How can an offline LED driver last 20 years?

[treez]

The Cu-Beam light is certainly the best LED light that the worldwide market has to offer, however, the spec says that the LED lasts 180,000 hours (20 years) and that the offline LED driver that ships with it lasts as long as the LED (ie, 20 years).

The Offline LED driver is said to last the 20 years because of having big heatsinks and special capacitors.

..Surely no offline LED driver can last 20 years?
I mean, in 20 years, it is likely to succumb to mains transients which will result in the MOV failing short and thereby blowing the input fuse...this happening well before 20 years?

If mains transients are not such a problem, then why does Littelfuse sell these LED lighting mains surge protectors?…
Littelfuse LED lighting surge protectors:-
https://www.littelfuse.com/~/media/...ghting_surge_protection_modules_flyer.pdf.pdf


Cu-Beam suspended LED light:-
https://www.dyson.co.uk/lighting/cu-beam-suspended-lights.aspx

Spec of Cu-Beam LED light:-
**broken link removed**

 

[schmitt trigger]

I hear your point.

Not only heat and electrical over stresses can make a circuit fail, but material fatigue during the heat/cool cycles (cracked solder joints and open via holes), environmental attack (SO2-laced humidity, so prevalent in the largest cities), connector intermittency, and simple manufacturing defects (i.e. leaving ionic contamination on the board).

Having said that, with excellent engineering, best manufacturing practices and very careful attention to all reliability aspects can produce incredible results.

To me the platinum standard for electronic circuit reliability has to be the Voyager 1 spacecraft. Despite having worked for close to 39 years in the most challenging of environments, it is still communicating with earth.

So, if cost is no object, it can be done.

 

[Easy peasy]

Our power supplies last 20 years, we have some in the field for 35 years, it is down to design (under stress), component choice and quality of manufacture ...
I imagine a lot of military gear would last for >20 years...

 

[c_mitra]

The Cu-Beam light is certainly the best LED light that the worldwide market has to offer, however, the spec says that the LED lasts 180,000 hours (20 years) and that the offline LED driver that ships with it lasts as long as the LED (ie, 20 years).

Surely they have not tested 100 of their units for 20 years and found that all of them are functioning as per specifications after 20 years, right?

Many semiconductor devices will last an amazing life if run somewhat below the specs. I believe this is also true for LEDs.

The company is confident that within 10 years you change your home decor and someone will convince you to change the lighting fixture to a more modern looking object.

The way they estimate the life time is certainly interesting reading. But I personally do not agree with the basic premise: I think that distributed lighting is best for routine work because it gives a daytime feel.

- - - Updated - - -

Our power supplies last 20 years, we have some in the field for 35 years,

Instrument failure is a first-order process- produced by random causes. Just like radioactive disintegration. A common term widely used is half-life, the time required by which half of the instruments have failed. Average life time can be computed from the half-life (excluding infant mortalities).

About 40 years back, GE used to advertise (scientific american) that one day LEDs will light up your home. Conversion of electricity to (visible) light has remained a very inefficient process so far.

But you are right when you say that conservative design, component quality and attention to details in manufacture decide the final life. Most consumer items are designed to fail within a certain time.

 

[treez]

thanks, this post specifically concerns mains connected power supplies..

..so the transients that are dealt with by the littelfuse surge arrestors in the top post dont exist in indoor (office) mains wiring?
Is their something at the service entrance to every building which quenchs these transients?

At the end of the day, if a MOV quench's X number of transients, then its a gonna....and when the MOVs a gonner, the whole power supply is a gonner, because MOV's fail short and blow the input fuse. ('crowbar' style)

So does the answer to offline power supply longevity involve using a really big MOV?

- - - Updated - - -

Lighting mains circuits are separate to the mains circuits of other indoor appliances right?, -so a luminaire would not be able to depend on some other appliance suppressing a big mains transient with its internal surge arrestor? This is why lighting is so vulnerable to mains transients?

(speaking here of say in office or factory environment)

- - - Updated - - -

The top “FAQ” of this link on isotera tells how it is mains transients which are usually responsible for the death of offline LED lights…

**broken link removed**

 

[FvM]

MTBF of power supplies is ruled by a few components, mostly affected by aging of power semiconductors and capacitors. As Easy peasy said, professional power supplies are often designed for long life, using higher quality capacitors and reducing the component temperatures. Apparently Dyson did the same with their lighting products. Achieving 20 years LED lifetime is surely the major challenge, apparently they did by using heat pipe cooling. Even if the statistical prediction has to be verified yet, the approach sounds at least plausible.

There's still the interesting question suggested by c_mitra if similar performance could be achieved with distributed lighting (multiple low power LEDs)?

 

[treez]

There's still the interesting question suggested by c_mitra if similar performance could be achieved with distributed lighting (multiple low power LEDs)?

The quote from isotera near the bottom of post #5 states how small offline LED lights have small transient suppressors in them which often leads to premature failure.

 

[FvM]

Distributed LED lighting doesn't necessarily involve distributed power supply, it could also use a few low voltage supplies.

I see that you are specifically harping about overvoltage transient endurance of LED lights. That's surely a point to consider. I didn't yet hear about a method to include overvoltage transients in lifetime calculations, presently the discussion seems to compare apples to oranges.

 

[FlapJack]

Both of the documents Treez posted say at the bottom 5 year LIMITED warranty.

So not only is the 20 years off the table but the 5 years has take backs also. This is an old trick. Make the limited warranty hard to find, long and difficult to read. In it you will see clauses such as not responsible for acts of god, in other words, "we can not warranty this, the damage is due to lightning strikes."

 

[treez]

yes, and i dont see why the Littelfuse Transient protection modules are just for "outdoor" lighting such as street lights.....i mean, a light in an office is powered by mains which comes ultimately from "outdoors"....and there may not be much distance between the "Outdoors" mains bit and the light in the office...so surely mains transients affect all lighting, and not just "outdoor" lighting such as street lights etc?
Also, in buildings, since lighting is always on its own separate mains circuit, there will be few other devices connected to the lighting circuit which have front_end MOVs which can help quench an overvoltage, therefore, lighting is especially vulnerable to mains transients i would have thought?

 

[FlapJack]

I looked at the littlefuse pdf. The surge protector is designed to pass various regulations for outdoor use, like ip66 dust tight 1st digit and water proof second digit. You can use it indoors but you are paying for the weather resistance.

https://en.wikipedia.org/wiki/IP_Code

 

[c_mitra]

I see that you are specifically harping about overvoltage transient endurance of LED lights. That's surely a point to consider. I didn't yet hear about a method to include overvoltage transients in lifetime calculations, presently the discussion seems to compare apples to oranges.

I took a second look at their site; they are using Cu heat pipes to cool the LEDs (not the power supplies; my guess). Some of the SONY laptops I have seen also use Cu heat pipes to cool the CPU. Modern power supplies are pretty efficient- >90% efficiency can be considered routine. If each light block (is that the correct way to describe?) has about 200W of LED then the LEDs dissipate 200W (well, almost) and the driver dissipates about 20W. Is it really so difficult to built-in high voltage transient protection into the driver? A big choke can do!

 

[treez]

Is it really so difficult to built-in high voltage transient protection into the driver? A big choke can do!

Thanks, but the thing about using a big choke for transient suppression is that it doesn’t protect the MOV from the transient, because the MOV is going to be upstream of the choke.
If the MOV was downstream of the choke then the MOV would end up quenching the natural switch-on ringing between the big choke and the input capacitors (x capacitors etc). –That would not be acceptable because the MOV gets worn out every time it quenches a transient.

With the MOV upstream of the big choke, as it should be, there is no protection for the MOV from the big voltage transient…and the fuse will blow when the MOV eventually gets totally degraded by the number of mains transients.

This is a major limiting factor to the lifetime of any offline switch mode power supply.

 

[smijesh]

treez

MOV's are not fast as TVS, You should use same voltage bidirectional TVS for transient protection.
See the link below
https://www.littelfuse.com/products/tvs-diodes.aspx

 

[treez]

i think the front end transient protection has to be a MOV by law.
This is because MOVs always fail short as required and thus blow the input fuse by the “crowbar” principle.

 

[smijesh]

Not a low but UL/CSA or other safety agencies describe that certain test(Transient test) will pass only when MOV connected to line input.

Hope combination of MOV and TVS are the best choice, MOV will take care comparatively big pulse and TVS will take care small and narrow pulse.

 

[c_mitra]

i think the front end transient protection has to be a MOV by law.

I doubt. These are apparently written by lawyers, not by engineers. Some AC tests were replaced by DC tests because no SMPS will pass some of the AC tests (I do not recollect what were they).

 

[Easy peasy]

Rather than MOV's etc, just design the front end with 1200V parts. This gives good immunity on 230Vac line, no mov's needed....

 

[treez]

Thanks, i trust 1200V rating will be enough, but i read this....

Lightning strikes produce transients as high as 6kv with a duration of about 100ns. Power network failures, (eg fuse breakage, switch flashover, etc) also cause many high voltage transients up to 1.2KV, duration up to 60us. Eighty percent of all power line transients are up to 1.2kv, duration 1-10us.

...this is from Raymond Mack's book "demystifying switching power supplies"

So 1200V might get breached. The 6kV one with duration of 100ns sounds a funny one, i can't see it propagating too far down the stray inductance of any mains wiring?

 

[Easy peasy]

Respectfully the above is a novice answer, as, at 1200V, all other items on the same line blow up first, protecting your equipment, also transient events inside the home above 1kV peak are very rare, which is why so much poorly designed equipment continues to survive.

If the lightning strike is close by, all bets are off, but at least yours (if 1200V) has the greatest chance of surviving...