LED Lighting

Forums › Home Workshop DIY › LED Lighting

1 | ... | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | ... | 47

3267 posts

Uber Geek

Trusted

#644753 22-Jun-2012 13:51

I have the day off with nothing to do so out of curiosity I've been digging into Viribright product certification. Most are here (but not all): http://www.viribright.com/index.ashx/company/certification
The evidence speaks for itself, this is not a cheap bulb. They use EPISTAR (Chimei) LEDs. Going through the test reports I saw the 4.5W GU10 lamp has no electrolytic but the 1.5m tube does. The jury is still out on the normal bulbs, but knowing what the inside of the "ball" looks like I know all the electronics is down in the fitting metal cap and there is very little space for large electrolytics as well as EMC components so it is highly likely their normal bulbs do not have electrolytics either.

Richms, you might have bought a few bulbs from a lower quality assembly batch, it happens (unfortunately). Next time try a different shop or at least a different box of stock. In 6 months since we moved into our new home I've had to replaced almost 1/4 of the Philips Softone bulbs. Some lasted only a few days, like the one that failed during the final inspection. Then there was the one that smoked, literally, my wife thought the wiring was on fire and wanted me to go into the roof to check.

You can never have enough Volvos!

Niel

3267 posts

Uber Geek

Trusted

#644763 22-Jun-2012 14:07

switchlighting: LED's are normally rated to L70 which is time to 70% of initial lumen. This is normally rated at maximum led temperature which is around 135 degree to 150 degree c depending on the LED.

70% is the recommendation of the Illuminating Engineering Society regarding lamps, not what the LED semiconductor manufacturer uses. It also does not specify the ambient temperature and humidity to achieve the life, which the semiconductor manufacturer does.

An LED running at 135 to 150 degree C has already died. Power derating needs to be applied from about 60 degree C at the LED, which means typically the bulb can be run full power at up to 45 degree C ambient + 15 degree C LED temperautre rise (rough numbers). You got your temperature units mixed up, it is 135 to 150 degree F, not C, which converts to 60 degree C.

You can never have enough Volvos!

switchlighting

18 posts

Geek

#644949 22-Jun-2012 22:04

Hello Niel,

I have spent the last 9 years working with LED's. My full time job is designing and assembling LED lights.

Just a few L70 results for various leds as supplied ny the manufacturer

Sharp zenegata 15W L70 results
- Long-lasting: Still delivering 70% lumen output after 40,000 hours at 90 °C case temperature

- Cree XML L70 results (case temperatures)
Reported TM-21 L70
Lifetime
105°C 1500 mA 96.9% 0.0017 L70(7k) > 42,300 hrs

- Luxeon rebel L70 at 105 case temp
TM‐21 L70(6k) >36,000min

Case temperature is the metal led base. The die temp is up to 25 degree higher C than this (depending on led)

Die temps are usually rated from 135 - 150 degree C.

During testing I have run our light to 120 degree heatsink temp for a few hours which equates to a die temp of 145 deg C

Cheers.

Niel

3267 posts

Uber Geek

Trusted

#644993 23-Jun-2012 00:48

LM-80-08 (08=2008) is the test procedure. It requires data to be collected every 1,000h for a minimum of 6,000h. At each measurement point the LED is allowed to cool down for 1h before luminance is measures, it is not measured while the LED is at the elevated temperature. Humidity is specified less than 65%. (Note that NIWA data shows that no city in NZ has a monthly average humidity below 65%.) The LED case temperature is controlled externally while an LED current specified by the manufacturer is applied, typically this is half the rated current. Sample size is not specified. The physical changes in case package (e.g. warping or even de-laminating) that would require a retest is not specified, the manufacturer decides this.

Energy Star requires LM-80 testing by an accredited lab using 25 samples. A minimum luminance (92-94% depending on class) is required at 6,000h of testing. This is measured at a temperature and current lower than what is used in the LM-80 test.

TM-21-11 (11=2011 and is version 1.0) is the method for extrapolating LM-80 data of minimum 10 samples to give a projected number that can be used to evaluate relative performance of LEDs. It is not an absolute LED life expectancy, just like the LCI (labour cost index) is not to be used for your pay rise. It is a statistical measure if everything else is constant and equal. The procedure is prescribed by 6 US and EU manufacturers, and 2 US government labs, specifically as a competitive tool for gauging relative performance of LED elements. Of the 6,000h data, the first 1,000h data is removed and calculations are made on the remaining 5,000h of data (more data can be collected and used, but only every 1,000h for 6,000h is required). Data is normalised. The end point as well as the start point is calculated by curve fitting. This is a very young test method and issues are still being worked through. The result changes depending on the time duration used, an example from Energy Star shows an actual test where the 6,000h data projects 60,000h and the 10,000h data projects 30,000h. Projection using data from a different LED showed exactly the opposite.

The L70 point obtained from the TM-21 data is the lower of the time predicted or 6x the TM-21 test time. In other words the algorithm was simplified (or is too complex) to predict any further than 6x the test data. L70 is used because B50 (time to 50% of lamps fail) is not suitable for LEDs. This is also why you cannot directly compare the L70 result of an LED with the B50 result of e.g. a halogen, it is not the same measure. For a start L70 is on the LED element, B50 is on the finished product.

None of the above is performed on an LED with a heat sink, it is performed with a TEM (thermoelectric module) regulating the package temperature and regulated ambient air temperature. The results of the test can be used to interpolate how the LED element will perform in a system with a heat sink and other parts added, but it is not a direct indication of how long the LED will last.

Some (but not all) of the above comes from:
http://www.energystar.gov/ia/partners/prod_development/new_specs/downloads/luminaires/TM-21%20Discussion.pdf

To recap, the TM-21 statistical procedure (where L70 is obtained) is a method developed by 6 Western LED manufacturers and 2 US labs for comparing relative performance, it is not an absolute measure of life expectancy. Other companies use it because they have to.

I have worked with LEDs since 1985... but I do not know everything. I am open to correction (if backed up by evidence).

EDIT: Converted text to link

You can never have enough Volvos!

Niel

3267 posts

Uber Geek

Trusted

#644994 23-Jun-2012 01:01

switchlighting: Die temps are usually rated from 135 - 150 degree C.

First time you refer to die temperature, previously you said LED temperature. Still, this die temperature is the absolute maximum and as you know from datasheets you cannot run at rated power at this temperature. In addition to de-rating the power, you also de-rate with ambient temperature because the heat sink temperature (and thus die temperature) is relative to ambient rather than absolute. If your 120 degree C heatsink test was running in 10 degree higher ambient then your heat sink would be 10 degree higher and the die also 10 degree higher than in your test.

We should probably back off a bit, not sure if this intense one-on-one discussion is allowed on the forum. Maybe okay while it is still related to the topic, and if nothing else it explains that you cannot compare a 35,000h LED life with a 5,000h halogen life.

You can never have enough Volvos!

Niel

3267 posts

Uber Geek

Trusted

#644996 23-Jun-2012 01:06

This article is easier to read than the previous reference.
http://ledsmagazine.com/features/8/11/10

You can never have enough Volvos!

Skolink

1081 posts

Uber Geek

#645010 23-Jun-2012 08:10

Niel: We should probably back off a bit, not sure if this intense one-on-one discussion is allowed on the forum.

Well I found all the technical details quite interesting.

Quic Broadband

Move to New Zealand's best fibre broadband service (affiliate link). Free setup code: R587125ERQ6VE. Note that to use Quic Broadband you must be comfortable with configuring your own router.

switchlighting

18 posts

Geek

#645037 23-Jun-2012 11:04

Hi Niel,

Yes you are correct, that is how L70 is calculated. It is a good even way of comparing led's. We use it so far as it has been tested or not. If it has not then we do not use it. In the end it is a tool we can use as we obviously cannot do real time testing.

Another good rating to look for is the mcadams rating. It specifies the allowable variation between each led. Our lights are Mcadams 3 step which means if you have a large room with lots of lights you will hardly notice any (if any at all) colour difference between each led in the room

Here is a link to a good capacitor life calculator if anyone is interested: I am not allowed to post links yet

Anyway, go to illinoiscapacitor click on tech centre then click on life calculators.

Cheaper led lights. Why it is often not better;

A good case is compact fluorescent lights - the cost has been reduced so much that although the tube itself may last the rated hours the electronics often do not. They are built to a cost rather than to support the expected life. My experience has been dreadful with some lights lasting months rather than years (and I was using brand names like Philips). If they spent another few $ on the electronics they would probably attain what they set out to do but customers often go for the cheapest available then wonder why it did not last.Has anyone purchased the cheap Chinese tools from the large hardware stores? They use the same resources as the high quality tools but cost a fraction of the price and last a smaller fraction of the life. I have on a couple of occasions succumbed to the costs and without fail regretted it.

If we had the same reliability as the Viribright I would be looking for a new job!! Why would you build and distribute a light that was this unreliable? Sure it’s cheap but who want to go back to Bunnings every few months to change lights again and again?

Some people are even buying lights off ebay for NZ$15 or so!! No backup here or anywhere probably, and what happens if they catch fire because you can rest assured they are not compliant with anything other than low cost (and good luck with the insurance). Before we started manufacture we were looking at purchasing from China. Everything was CE which looked great. One company we asked to see the CE test reports (they had sent us the

switchlighting

18 posts

Geek

#645042 23-Jun-2012 11:08

Damn, my message was cut short....

Some people are even buying lights off ebay for NZ$15 or so!! No backup here or anywhere probably, and what happens if they catch fire because you can rest assured they are not compliant with anything other than low cost (and good luck with the insurance). Before we started manufacture we were looking at purchasing from China. Everything was CE which looked great. We asked to see the CE test reports and were told “ohh no we no comply with CE, we just make that up ourselves for looks”.

Well, thats my rant for the day.

Cheers.

robjg63

4100 posts

Uber Geek

Subscriber

#645216 23-Jun-2012 19:47

Was in Bunnings Mt Roskill today and saw the viribright LED bulbs (8w dimmable) and thought for $17 each I would buy one white and one warm white and see what they are like..

Popped the warm white into the lounge (dimmer on the circuit) with 5 other 75w incandescent bulbs.
Dont know what sort of dimmer - house around 10 years old if that means anything.
I had to bend the metal contact at the centre of the fitting a couple of mm so that the centre contact of the ES bulb would make contact. Fired it up and was pretty impressed. I thought it looked pretty close to the 75 soft tone bulbs. Tried the dimmer. The earlier comments are correct. They dim a little and quite slowly - so not a great range of dimming. I think form memory on the lowest dimming setting they were still on. The incandescent bulbs were completely off.

Anyway - tried it for a couple of minutes and then turned it off. When it got properly dark a few minutes later I turned the lights back on and the viribright was dead :-( I will take it back - but not a good start.

Tried the white one in the lamp next to the bed. Looks nice and bright! I quite like that bright white light for reading. Hope it lasts the 25000 hours they mention. So far a 50% failure rate not so good....

Nothing is impossible for the man who doesn't have to do it himself - A. H. Weiler

robjg63

4100 posts

Uber Geek

Subscriber

#645442 24-Jun-2012 15:56

Took the dud viribright LED bulb back to Bunning and they exchanged it.
This is the warm white 8w bulb.

Installed it in a different location and away it went. Had it running for around 30 minutes. Turned it off and on a couple of times - all fine.

Left it turned off and came back an hour later - dead!

I dont know if there is something seriously wrong with these things (the cool white one still works), but I cant say that I am impressed. $17 for 30 minutes of light?????

I dont think its worth the petrol taking another one back again to be honest.

Has quite put me off LED lights for now.....

Nothing is impossible for the man who doesn't have to do it himself - A. H. Weiler

Niel

3267 posts

Uber Geek

Trusted

#645506 24-Jun-2012 18:05

Sounds like a batch fault. Last year I've bought 2x heated towel rail timers, the ones that are hard wired inside the wall. First one failed after a month, the second after 2 months. It was a batch with a counterfeit capacitor (I'm an electronic engineer, fixed it myself and we had exactly the same brand/model/size/genuine capacitor in stock).

Of my eight 8W warm white Viribright lamps which are used for about 8 to 10 hours per day and over a month old, only 1 had a fault and it was 1 of the 6 LED arrays had an intermittent connection.

You can never have enough Volvos!

robjg63

4100 posts

Uber Geek

Subscriber

#645509 24-Jun-2012 18:13

I found a couple of postings on the consumer.org website:
http://www.consumer.org.nz/reports/led-lighting/comments/9183

Seems some others have had problems with them. If Bunnings has had issues you would think they would withdraw them....

Might ring them tomorrow and ask if they have had a high failure/return rate. While $17 is cheap for an LED - its not when it only works for 30m!

Nothing is impossible for the man who doesn't have to do it himself - A. H. Weiler

oxnsox

1923 posts

Uber Geek

#645517 24-Jun-2012 18:46

robjg63: Took the dud viribright LED bulb back to Bunning and they exchanged it.
This is the warm white 8w bulb.

Installed it in a different location and away it went. Had it running for around 30 minutes. Turned it off and on a couple of times - all fine.

Left it turned off and came back an hour later - dead!

I dont know if there is something seriously wrong with these things (the cool white one still works), but I cant say that I am impressed. $17 for 30 minutes of light?????

I dont think its worth the petrol taking another one back again to be honest.

Has quite put me off LED lights for now.....

What sort of installation is/was it?

Downlights fittings, for example, where lamps are vertical and well vented may run more reliably????

richms

28198 posts

Uber Geek

Trusted

Lifetime subscriber

#645537 24-Jun-2012 19:33

My 3 have been a BC batten holder in the kitchen so not subject to frequent cycles, an E27 reflector downlight in the bedroom and a E27 3 spots on a bar in the kitchen.

the ones in the hallway that get turned on and off all the time are fine so far, but the downstairs ones are 5w ones that have the flicker problem since I assume they are too cheap to properly filter the incoming mains on the cheaper $11 ones, so less to go wrong in it.

Richard rich.ms

1 | ... | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | ... | 47