning Man Festival…

  3. Glad to see I’m not the only one who marks CFL bulbs and batteries with the date they were put in use. I now feel a little bit more normal :-)

    Most of the bulbs I’ve dissected have suffered from burned-out electrodes, that is, the filament that heats up (just as an incandescent, but only during starting) has burned through. They were still usable with a CCFL driver and they lit up fine in strong RF fields. I’ve modified a few with burned out electrodes to use as 12V CFLs, by removing the old electronics and installing the CCFL driver out of an old scanner. Worked fine.

    Only thing is, don’t forget to properly mark the bulbs as being 12Vdc. Or you’d get a lot of flash-and-bang when inadvertently screwing them into a 230Vac outlet…..

    12V CFLs were very expensive at the time and, contrary to what I had expected of consumer electronics (CCFL drivers), they still work years later to my slight amazement, after many hours of use.

    Back to the point: I’ve got an old CFL (anno 1992) that’s still working today. An old Philips one, model ‘jam jar’. Indestructible. Bought it back in the day when CFLs were pretty expensive; bought one for myself and as a gift for someone. 20 years later and it’s still going strong. Think it has to do with the fact that they didn’t try to stuff all the electronics in a tiny package (‘must not be bigger than a plain incandescent!’, the marketing-manager said), and probably the fact that the electronics itsself was built to a higher standard, not trying to squeeze the last penny of manufacturing costs out of it.

    BTW, my first job was as business engineer at Philips Lighting. Know what the most expensive component of the B.O.M. of an incandescent lamp was? The cardboard box it came in :-)

    1. a little bit more normal

      At least around here. At one Family Gathering I asked how many others dated stuff… pretty nearly all my blood relatives said yes, pretty nearly all their spouses just shook their heads in long-suffering tolerance. I think it’s genetic from my distant Pennsylvania Dutch ancestors.

      An old Philips one, model ‘jam jar’

      I had one of those, too, and it worked great for many years. Huge, butt-ugly, and grossly expensive, which probably accounts for its lack of enthusiastic acceptance.

      Of course, those big new LED bulbs with weird external yellow filters don’t seem to be leaping off the shelves, either…

      Philips LED bulb

      the most expensive component

      I would never have guessed that!

  4. Also, the dimmable CFLs aren’t ready for prime time. I installed two in our dining area early last year. After 3 warranty replacements in 5 months, I gave up and went back to standard bulbs. Maybe I’ll try again with some LED bulbs sometime.

    1. I’m biased (since we design led systems) but I have yet to see a dimmable CFL that worked. We’ve bought, tested, and disassembled probably 20 different varieties, and they all pretty much sucked either on startup time, brightness stability, dimming range, longevity, or all of the above.

      1. As soon as LEDs cross some magic economic threshold, CFLs will vanish into the trash heap of history…

        Even before then, they vanish into the trash heap all too quickly!

        1. They’re probably already across it if you’re willing to take a very long view: we’re down to a controller chip, an inductor, a diode, a couple small ceramic caps, and a couple smt resistors, so the only stuff with less than a million hour lifetime is the LED itself and possibly the controller. We’re working on closed-loop color and brightness control so they can maintain color/brightness out to 200,000 hours. It’s a matter of time. (They’ll never be incandescent-cheap, though.)

          1. closed-loop color and brightness control

            I like it already! Put a color-control knob on the side and that’s even better!

            (Of course, there goes your reliability, but …)

        2. Finally, a chance to ask someone who designs/builds LED lights to please, pretty pretty please, make sure that the noise it radiates is limited….

          Have a LED lamp I use at the workbench which generates a lot of broadband noise. Before making measurements with the oscilloscope or spectrum analyzer, I have to turn off the light, otherwise the measurements get all messed up. It’s terrible.

          Just turned on the SA and had a closer look: with 10 cm of wire at the input, the noise is 42 dB above the noisefloor with the lamp on; most of it is in the HF bands, especially concentrated at 10 MHz but also a large noise bump around 20 MHz. With the lamp off the noise is nearly completely gone (the 5dB over noisefloor that are left are caused by the laptop and its power supply).

          For the record, it’s an XQ-Lite XQ0827, a 4.2W LED bulb.

          There are rules and regulations w.r.t. EMC compatibility, but it seems manufacturers/importers don’t care. And of course, most consumers don’t care either.

          I dread the day that houses will be filled with LED bulbs with their own little noisy switchmode PSUs :-(

          Apart from that, I’m a fan of LEDs over incandescents or CFL. But the noise, the noise…..

  5. Ed – I started writing dates on my dimmable PAR20 bulbs 2 years ago as I was getting many failures. I have called Phillips several times (when I can find the receipts) and they are fantastic about sending replacement Home Depot gift cards. Of course it would be better if I didnt need to do that at all…

    1. better if I didnt need to do that

      Just like automobiles: I really don’t want to be on a first-name basis with the mechanics out in the repair bays.

      (That said, our Sienna has been quite good over the last 11 years. I haven’t yet forgiven ’em over that freezing January day when I replaced the Mass Air Flow Sensor in the Autozone parking lot, but …)

  6. Since we’ve run the comment thread down, I’ll answer here instead. Ed: we’re probably (very, very, very probably) going to see the color and brightness control done via remote control of some sort, with everything handled within the silicon. Wireless, IR, dedicated-wired, and powerline communication are all in active development and fighting it out.

    Peter: I have no good news. Those bulbs probably pass federal EMI specifications (which are expensive and miserable to pass — we spend months on each project, goading stuff through our homebrew and then accredited EMI test systems) and nobody is going to go beyond that simply because of the difficulty (and nobody will pay for it.) The weak spot of LED’s aside from the LED’s themselves was failure of capacitors, mostly the ones used for smoothing the output into the LED, and so everyone has stopped/is stopping using those. That means the noise isn’t just in the switch node anymore: it’s the whole LED unit that’s flickering at the switch frequency. We’ve yet to see if this is going to play havoc with consumers when they try to use video/still cameras in LED-illuminated areas. But reliability is the main selling point, and price the main drawback, so they drive development.
    For what it’s worth, most of the stuff *we* make has power factor correction, so at least it shouldn’t be shoving noise and distortion back into the powerlines and using the whole building as an antenna.

    1. it’s the whole LED unit that’s flickering

      I have trouble with automobile taillight flicker fusion and, although these are surely higher than that, it’ll be fun to watch ’em strobe against each other.

      Plus, it’ll drive amateur radio operators nuts: the harmonics surely go up to about dull red.

      1. People have done a lot of research on flicker and vision. One of the interesting bits is that if you’re sitting still, staring at a fixed light, you can’t see flicker faster than about 200 Hz. However, if you’re moving your eyes quickly side-to-side, and the light in question is also moving quickly in the other direction (like happens in traffic) you can see serious flicker up nearly to the megahertz range. Most of our stuff and most of the other stuff I’ve gotten to look at are in the 10-100khz range.

        1. Trivia Time!

          Eye movements come in two kinds: slow and fast. The slow ones are optokinetic tracking (follow someone’s moving finger, not your own because the feedback is internal in this case!), slow phase of nystagmus (the tracking part when you look out of a side window of a car), and vestibulo-oculomotor reflex (VOR) that inertially stabilizes your eyeball. The fast movements are saccades. Their peak velocity is roughly proportional to their amplitude (length). They make up the fast phase of the nystagmus (retrace), and, most importantly, they relocate our gaze.

          An initiation of the saccade sets a saccadic blanking flip-flop. This flip-flop is reset when there’s no more image slip on the retina, and an image sample is then taken. What we normally see from our fovea is thus a bunch of snapshots, each taken when a saccade ends. The sample will obviously happen when you’re saccading against a moving image and instantaneously the angular velocities match. Look out the window on an adjacent track when on a train, and move your gaze ahead and behind. When things are just so, you’ll see freeze-frame images of the detritus on the tracks — that’s when your saccade happens to catch up with the tracks.

          To see flicker during the quick (saccadic) eye motion, you have to reset the blanking flip-flop to acquire the image while the eye is moving. Looking at a large spinning disk with some bars on it may be plenty enough. The image sample is long enough to catch a whole bunch of blinks if you’re doing them at kHz+.

          The eyeball motion is viscous-limited by the tear film; the eye inertia could be an order of magnitude higher without any noticeable effects :)

Comments are closed.