LED lifetime claims need a reality check (MAGAZINE)

We’ve launched a lot of discussions collectively about what the LED lighting industry needs in order to make a big play for growth in the next few years. One could argue that the talk about emerging applications and integrated intelligence will make little to no difference if the foundations of the commercial offerings are unstable. And the notion of challenges in the solid-state lighting (SSL) market is less about external forces and more about the industry shaping up and getting out of its own way.

If all of your friends jumped off a cliff, would you jump, too? So goes a classic quip of American parental wisdom, asking if doing something wrong is worthwhile just because many others are doing it. Ignoring wisecracks about the appeal of hang gliding and BASE jumping and such, this question invites kids to realize that self-preservation begins first with a sense of self, a sense of control over one’s destiny, and a sense of empowerment to make decisions for one’s own good. Though all the popular kids may be sprinting toward the cliff’s edge, one has a choice to make: Run with them, or slow down now before you’ve gone too far.

If you’ll indulge me, we need to have a sit-down conversation with our collective teenager, the LED lighting industry, about the topic of peer pressure. In our still-nascent SSL industry, many companies are trying to do the right thing, making responsible LED lifetime claims fully informed by industry standards, while others are clearly attempting to profit from the uninformed and underinformed. Many more succumb to peer pressure and play the middle ground, playing the same games with lifetime claims while being somewhat transparent about it. A whole lot of manufacturers are doing what their competitor does while knowing full well that it’s wrong. They take the same shortcuts seemingly for short-term gain — no matter the potential for long-term pain resulting from today’s bad choices.

As a reader of LEDs Magazine, chances are you’re familiar with LM-80 and TM-21, standards published by the Illuminating Engineering Society (IES). For the uninitiated, whereas lifetime for pre-SSL light sources was informed by burning a population of lamps until half of them failed, with LEDs we instead rely on luminous flux maintenance projections to inform lifetime claims. Test laboratories measure luminous flux maintenance (erroneously lumen maintenance; the lumen needs no maintaining) of LED packages/arrays/modules for X thousands of hours (i.e., LM-80) and use a standardized calculation (i.e., TM-21) to project into the future, estimating the number of hours until luminous flux will depreciate to a selected percentage of initial light output. For example, L₇₀ denotes the projected number of operating hours until LEDs are expected to reach 70% of initial output. Lamp and luminaire manufacturers use LM-80 test report data and in-situ operating temperature measurements of the LEDs to run TM-21 projection calculations. With TM-21, manufacturers calculate and subsequently market LED lamp and luminaire lifetime claims defined in tens of thousands of hours. In some instances, proper TM-21 calculations enable manufacturers to make legitimate claims of 100,000 hours or more to L₉₀, L₈₀, L₇₀, etc.

From the Port of New York to Guangzhou,
from corporate multinationals to
the local legacy metal benders,
and every company in between,
it seems nearly every manufacturer is
misusing this standard to some degree.

In a great many other instances, however, manufacturers have either doctored LM-80 data to skew projections to their advantage, selectively enforced TM-21’s calculation requirements, or maybe just made up a bunch of crap. It’s not hard to find the offenders. There’s no shortage of street lights marketed with lifetime claims of 150,000, 200,000, 250,000, or 300,000 hours to L₇₀. That last one — 300,000 hours to L₇₀ — translates to more than 34 years of continuous operation. At 12 hours on/12 off per day, that would be more than 68 years. Some LED manufacturers’ product lifetime claims are now in the territory of average human life expectancy for many countries of the world. How about a wall pack from a corporate multinational rated >560,000 hours to L₇₀? That’s nearly 128 years at 12 hours per day. I’m a believer in SSL technology and appreciate that with thoughtful engineering, testing, and design iteration one could conceivably achieve decades of reliable operation, but today’s SSL market has run amok with overrated LED lifetime claims that make direct or implied (“L₇₀”) reference to TM-21, the very industry standard that cautions against statistical invalidity, while conveniently ignoring the standard’s most important statistical tenet and glancing over more obvious weak points like LED driver failure.

Stick to the rules

There is one rule in IES TM-21-11 that is sacrosanct. It is absolutely crucial to the proper use of the TM-21 projection method. To violate this rule is to venture into the land of make believe where numbers are meaningless, however large and commercially advantageous. To willfully ignore this rule is tantamount to writing whatever large six-figure number one prefers on a piece of paper and calling it hours to L₇₀. And so many in the LED marketplace are breaking this rule all the time.

As per TM-21 clause 5.2.5, it is the famed 6X rule: “Luminous flux values must not be projected beyond 6 times the total test duration (in hours) of measured data.” This rule is then repeated nine times in the standard. Nine times.

In most instances, the TM-21 calculations provide three results, in which p is the percentage of initial light output maintained, and D is the LM-80 test duration total in hours divided by 1000 and rounded to the nearest integer:

Calculated Lp (Dk) — e.g., Calculated L₇₀ (10k) = 189,965 hours
Low statistical confidence and not intended for application use
Projected Lp (Dk) — e.g., Projected L₇₀ (10k) = 130,131 hours
Interpolated by temperature, low statistical confidence and not intended for application use
Reported Lp (Dk) — e.g., Reported L₇₀ (10k) > 60,000 hours
Interpolated by temperature and capped at 5.5 or 6 times the LM-80 test duration, high degree of statistical confidence, intended for application use

TM-21’s calculation procedure first produces the Calculated Lp (Dk) for the individual LM-80 case temperature(s). Calculated Lp (Dk) values are sometimes very large, always of low statistical confidence, and never intended for designing a lamp or luminaire. The Projected Lp (Dk) is then calculated by interpolating for the case temperature of the hottest LED package/array/module measured in situ within a lamp or luminaire; it is also not intended for designing products.

The backstop for the low statistical confidence in the Calculated Lp (Dk) and Projected Lp (Dk) values is implementation of the 6X cap, resulting in the Reported Lp (Dk), the only TM-21 result with a high degree of statistical confidence and intended for application use (i.e., product design). It is also the only value intended to be included in a TM-21 report. As per TM-21’s Addendum B (a free download at ies.org which replaced the defunct Addendum A), the Calculated and Projected values are not to be reported. ENERGY STAR’s TM-21 Calculator was adjusted accordingly in 2016, so those continuing to report Calculated and Projected values are using a calculator of whatever credibility devised by someone else.

This problem isn’t limited to fly-by-night upstarts trying to find advantages over bigger competitors. From the Port of New York to Guangzhou, from corporate multinationals to the local legacy metal benders, and every company in between, it seems nearly every manufacturer is misusing this standard to some degree. Where it begins may depend on the company. Are product managers not understanding which TM-21 L₇₀ value to pass to marketing? Is marketing misinformed, or willfully exaggerating with a shrug of the shoulders as the uncapped 681,000-hour L₇₀ value in the TM-21 report becomes a selling point of the company’s newest high bay? That’s nearly 78 years of continuous operation, and it’s a real claim from an American manufacturer. Geez, I hope I make it to 78 years.

Leveling the playing field

When IES published TM-21-11 in 2011, few understood that lamp and luminaire companies would need to train several departments about its correct usage. That not only product management and engineering would need to understand it, but marketing would also need to know what TM-21 values mean, what kinds of claims comply with TM-21, and importantly, which claims do not. That finance and legal — who hopefully heard what happened with Detroit’s street lights — would need to understand the importance of this and all related LED standards in the market to inform warranty provisions. And that all of this would be necessary to enable robust comparisons of LED products.

And isn’t that the point of all of this, to enable industry standards to inform the market? The millions of dollars spent annually in LM-80 testing, the reporting complexities, the administrative headaches of supporting product certification and qualification programs, the confidentiality concerns — Shouldn’t those millions spent translate to something meaningful in the market?

On staff at IES and previously during my tenure on the IES Testing Procedures Committee (TPC), I’ve been approached for years by industry colleagues around the world dismissing TM-21 while saying they had a better way to project luminous flux maintenance. I asked each to bring their ideas to the TPC, but few if any actually did. ANSI/IES TM-21-19 was just published, the first revision of this standard, and an ANSI-approved IES TM-21-19 Calculator file is also in development for publication in 2020. Both will uphold the 6X cap and will codify changes in Addendum B as well as many other improvements to the standard.

As the modern adage goes, don’t hate the player, hate the game…well, there’s no game without the players and the players are playing games. LED package and array vendors, to my knowledge none of you have LM-80 reports longer than 20,000 hours, meaning your customers’ Lxx claims should never exceed 120,000 hours, and neither should yours. Even if you have longer reports, what do such reports mean when they represent the performance of now-B-listed parts you put on test nearly 28 months ago? That’s an eon in LED product development, and everyone knows that.