Cheap cobs

McCheeze

Member

frica

Well-Known Member
Re: Cheap Cobs.
Don't expect more than ~12000hrs from them. That is the result of my tests going through several.
I have no opinion on the 50000hr claim from CREE. Has anyone here run them that long? Has anyone achieved more than 12000 hrs of running time? :-?
http://www.cree.com/~/media/Files/Cree/LED Components and Modules/XLamp/XLamp Application Notes/LM80_Results.pdf

Cree has published their LM-80 results.
The newer chips only have a test duration of 6000 hours more is impossible in a short amount of time unless they have time machines.
But it's plenty enough that to project the lifetime of the leds.

This is the CXA2530

 

heckler73

Well-Known Member

Cree has published their LM-80 results.
The newer chips only have a test duration of 6000 hours more is impossible in a short amount of time unless they have time machines.
But it's plenty enough that to project the lifetime of the leds.
Projections don't always work out, though (ask the IPCC). Fortunately, there are enough people on here using them, so within another year or so, those questions I have will be better answered, I imagine. After all, how large were their sample sets? 25?
Realstyles probably has that many over one part of his garden, by himself. So altogether, between the regular posters there must be a good couple hundred COBs burning at any one time. :lol:
 

needsomebeans

Well-Known Member
What kind of yield were you getting out of the junk ones ?? I can build a l00 watt light using generic chips for $17 if o could get .5 g/w out of it I'd be happy
It wasn't good enough to justify the time to keep them going. There was at least 4 lights burnt out of one of them after less than 1 year of use. Here is a link to the only thread that I could find where someone is having any luck with them. https://www.rollitup.org/t/economical-multi-led-chip-projects-for-growing.852256/


Good Luck
 
Last edited:

heckler73

Well-Known Member
That's true but LM-80 wouldn't be the industry standard if it wasn't reliable.
The LM-80 unfortunately doesn't say anything beyond the testing time. It is an accepted methodology for maintaining a standard of comparison built out of the LM-79 framework.
The TM-21 "analysis" is what provides the "lifetime" guesstimates via a least-squares approximation that spits out two constants (which are supposed to be reported in the LM-80). If one knows those constants, a projection can be plotted more easily. However, they can be roughly derived from the L90(t)-L70(t) values, I believe.

But is it reliable? That depends on how well the LED follows Boltzmann statistics. It also depends on what the standard deviation of those averages are, too. There is no guideline regarding that in the TM-21 (in fact, it is not considered) from what I can see. And then there are the surrounding hardware issues...
Ultimately, the only reliable test is the one you perform. The potential may be out there, but how well can one create the optimal, external conditions so that potential can be reached? I think it is more accurate and safer to say "the LEDs will last at least 6-10khrs". Until one achieves the 50khrs mark, it should not be used as part of marketing or at least, caveat emptor knowing those values are just 5.5-6x multipliers--sample size dependent--of the actual testing values.


Hell, if someone makes it to 20khrs on 1 set of lights, that will be worth celebrating on its own, from a technological standpoint.
 

McCheeze

Member
The LM-80 unfortunately doesn't say anything beyond the testing time. It is an accepted methodology for maintaining a standard of comparison built out of the LM-79 framework.
The TM-21 "analysis" is what provides the "lifetime" guesstimates via a least-squares approximation that spits out two constants (which are supposed to be reported in the LM-80). If one knows those constants, a projection can be plotted more easily. However, they can be roughly derived from the L90(t)-L70(t) values, I believe.

But is it reliable? That depends on how well the LED follows Boltzmann statistics. It also depends on what the standard deviation of those averages are, too. There is no guideline regarding that in the TM-21 (in fact, it is not considered) from what I can see. And then there are the surrounding hardware issues...
Ultimately, the only reliable test is the one you perform. The potential may be out there, but how well can one create the optimal, external conditions so that potential can be reached? I think it is more accurate and safer to say "the LEDs will last at least 6-10khrs". Until one achieves the 50khrs mark, it should not be used as part of marketing or at least, caveat emptor knowing those values are just 5.5-6x multipliers--sample size dependent--of the actual testing values.


Hell, if someone makes it to 20khrs on 1 set of lights, that will be worth celebrating on its own, from a technological standpoint.
If that's the case why do they say 50khr lifespan?
 

heckler73

Well-Known Member
If that's the case why do they say 50khr lifespan?
A simple multiplication of the test duration is allowed for under the TM-21 protocol to describe limits. But it is only considered accurate under the luminaire design used. The way we all slap lights together on here is definitely outside their testing parameters. Hence, the best way to use that data is to gather one's own and compare to the values presented, to see if there is a deviation from the projection. In essence, everyone has to do their own TM-21 tests, and I certainly would not expect the values to be similar. In fact, some designs on here may even be better than spec.

http://www.ledsmagazine.com/articles/2012/11/the-elusive-life-of-leds-how-tm-21-contributes-to-the-solution-magazine.html



The WG [Working Group, from the IES] analyzed 40 sets of data to determine model-fit uncertainty and its relationship to prediction limits. The recommended extrapolation limits are:


• For a sample size of 20 units, the maximum projection = 6× the test duration

• For a sample size of only 10 units, the maximum projection = 5.5× the test duration.
 

frica

Well-Known Member
In fact, some designs on here may even be better than spec.
I'm very certain that's the case.

Here many people use CPU coolers with their (underdriven) Cree cobs while normally they're being operated in a closed environment with a bare minimum heatsink, which is why they're being tested at 85 and 105 degrees celcius too.
 

littlejacob

Well-Known Member
Bonjour
I should use my cxb3590 @105ºc...over water boiling point...and cook on the aluminum sheet...lol!
CU
 

McCheeze

Member
So how hard would it be to get a 3590 to stay around 70f ? Electronics run best around that temp and our plants like that temp....just thinking why not treat your light the same way as the plant give it the best environment possible and it would be good to you right?
 

ttystikk

Well-Known Member
Re: Cheap Cobs.
Don't expect more than ~12000hrs from them. That is the result of my tests going through several.
I have no opinion on the 50000hr claim from CREE. Has anyone here run them that long? Has anyone achieved more than 12000 hrs of running time? :-?
They're only a year or two old, right? So no one will have had them running for longer than 10-12k hours yet!

However, I did have mine designed for the long haul; sealed under glass, overkill heat sinks, under driven by half. I'll be testing their long term performance, because I'll be keeping them that long! lol
 

McCheeze

Member
Or up u can use your cxb3590 as a heat source for a vaporizer just build a pipe off the top and you can smoke some meds while you watch your meds grow...:)
 

McCheeze

Member
They're only a year or two old, right? So no one will have had them running for longer than 10-12k hours yet!

However, I did have mine designed for the long haul; sealed under glass, overkill heat sinks, under driven by half. I'll be testing their long term performance, because I'll be keeping them that long! lol
What kind of setup do you have?
 

heckler73

Well-Known Member
They're only a year or two old, right? So no one will have had them running for longer than 10-12k hours yet!
That's essentially the grand point I was trying to raise. But I can be comfortable saying Zhong Guo Cheapos will give 12khrs anyway (+/- 2khrs) . So I presume that is a safe estimate for any DIY rig. :-?
But I was fairly sure someone like SDS or Supra have been running theirs for that long. Then again, many COBs seem to get changed like tires in F1-racing around here. Everyone is trying to sit on the bleeding edge :lol:
 
Top