We’ve just published the results of testing a batch of LED light bulbs for nearly 14,000 hours “on” time in our test rig. Most of these are now superseded models but we thought you’d want to to know the results.
My question to the CHOICE community: should we keep on testing new batches of LED general lighting service light bulbs?
(General lighting service = common light bulb type for use in light fittings, lamps etc)
I feel that these have now become a relatively inexpensive and generally reliable commodity item and we are better off putting our efforts elsewhere. Also, because it takes a few thousand hours to usefully test LED light bulbs, some of the tested models are often discontinued by the time we publish results.
We are now looking at testing downlights and drivers instead, as we’ve often been requested to test these. It will require a new test rig to be built and we don’t really have the lab space to test both downlights and light bulbs, so most likely the old lighting test rig will be dismantled.
Did the testing regime look at turning the bulbs off after short use (being turned on), bulbs/circuitry allowed to cool, then switched on completing this cycle indefinitely?
I would imagine that such cycling, which would reflect homeowner use, may affect longevity.
We are slowly switching to leds throught the house (replacing cfls when the fail with leds). The only leds which have failed from our own use was smd leds which had about 24 smds in a gu10 downlight. The bulbs had two circuits, one circuit failed on both lights after about 15-18 months use, resulting in only half of the smds lighting when switched on.
The larger chip leds (1-4 chips per bulb) appear more reliable and yet to have one fail.
As you say Chris, by the time testing is complete, new models have often replaced the ones being tested, so I think testing of drivers/power supplies would be a more useful exercise now.
Testing for say 4-6 hours on with enough time for the elctronics to completely cool down, then cycling on again, to simulate lights that are on every evening for lengthy periods, and perhaps cycling on/off at 10 mins would be useful, to test if frequent power cycling affects their life.
@phb Our test method involves running the light bulbs in a continuous switching cycle of 165 minutes on, 15 minutes off, so they’re on for a total of 22 hours per day.
We have in the past also included a short cycle test. This involves switching them on and off rapidly so that they are on for 270 seconds and off for 30 seconds, for several hundred hours. We mainly did this back when we were testing CFLs, as they are particularly susceptible to failure under very frequent switching. We have also done this test on some LED light bulbs; the test showed that LEDs are much less susceptible so we generally don’t do this test for LEDs.
When we come to testing downlights, I’m planning to put them through both long cycle and short cycle tests.
My experience with LED lights is a bit hit and miss. Some last and last, others die unexpectedly.
It is probably a better outcome to move onto testing something else, like the downlights, next. BUT, I would also like it if you to sporadically test newer entrants to the LED lighting market, as the products evolve, modify, and change. There’s so much to chose from. Strip lights, flood lights, torches, driver vs driverless, etc. They don’t have to be exhaustive lifetime tests, just indicative testing.
@grahroll > My only concern with LEDs these days is keeping them honest on lumens…I am guessing this could be done easily and would not require large amounts of test time.
Actually that does take a fair bit of time, though of course it’s a lot less elapsed time than a lifetime test. The light bulbs are set up in the test rig and burned in for 100 hours before measurement - that’s the standard method - and luminosity measurements can take about a week to complete, assuming it’s a batch of of about 15 bulbs. I predict we’d get a certain amount of member dissatisfaction with the results if there was no lifetime testing component. But still, it’s an option to consider.
We could just test them out of the box without the burn-in, which would eliminate the need for the test rig, but the burn-in period detects any obviously faulty bulbs and makes the test results more robust.
@meltam Thanks for the suggestion - as newer forms of LED lighting become more common we’ll certainly look at testing them. But re torches and flood lights, there is a huge amount of these on the market and I think these tests would be a much bigger exercise than we want to get into at this stage. Accurate light measurement of a torch or floodlight is quite different to testing a light bulb - for example you want some measure of the distance the light reaches (the throw) and that’s a whole new test.
I have health problems with dirty electricity and LED lights are supposed to be bad for this. It would be useful for those with electrosensitivity to make appropriate measurements. http://www.electrosensitivesociety.com/
Pity no tests were conducted regarding RF interference. They all interfere to some degree. Some much more than others. RF interference causes severe problems with receiving radio stations in the HF spectrum amateur radio bands as well as AM reception.
I hope a future test will include testing for RF interference as per the above.
If the history of light bulbs is anything to go by; Once incandescent light bulbs became commodities, the major bulb manufacturers of the world banded together to limit the life of light bulbs to 1000 hours. This ensured a steady income. Those that didn’t limit the time needed to earn more from each sale and went out of business.
I think that it is important for organisations like Choice to do periodic life expectancy testing to ensure that the LED bulb manufacturers are not playing games with the consumers and introducing methods to limit the life of the bulbs. There was even talk of manufacturers putting firmware into LED bulbs to limit their life. This firmware would have an encrypted key which would be protected by the DCMA which was shoved down our throats by the US, as a part of a “free trade” deal.
I don’t doubt your RF interference problems but our usage has not been an issue. Of course if it has not been mentioned to Choice then it would probably never be discussed as an option for testing and maybe it is a bigger issue than we understand. Also of note is that in many cheaper CF and LED products satisfactory shielding of the Electronics has not been done and this certainly contributes to RFI. Buying cheap bulbs does cause issues and also they tend to fail more regularly. Some of the RFI can be substantially decreased by using Ferrite Chokes around the power lines supplying these bulbs/lights
The house is completely outfitted with LEDs and before that with Standard Fluros & Compact Fluros (CF). From our changeover from older style incandescents (including halogen), we have not noticed a difference in any reception or transmission we do. This includes our use of a Wifi shared internet, Mobile phones, AM and FM radio listening, wireless house phones, TV, and a large wireless controlled Refrig AC unit.
I am not an amateur Radio/HAM user and in this area cannot then speak to that particular issue but in my general normal household usage it has not been a problem. I also think that RF interference from a multitude of other sources plays great havoc with reception and this includes but not limited to powerlines, nearby street power transformers and sub stations, motor vehicles, toasters, vacuum cleaners, microwave ovens, electric drills and similar equipment, welders and pool pumps.