TM-21, the right way to calculate LED lifetime

In the article LED Lifetime - how to calculate it? 
we have explained that the proper way to calculate the lifetime of an LED is with a pair of two standards LM-80 and TM-21.

 LM-80 (explained here) sets the standard of the actual testing while TM-21 specifies how to extrapolate the LM-80 results to times beyond the LM-80 test time.

Now we will briefly explain how the TM-21 extrapolation method can be used to predict LED lifetime and obtain a graph that looks like this:

From the LM-80 report, spreadsheets with a lot of detailed raw data obtained from testing:

We take the following:

• lumen maintenance data recorded at maximum 1000 hour intervals, over a period of at least 6000 hours for the three Case Temperatures (55°C, 85°C and manufacturer chosen)
• sample size where 10 samples is minimum and 20 samples is the optimum for each of the three data sets
• Test duration (hours)
• Tested drive current (mA)
• Number of failures

...and use the calculation method detailed in the TM-21 to predict the lifetime of an LED up to a maximum of 6 times the measured interval.

One can use the following calculator which has all the formulas hardwired.

In conclusion, the industry has moved to set standards which lead to clear rules in the race to acquire customers with lifetime and performance claims. Now it is up to the customers too become informed and ask for supporting data when faced with a lifetime graph like this one:



For 75000 hours at L70 the manufacturer must provide: LM-80 raw test data for 12.500 hours and extrapolation based on the TM-21. Without them the lifetime claim can simply be ignored when making a purchase...

LED lifetime test: the IESNA LM-80 standard

Related articles:

LED Lifetime - how to calculate it?

The IESNA LM-80 standard is the internationally recognized way for Lumen Maintenance Testing of LED light sources, arrays and modules. In other words, it clarifies the way LED brightness depreciation over time is measured. Without this standard, a manufacturer could manipulate the testing in order to obtain values that makes his LEDs stand out from the competition even if their actual quality is sub-standard. A market full with low cost products with 50.000 claimed LED lifetime shows how dire the need is for standards.

The result of IESNA LM-80 can look like this:

 What does it show:

The Lumen Maintenance at 6000 hours, as percentage, where 100% is initial, measured value.

What does not show:

 Determination or estimation of expected life or lumen output beyond test data.

Details required by LM-80:

• drive current for the LEDs tested. The manufacturer can decide the drive current for each LED type which usually includes the typical value, the maximum value and an intermediary value from it datasheet.

• for each drive current the LEDs must be tested at three ambient temperatures (Ta). The manufacturer has the freedom to choose one, the others are 55°C and 85°C.   

How LM-80 prevents misleading test results:

• Testing must be made in a 3-rd party Internationally Recognized Testing Laboratory. The manufacturer can use its own lab only if it accredited for the specific testing by a national or international Accreditation body.
• Intermediary results must be measured and recorded, at least at each 1000 hours.
• Testing and measuring equipment is properly calibreated
• Every aspect of testing is recorded

Results of the test must include: 

• Description of Sources tested
• Sample size
• Ambient conditions (Airflow, temperature, RH etc)
• Case Test point temperature
• Electrical conditions
• Lumen maintenance data
• Observation of failures
• LED monitoring interval
• Chromaticity shift over time

Without testing according to the LM-80 standard any measured lifetime claim is unfounded and must be received with a high degree of scepticism.

All major LED producers now follow the LM-80 testing standard, among them Nichia, Cree, Philips, Osram, Bridgelux and many others.

LED Lifetime, how to calculate it?

 This article is related to the articles that try to come up with answers to the question: "Are the savings promised by LED technology always real?" 

 The fact that a series of articles is needed shows the complexity of this issue.

Before this year, the lack of standards in the LED industry made things even more complicated.
In 2013, we finally have a standard way to predict the L70 lifetime of an LED. And once the LED lifetime is known the it becomes easier to calculate the return of investment for a LED installation.

 I remind what L70 lifetime means:

LED lifetime for use in general lighting = the time when the light output (lumens) has reached 70% of the value from the beginning. It is called L70 lifetime and is presented in the form of a graph:

The fact that LEDs can have a lifetime which can be up to 100 times longer than the time many producers or distributors test them on their premises has raised serious concerns about the validity of LED lifetime claims of tens of thousands of hours, 50.000 hours being the one most used.

To address the issue the IESNA (Illuminating Engineering Society of North America) has created two standards, both of international reach and a must respect for any serious LED producer.

The IESNA standards are the LM-80 and the TM-21. The need to have two standards is based on the fact that a LED lifetime claim has two parts:

1. Actual testing of the LED for a variable amount of time: 500 to 10.000 hours in which the evolution of its luminous flux is observed
2. Extrapolation of the decrease in the luminous of the LED over time with the focus on the moment its value reaches 70% (L70).

LM-80 sets the standard of the actual testing while TM-21 specifies how to extrapolate the LM-80 results to times beyond the LM-80 test time.

Without these standards a producer or distributor could pretty much obtain a LED lifetime to suit his advertising needs, by:

- testing too few LEDs
- testing the LEDs in the best possible ambient conditions
- observing the LEDs for only a short time
- using its own (and often secret) method to extrapolate the results, "adjusting" the calculations to reach the desired L70 lifetime.

In the next posts i will briefly present the LM-80 and TM-21 standards because the market needs to be made aware about them and make use of the knowledge when purchasing LED technology. Without LM-80 and TM-21 we can only find out if the savings promised by LED technology are real via costly and time consuming trial-and-error purchases.

LED lifetime, real versus advertised

 We continue the quest to find the answer to the question "Are the savings promised by LED technology always real?"

 First we saw how LED performance in real life can be very different from the one advertised, either because of the influence of temperature on brightness or because some producers simply provide false data.

Now we will discuss about the main argument that every seller or producer of LED lamps out there uses to convince: LED lifetime.

LEDs have the potential to last a long time, tens of thousands of hours, much longer than all the other lighting technologies. Therefore, the savings that come from not replacing  the lamps or fixtures so often offset by a significant degree the initial, higher cost, of the LED lamp or fixture. In some cases, replacements cost for a lamp can be 16 times higher than the cost of the lamp itself and thus the long life LED is a very convincing argument.

I have read recently about the plan of the city of Los Angeles to replace, over the next 4 years, 140.000 of the cities street lights with LED lamps. It is, to date, the largest such street LED light replacement program in the world, with an estimated total budget of $57 million. As expected, the LED fixture lifetime was the main reason for this retrofit. It is reported that the lifetime of the LED retrofits that will be used is 10 to 12 years compared to four to six years for a typical street lamp.

LED lifetime vs LED fixture lifetime

To discuss about the lifetime of a LED fixture we first need to approach the issue of LED life.

All lighting technologies (LED, incandescent, halogen, fluorescent, HID and so on) see their brightness diminish as time passes. An incandescent bulb can loose 50% or more of its light output in 40.000 hours, as this 110 years old bulb shows but out of any number of bulbs used, in 1000 hours half of them will fail outright. Therefore, the advertised lifetime of any incandescent bulb is around 1000 hours, being equal to the time it takes for 50% of the installed bulbs to fail and not the 40.000 hours it takes too loose 50% of its brightness.

The same way is used to calculate the lifetime of any other lamp or fixture, except LED:

Lamp lifetime = the time when 50% of the tested or installed lamps fail

LED is the only lighting technology to date for which it takes longer for the lamp to fail than to loose enough light output to make the lamp unusable.

LED lifetime for use in general lighting = the time when the light output (lumens) has reached 70% of the value from the beginning. It is called L70 lifetime.

 For a high performance LED the L70 lifetime could be equal with 80.000 hours or more, as the graph above shows. 

The obvious question that appears from such a graph:  How does the LED producer know the LED has a L70 of 80.000 hours, did he test it for so long? What about this LED bulb producer who claims his bulb has lifespan of 50.000 hours:

Because it would take more than 5 years to test a L70 claim of 50.000 hours and almost 9 years to test one of  80.000 hours, no producer in the world does it. It would make no sense, as the product tested will be long outdated by the time the test results arrive. And because such a test requires specialized equipment, a lab and plenty of technical knowledge, now one else does it either.

Still the market is saturated with products that promise a L70 lifetime of 15.000 to 70.000 hours! Where do these numbers come from?

The source of the L70 lifetime data divides and creates chaos, turmoil in the LED market. Some say the market is a wild west, others a jungle because, until very recently, there was no standard way to calculate LED lifetime. L70 LED lifetime information comes from a combination of testing the LED for 100 to 10.000 hours and then extrapolating the results for X0.000 hours.  How rigorously the test is done (sample size, measurements, time) and how good is the projecting method used decides the outcome.

When there are no rules and the interest is to have a higher LED L70 lifetime, a significant phenomenon of false advertising appears.  As with LED luminous flux, the savings promised by LED technology can occur only if the producer gives true information about lifetime.

In futures posts we will briefly present the standard way in which to obtain L70 lifetime, stay tuned!