Warranty Coverage in the Age of Obsolescence

Posted: November 8, 2016 in General Commentary

warrantyThere is a great deal of noise in the SSL market on the topic of warranty coverage. The usual lighting industry one year warranty has been determined as obsolete as the glass bulbs they housed, with new demand now for coverage of LED products of 5 years, 7 years, even 10 years. The lofty demand is founded on the concept that if the LEDs are indeed going to last 50,000, 70,000 or even 100,000 hours (13 to 26 years in heavy commercial use), then why not back that up with a warranty that reflects that performance claim? This is further supported by the reasoning that such a warranty can be used to justify the added expense of LED products. All seems reasonable enough, and there are certainly producers out there willing to deliver warranties with extended coverage.

The question is, are these real warranties, or just paper claims? Perhaps more important, are the demand for this type of warranty coverage even rational?

Historic Perspective

Conventional lighting products can last many decades in service. Troffers and down-lights can remain in service for decades, as they are primarily just sheet metal boxes or cans, with replaceable lamp and ballast modules. Setting aside the exaggerated claims of SSL marketers, the cost of keeping conventional products in operation in residential, through medium and small commercial applications is very low. Most of these structures are maintained by owners or existing maintenance staff, who exist whether lighting needs maintenance or not. The cost of lamps is minimal in the grand scheme of maintaining any space. Further, product failures, beyond the normal lamp/ballast life expectations are essentially zero, as the rest of the fixture has no wear components. This extends to decorative products for the most part, with the occasional exception of a broken lens or trim, as well as track lights and other architectural lighting. In this, conventional lighting is expected to last for as long as the owner wishes, with replacement generally coming under a refit, remodel, or building abandonment. Yet, all of these products have been accepted and put in service under one year warranty coverage. Even in the case of roadway and site lighting, where the cost of replacement and repair is many times that of typical interior lighting, warranties have generally been one year. The one exception to this, and this is rare, is landscape product, particularly in-grade, where water intrusion is an issue. Lamps come with virtually no warranty coverage at all, beyond initial start-up. Ballasts and sockets are all wrapped up in a one year coverage envelope, which is often separate of the product in which they are installed. Even the XL T8 lamp with its 50,000hr life, connected to electronic program start ballasts expected to last >100,000 hrs, fall under one year warranty coverage.

Using the current logic applied to SSL  products, any conventional luminaire should be covered under a 10 year warranty, easily. Fluorescent lamps with 20,000 hr service life, should easily be covered under a 3 year warranty, based on well established bathtub mortality curves, such lamps surviving their infant mortality state will last at least 60% of their rated life before they begin to experience end of life failures leading to their eventual rated 50% failure mode. XL lamps rated for 50,000+ hours should easily carry a 7 year warranty, with virtually no impact on manufacturers, since the infant mortality they now cover under a 1 year warranty would constitute the bulk of claims, as it does now. Ballasts are expected to last at least 50,000 hrs, with most modern products expected to operate for over 70,000hrs. Yet, these remain covered under 1 year warranties as well. Virtually all reasonably good quality electronic ballasts, placed inside fixtures to control temperatures, could be covered under 7 year warranties, if not 10 years, with virtually no impact on the manufacturers who produce them. Yet, their is no pressure on them to extend such coverage.

Interestingly, in all but a very small percentage of products, lamps and replacement ballasts, remain available and have been unchanged in decades. A T8 product installed in 1996 can be serviced today, 20 years later, to restore original light output, with minimal effort. Anyone working in an office or industrial structure that is 20+ years old, is likely to be working under the fixtures installed in that building when it was built. In my case, the building we lease space in is over 30 years old, and is lighted with T12 troffers that were retrofit to T8 21 years ago. The fixtures are more than 30 years old, installed under 1 year warranty coverage, that has expired over 29 years ago.

SSL Product Mortality Curves are Similar to Fluorescent in Terms of Lumen Depreciation

Over the time SSL products have been deployed, the rationale for stating service life in terms of lumen depreciation made some sense, and weighed against LEDs. While claims of 50,000hrs to 70% of full light output seemed lofty and unique, the reality is, that level of lumen loss was significantly greater than what modern fluorescent lamps produce. Most of the lofty comparisons to LED technology were made against CFL lamps, which have short lives and steeper lumen mortality curves. A typical 32W CFL lamp will enter the failure stage of life at around 8,000 hours, with lumen maintenance of around 85%. Against this, an LED producing 70% out to 50,000 hrs seems quite impressive, and it is. Yet, the question then is, why not demand a 3 year warranty on the CFL product? It seems the risk of failure there is far greater, in both lumen maintenance and product failure, inside a reasonable capitalization period, than exists with LED. Why are LED products required to deliver a 5 year warranty, or more, in a market that has accepted a 1 year warranty from a known inferior product?

Now, with LED entering the global general illumination market, it presents a similar value advance. A T5 or T8 lamp will effectively maintain 90% of their lumen output until they enter the end of life cycle, somewhere in the neighborhood of 80% of rated life, or 16,000 to 40,000 hours. Modern LED products will deliver a similar lumen mortality performance, with most products rated at 70,000hrs (based in actual TM-21 calculation from large sample/long duration LM-80 test protocol), delivering 90% of their original lumen output for at least 20,000 to 40,000 hours. If the luminaire manufacture designs their product to hold the LEDs well below their maximum Tj temperature, say in the operating range under 25C above ambient, service lives of 90% of initial lumens can be extended to 50,000+ hours. Yet, the demand is for the LED product to carry a 7 or 10 year warranty, while the competing T5 or T8 product is considered acceptable with a 1 year warranty?

The Real Service Life Issue is in the Driver

At the heart of long service life is the ballast for the fluorescent and HID markets, and drivers and/or power supplies for the SSL market. In this, there is virtually no differentiation between the two. LED drivers are essentially the same product as an electronic fluorescent ballast. The only real difference is in the power supply side, where AC is converted to DC. In this, there are some failure modes that exist that are unique to LED power supplies. However, if operated within the manufacturers temperature range, there is no reason to expect LED drivers to have any greater mortality or shorter service life than modern T5 ballasts. Yet, the same producer of the 1 year warranty covered T5 ballast, is seeing a demand to cover the LED driver with a 5, 7, or 10 year warranty to be compatible with the demand made on luminaire manufacturers for such coverage terms. This appears to be a nonsense demand.

The demand for driver coverage is likely rooted in early LED product issues, where the drivers were being integrated into products, like retrofit lamps. In these products, failure of the driver is a product failure, as there are no replaceable components inside. As I have stated many times over the years, the rush to jam LED retrofit lamps into the market was a bad idea, and this illustrates why. LED retrofit products have dominated the attention of customers, and their failures, from color performance, lumen output below claims, and short life characteristics, have stained LED technology before it was even fully deployed. The stink of retrofit products, made by questionable producers, operating LEDs at extremely high temperatures from poor thermals, coupled to bespoke cheap drivers crammed into impossibly hot little cavities, remains on LED technology today. Those numerous early failures and ugly light output has resulted in a market that does not trust LED technology, which has led to warranty demands that are irrational. Oddly enough, the very product category that created this distrust, remains immune to the extended warranties it has caused. Most retrofit lamps today are marketed with 90 day warranties with a few high quality products extended with pro-rated coverage to 3 years.

Meanwhile, legitimate products designed to operate LEDs within reasonable temperatures, operated from robust replaceable drivers, are now required to offer warranties exceeding those of fluorescent products by a factor of 5X to 10X. Most driver warranties, not purchased in a system, as single components, are covered under 3 or 5 year warranties, similar to the coverage of electronic fluorescent products. This means that, when the LEDs are operated at low temperatures to product extended service life, the driver is elevated to the failure point in the product. In this, it is unreasonable to expect a fixture manufacturer to offer warrantee coverage for a complete luminaire that exceeds the most expensive component in that fixture. Since this is generally included in warranty statements, the reality is, that a 5, 7 or 10 year fixture warranty with a driver disclaimer limited to the component used, is actually a 3 year warranty, since the likelihood of the remainder of the fixture failing before the driver is very very small – unless the product is badly designed. In which case failure will likely occur inside the first year.

5, 7 and 10 Years is a Very Long Time in the SSL Universe

Regardless of failure modes, the real issue of warranty coverage is how a claim is remedied. In the conventional product space, since the lamp and ballast are stable technologies, replacing a failed product is readily accomplished. A new fixture with a fresh lamp will perform essentially the same as a 5 year old fixture with the same specified lamp. In general, replacing 15 year old T5 fixtures can be done on a spot bases and still not present any challenges – stretching t 25 years with T8 and most CFL technologies.

The same cannot be said for LED products. As LED technology evolves, lumen output increases rapidly year over year. LEDs available today will likely no be available in 24 months, as they are EOL’d and replaced with higher performing products. Further Voltage characteristics of arrays changes, which changes the driver specification. The combination of current and driver characteristics to maintain a given light output with ever evolving LED sources, makes it virtually impossible to assume that any product made today will exist of the same specification in light output and energy consumption, for more than 24 months. It is very likely that inside a 5 year warranty cycle, the product itself will be so changed, it will no longer be a suitable replacement for itself – perhaps by a factor of 2 generations. Over 7 or 10 years, it is guaranteed that the replacement product will be significantly changed from the original, assuming its even still available at all.

If the issue of lumen depreciation is considered, on top of the advancement in performance, a replacement product 5 years from today (within a 7 year warranty) will be significantly brighter, using LEDs with noticeably different color performance, while using less energy. This is a more significant change than the T12 to T8/T5 shift presented a decade ago, which was managed with group retrofits, which is not going to be acceptable as a warranty replacement demand.

Driver availability in most products will mean that products that fail due to driver losses will be readily repaired in the same manner as fluorescent products today, for products using discrete driver components. Since replacing a driver will likely not change product appearance or performance, this is an invisible warranty remedy. This then means that the warranty coverage of a product is essentially limited to the driver warranty period and driver manufacturer remedy policy, regardless of any demand on the product manufacturer to offer more.

Modular LED components present a challenge. LED evolution will make it virtually impossible to simply replace an LED to return product to its original output. Further, there is a high likelihood that the LED may have been obsoleted, such has recently occurred with Bridgelux abandoning the BXRA array. In that case, every fixture that used that array are no longer capable of being repaired in kind, requiring new LEDs, and very likely new drivers to match the characteristics of those LEDs. In the case of the BXRA, most of the Vf characteristics were lower than replacement LEDs that deliver similar light output, which demand driver changes. This is happening all over the LED industry, with changes in module physical configurations, power requirements, color specifications, etc.. Zahga compliant or not, the industry is not stable enough, nor has the foundation modules become uniform enough to support warranty repairs that deliver identical performance to past products. The LED module business is not the lamp business, and is likely to become that uniform for another decade.

For products using proprietary integrated driver components, there is little remedy for a warranty failure over extended periods of time. That means in linear products, and many reduced envelope integrated products, any failure that occurs after 24 months of service, are going to present serious issues in matching light output and character between the remaining existing products and the new replacements. It will simply be impossible, using components available in the future, to produce an identical replacement. This means that the manufacturer will likely have no choice but to pass down any of this impact on the customer, by offering a replacement product for the failed units with new generations, regardless of the implication. It is unrealistic to expect that a fixture manufacturer would be required to replace all products in an installation as a remedy for a small percentage of failures.

Some manufacturers will maintain a limited inventory of obsolete components and products to cover warranty repairs for a period of time. However, this is only reliable for products that exhibit a small percentage of failures over a long warranty period. Unfortunately, electronics do not store well. Capacitors and other components break down over time, rendering many old stock items useless before that are put in service. Further, tax law makes it undesirable to have any significant level of old stock inventory carried over from year to year. Lastly, product manufacturers who are rapidly expanding product lines, are not going to dedicate costly shelf space to old generation products, especially those with modest failure rates. This means that by the time the products enter the failure period of their bathtub curve, the inventory will have been scrapped. Extending warranties to 5+ years pretty much guarantees this will be the case.

What is in a Warranty

Any warranty that over-reaches, is essentially nothing but paper, a marketing exercise without merit.  Lighting industry warranties have generally been pretty lame, with little real risk of exposure to the manufacturers. When ballasts or lamps fail, they are the fault of those suppliers, with many now managing those replacements independently of the fixture maker. Lighting warranties that cover “Defects of materials and workmanship” generally do not assume that driver and LED components are included, as they are covered in other text indicating coverage independent of the luminaire itself. What is really not covered, and at this moment, cannot be covered, is matching the performance of an existing product with new.  For the most part, warranties will state “of equal of greater performance”, which is great for a power drill, but is not so great in lighting. In a linear cove arrangement, having three warranty replaced fixtures generating 25% more light than the remaining 50 fixtures is not going to work. Neither is replacing 10 down-lights with warranty products producing 30% more light from a smaller diameter emitter, in a ceiling of 40 identical products using an obsolete LED no longer available.

Demanding long warranties seems like a reasonable demand. The products certainly can support this life expectation. But, when you dig into this deeper, it is readily apparent that regardless of marketing claims and promises, warranties exceeding three years in the SSL universe are unsupportable, and will be until numerous factors within SSL technology settle down and stabilize. That is not expected to happen for at least another decade, if ever.

This is Not all on the Manufacturers Heads

The glass lamp business sets a background of expectation in lumen output and optical performance that the industry has worked with, almost without thinking about it. This means that most 32W CFL down-lights with a wide beam pattern perform the same, so the lamp drives the performance expectation. This applies to all products using off-shelf lamps. Now, LED products have no such boundaries, so down-lights are available in a wide range of light outputs and even wider array of optical distributions. Manufacturers are keen to create a unique space for themselves, so this wide open landscape is filled with an infinite array of performances. These performances are under constant pressure to escalate and improve, more light, higher efficacy, less watts, etc… Lighting decision makers avail themselves of this expanding offering, to create finer tuned designs, more efficient systems, etc…

Stabilization requires restraint, that is not attractive at this moment in time. Eventually selective evolution will lead to a certain volume of like-performing products. At the same time, the technology drivers will move from performance based metrics to cost reduction, which will slow the change of performance characteristics year over year, allowing manufacturers the capacity to replace failed products with identical new products reasonably.

How can any manufacturer be expected to keep anchors in the past, while pursuing advancements into the future?

The solid-state revolution of lighting is still in a nascent state. There is a great deal more progress to come. This means that while there is certainly a desire to see customers be protected from product failures as new technologies are adopted, making these demands too early is a cart-before the horse fallacy. Warranties only offer a reasonable expectation of remedy that is fair to all parties. In this, it is fair to demand that customers understand the technology and accept that there are risks to adopting it before it has become completely stable. In this, it is fair to expect manufacturers to make every effort to deliver a reliable product, and demand their suppliers to the same for them, while accepting that there are risks, and ongoing changes that will frustrate warranty remedies into the future.

In this, it appears that 3 year warranties are about the limit of coverage at this time. Any extended periods beyond this, not matter how desirable the idea might be, creates a growing level of uncertainty to the possible remedies of any failed product, that will render the warranty essentially moot. Five year warranties are pushing the limit of credibility. Seven and ten year warranties are simply unreasonable. Think back to 2006, and ask, what product made then could be replaced today one-for-one? Now, consider the rate of change and proliferation of components and change within manufacturers delivering modern lighting products is operating at a velocity several times what has occurred in the last decade. This is not a trivial issue, it is one of balancing realistic expectations within the bounds of the market and its movement forward.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s