bulbs burning out prematurely

[peterlonz]

Store lighting

Store lighting

First I am not a licensed electrician.
My interest is that of an engineer, wishing if possible, to establish a "best practice".

It seems to me that there are divergent views on the exact choice of fluoros, incandescents, & LEDs.
My comments in no particular order:
1) Surely there is no place today for the use of incandescents in store lighting scenarios. Most incandescents are now Chinese & the quality is generally accepted as poor. This may or may not be true, but it is a generally held belief.
These devices are vibration & voltage sensitive, "extreme service" globes with reinforced filament support are available to counter vibration susceptibility. Generally higher voltage rated globes are also available.
2) Fluorescent lighting tubes remain, even today, as the best bang for buck, but exactly what life may be expected & what is the best voltage choice for a given supply is a matter that should be determined by the manufacturer. Whilst Chinese products are now available I believe US & European manufacturers products are still available & recommendations from such traditional manufacturers should be trustworthy. If your job involves maintenance of such fittings & tubes, I'd recommend making the appropriate contacts.
3) The labour costs of maintenance on fluorescents can be high, but it is probably better to do planned replacement of groups rather than periodic replacement of outages. Most definitely the change date should be recorded on the tube or fitting - knowledge is power.
4) LED's I believe should not yet be considered for other than experimental purposes in a commercial applications. They are generally of Chinese origin & utilise electronic components subject to considerable heat. Most lighting engineers commenting on the web believe the electronics are failure prone at a life well below that claimed. You might get 5000 hours but anything more is hope based.

I'd be happy to see myself corrected by anyone with fact based experience.

 

[Hv&Lv]

Her branch circuits were running right at 126 volts.

126 is borderline too high.

If it was 126.1 I would make a call to the POCO, ask them to check their regulators.
They like to run high, but with todays smart meters and the information they can provide, there is little need for such a wide or high band setting on the regs.

 

[Electric-Light]

126 is borderline too high.

If it was 126.1 I would make a call to the POCO, ask them to check their regulators.
They like to run high, but with todays smart meters and the information they can provide, there is little need for such a wide or high band setting on the regs.

He said retails stores. Do they get their 120 from PoCo or is that made on-site?
Don't many complexes get MV or 480v?

If it's a smaller scale store with 208/120 service, it might just better to let it be than push 208/230 dual rated equipment like A/C and refrigeration even closer to their lower limit.

 

[kwired]

Not they won't

Time adds up the same way. No quicker/less quicker

My point was that run time accumulates faster than some may realize. If you have a lot of lamps to maintain and you only replace them when they stop working, you may feel like they are not lasting very long, when in fact you are constantly changing different ones. By the time you get through changing the last ones the first ones are near end of life already, so you never stop changing lamps.

 

[jmellc]

First I am not a licensed electrician.
My interest is that of an engineer, wishing if possible, to establish a "best practice".

It seems to me that there are divergent views on the exact choice of fluoros, incandescents, & LEDs.
My comments in no particular order:
1) Surely there is no place today for the use of incandescents in store lighting scenarios. Most incandescents are now Chinese & the quality is generally accepted as poor. This may or may not be true, but it is a generally held belief.
These devices are vibration & voltage sensitive, "extreme service" globes with reinforced filament support are available to counter vibration susceptibility. Generally higher voltage rated globes are also available.
2) Fluorescent lighting tubes remain, even today, as the best bang for buck, but exactly what life may be expected & what is the best voltage choice for a given supply is a matter that should be determined by the manufacturer. Whilst Chinese products are now available I believe US & European manufacturers products are still available & recommendations from such traditional manufacturers should be trustworthy. If your job involves maintenance of such fittings & tubes, I'd recommend making the appropriate contacts.
3) The labour costs of maintenance on fluorescents can be high, but it is probably better to do planned replacement of groups rather than periodic replacement of outages. Most definitely the change date should be recorded on the tube or fitting - knowledge is power.
4) LED's I believe should not yet be considered for other than experimental purposes in a commercial applications. They are generally of Chinese origin & utilise electronic components subject to considerable heat. Most lighting engineers commenting on the web believe the electronics are failure prone at a life well below that claimed. You might get 5000 hours but anything more is hope based.

I'd be happy to see myself corrected by anyone with fact based experience.

I talked with a guy at a supply house who said they were getting lots of complaints on LED's not living up to their claims. He thinks problem is mfr labels a bulb for 50,000 hours. As mft tested it, it would last that long. But mfr simply plugs in, never turns off & counts hours until it dies. He says LED's get weakened with every on/off cycle & customer gets a lot less life than expected.

 

[kwired]

I talked with a guy at a supply house who said they were getting lots of complaints on LED's not living up to their claims. He thinks problem is mfr labels a bulb for 50,000 hours. As mft tested it, it would last that long. But mfr simply plugs in, never turns off & counts hours until it dies. He says LED's get weakened with every on/off cycle & customer gets a lot less life than expected.

That could be an issue, also the users have a bigger variety of surges, over/undervoltage conditions that they may not even be aware of. Testing/rating likely assumes stable and correct voltage. Field conditions also will effect performance, like dissipation of heat. If heat is not properly dissipated that will be a killer on lifespan. Sure there is not as much heat as an incandescent lamp puts out, but that don't mean LED's were ever designed for that kind of heat in the first place, they still create their own level of heat and are only designed to dissipate that heat, putting them in enclosed space may not allow enough heat dissipation if they were not designed for enclosed space.

Has any LED manufacturers actually tested the expected life or are they all still estimated based on what testing was done?

 

[Electric-Light]

That could be an issue, also the users have a bigger variety of surges, over/undervoltage conditions that they may not even be aware of. Testing/rating likely assumes stable and correct voltage. Field conditions also will effect performance, like dissipation of heat. If heat is not properly dissipated that will be a killer on lifespan. Sure there is not as much heat as an incandescent lamp puts out, but that don't mean LED's were ever designed for that kind of heat in the first place, they still create their own level of heat and are only designed to dissipate that heat, putting them in enclosed space may not allow enough heat dissipation if they were not designed for enclosed space.

Has any LED manufacturers actually tested the expected life or are they all still estimated based on what testing was done?

Yes, LM-79/LM-80. Google Philips L Prize. You'll see very extensive test on that thing. However, lifetime on all LED products are based on extrapolation. That is, they test for certain duration or may use accelerated degradation at elevated temperatures and use mathematical models to predict life at ordinary conditions. Common sense will tell you this. What did we have LED lighting wise in 2004? What did we have in 2010?

It takes 5.7 years to do real 50,000 hour test.... and virtually no current LED lighting product have remained the same for 6 years.

 

[kwired]

Yes, LM-79/LM-80. Google Philips L Prize. You'll see very extensive test on that thing. However, lifetime on all LED products are based on extrapolation. That is, they test for certain duration or may use accelerated degradation at elevated temperatures and use mathematical models to predict life at ordinary conditions. Common sense will tell you this. What did we have LED lighting wise in 2004? What did we have in 2010?

It takes 5.7 years to do real 50,000 hour test.... and virtually no current LED lighting product have remained the same for 6 years.

So you are saying the rated hours of most (maybe all) LED's is not based on actual conditions but instead is based on simulated conditions, predictions, or methods that do not actually run them under normal conditions until a certain percentage of them actually do fail?

 

[G._S._Ohm]

So you are saying the rated hours of most (maybe all) LED's is not based on actual conditions but instead is based on simulated conditions, predictions, or methods that do not actually run them under normal conditions until a certain percentage of them actually do fail?

The consumer is the ultimate test lab.
Speaking of extrapolation, I can't imagine a test protocol for any electrical or mechanical thing that will understate the hours you are supposed to get in the real world.
If makers posted the average lifetime plus the standard deviation, informed consumers having read almost any text on statistics and hypothesis testing would put them out of business. Possibly no societies can stand all truths (or all lies).

 

[kwired]

The consumer is the ultimate test lab.
Speaking of extrapolation, I can't imagine a test protocol for any electrical or mechanical thing that will understate the hours you are supposed to get in the real world.
If makers posted the average lifetime plus the standard deviation, informed consumers having read almost any text on statistics and hypothesis testing would put them out of business. Possibly no societies can stand all truths (or all lies).

Ok, but the conditions used to come up with the estimated hours still must be adhered to for the estimate to have much validity. If user has other conditions that were not considered in the estimate, then the estimate means nothing. This would include transient voltages and ambient temperature differences. Two things that can easily vary once the user is involved.

 

[Speshulk]

So you are saying the rated hours of most (maybe all) LED's is not based on actual conditions but instead is based on simulated conditions, predictions, or methods that do not actually run them under normal conditions until a certain percentage of them actually do fail?

I've been told that the LED itself may last the 50,000 hours, but often the driver fails well before that.

 

[kwired]

I've been told that the LED itself may last the 50,000 hours, but often the driver fails well before that.

I can believe that. For self contained units with driver and LED all in one, it doesn't matter what fails, the entire unit becomes useless.

 

[BullsnPyrs]

You Get What You pay For

You Get What You pay For

I can believe that. For self contained units with driver and LED all in one, it doesn't matter what fails, the entire unit becomes useless.

A good reason to pay more for a quality fixture with replaceable components. We are seeing the same thing with LED as we did with induction. The cheap fixtures at the lower price points do not perform as advertised or hold up the way they should. Cooper and a couple of other manufacturers are listing there new exterior fixtures at over 100,000 hours.

 

[Electric-Light]

So you are saying the rated hours of most (maybe all) LED's is not based on actual conditions but instead is based on simulated conditions, predictions, or methods that do not actually run them under normal conditions until a certain percentage of them actually do fail?

Yes and no... Certain parameters can be tested at ordinary condition and observe the variation, then extrapolate from there. I haven't read the technical documentation in a while, but I believe that it is tested to 6,000 hours or so in-lab with a fairly large sample size. This is still a 9 months commitment. My recollection is that the decay is the most significant during the first 1,000 hours or so. From there, an equation is obtained and it is used to extrapolate the predicted time to reach 70% output. These tests often show the difference in predicted time depending on temperature.

Ambient temperature of space surrounding the LEDs face affects the decay rate of encapsulating optical material. Heatsink sink temperature affects the decay rate of phosphor and die. White LEDs used today are predominantly of solid state fluorescent lamp design utilizing visible blue excited phosphor blend.

The picture in my avatar is a Philips remote phosphor LED lamp. Rather than use phosphor blend right on the chip, it uses colored LEDs,then use a phosphor embedded cover that is not in contact with chip and have a greater surface area to reduce the rate of phosphor decay and diffuse the light. Of course, this design is rather useless if directional light is desired.

The consumer is the ultimate test lab.
Speaking of extrapolation, I can't imagine a test protocol for any electrical or mechanical thing that will understate the hours you are supposed to get in the real world.
If makers posted the average lifetime plus the standard deviation, informed consumers having read almost any text on statistics and hypothesis testing would put them out of business. Possibly no societies can stand all truths (or all lies).

Accelerated or simulated testing is done with just about anything that is economically, or time unfeasible to do under real life conditions. Such requires extensive field study. One such example is motor oil life analysis. Instead of in-lab test, studies have been done on a fairly large number of taxi cabs and oil was periodically sampled from each car along with odo reading lab testing and data capturing. Even then, racking up 15,000 miles is like a one year deal while racking up 50,000 real hours on LED is impractical.

We do have plenty of field data to validate LEDs can often survive 100K+ hours before actually failing, as evidenced by those yellow, amber and green "power on" indicator lamps on various household and industrial things that usually remains on 24/7 and stay in service for over a decade. It's well understood in the industry that failure is uncommon in LEDs unless abused, so LM-79 and LM-80
focuses on decay.