Life of a driver

[boulder]

Just had a random thought...

I was thinking about how most LED's lifespans are rated at about 50,000 hours. Then, a thought popped into my mind....Would a driver also last a lifetime of 50,000 hours? I always see ratings for the life of the emitter and never the driver. I was wondering if anyone knows about the lifetime of drivers.

Has anyone else ever wondered this?

I just did some quick math too...50,000 hours is almost 6 years! Think about how long it would take to put 6 years of use on one flashlight!

 

[waddup]

"long would it take to put 6 years of use on one flashlight"




im guessing 6 years.

 

[boulder]

That would be A LOT of batteries to put 6 years of runtime on a light.

But does anyone know if a driver will last that long?

 

[Light Sabre]

My guess would be that one factor in the lifetime of the driver would depend on how many times the flashlight is dropped. There have been discussions here on CPF about all the factors that could shorten the life of a flashlight. Dropping the flashlight and cracking the driver PCB was one that was discussed. If a flashlight was designed for alkalines and lithiums were installed instead, then your driver's life may not be very long at all. I have 1 dead flashlight here because it couldn't handle the L91 lithiums I put in it.

 

[gcbryan]

I'm sure it's like most electronics, if it's designed well and used within its design limitations and not subjected to undue heat or shock its life is more or less limitless.

The 50,000 hour thing is a bit of a misnomer according to one poster who seems to be in the industry and know what he's talking about. He's explanation seems reasonable given my experience with manufacturer claims.

The 50,000 hour thing (according to him) isn't the expected life which is the way it's often used. It's simply when the last emitter out of a large sample would be expected to fail.

In other words some could fail after 5 minutes, most would fail somewhere before 50,000 hours and the last one would fail at 50,000 (statistically speaking).

It doesn't really matter much to most of use as I would expect most emitter to still last far longer than we would have a use for them and we would probably lose the flashlight before we hit any of those numbers.

I'm sure drivers wear out more from being dropped, being exposed to heat and humidity and from people using battteries that they aren't designed for.

 

[McAllan]

I'm sure it's like most electronics, if it's designed well and used within its design limitations and not subjected to undue heat or shock its live is more or less limitless.

Agreed but there's another view on it too. It also depends on the driver type. A simple driver without modes or simple two mode like Fenix TK20 to take an example is much more likely to to work flawlessly in 30, 50 or 100 years from now.

The reason is that those advanced multimode drivers use a microcontroller. If designed right (but few with memory is, imho sadly enough) you doesn't wear on the microcontrollers flash or eeprom. But today microcontrollers use eeprom for storing their program. The basics is the same as flash (although access and rewrite is different but that doesn't really matter here) which is a charge is stored in a cell. Since no insulation is perfect it evaporates over time. Like rubbing a balloon and attaching it to the ceiling. The typical microcontrollers used in flashlights today are only "guaranteed" for 10 to 40 years by their respective manufacturers.
Don't know if you can get some guaranteed for up to 200 years. But I can't see why since you can still get the famous 27Cxxx eproms today - many of them with a data retention advertised as 200 years. Of course the cells are larger than typically used today. But with the very small program need for a typical flashlight program storage space shouldn't be much of an issue almost no matter how small.
The phenomenon has a name - bit rot!

In the good old days which graduately ended in the late 80es to mid 90es it was common practice that microcontrollers were either programmed using fusible links - or programmed at the time of make at the factory by altering the stencil - hard coded. Both of which do not suffer from bit rot (as there's no charge). So a little care and perhaps exchange of worn out/old capacitors and electronics from that era could very probably be working almost eons from now. But newer electronics - no way.

 

[asdalton]

The 50,000 hour thing (according to him) isn't the expected life which is the way it's often used. It's simply when the last emitter out of a large sample would be expected to fail.

That's not how LED lifetimes are measured. LED "lifetime" is based upon a threshold of percentage decrease in brightness relative to when the LED was new. If an emitter is driven and heatsinked properly, this fading will occur long before the emitter actually burns out.

Also, the last failure would not be much of a useful metric for any failures having a wide distribution (i.e., anything in the real world), since the last failure time in a large sample could be absurdly long. Remember that there is an incandescent bulb somewhere that has been burning for over 100 years.

 

[atlas]

Almost all (if not all) electronic components carry a MTBF rating. MTBF is meantime between failures, which equates to the average lifetime of the device. Some will spontaneously combust, some will last much longer than the MTBF rating. Naturally, the MTBF rating assumes that the component is being operated within it's safe limits.

 

[HarryN]

Drivers in general are not nearly as reliable as the LEDs. Philips ran into this head on when they first started pushing LEDs into the commercial market. You can imagine that the drivers that sell for $ 5 and less can't be much when quality parts in volume cost more than this - each.

The way these lower cost drivers keep such low prices is to use off spec material, and not worry too much about the lifetime. In some ways it makes sense - how many flashlight have you owned that you put even 100 battery cycles on? (100 - 200 hrs). I suspect that most people, including most CPFrs, never even come close.

 

[georges80]

A LED driver is not much different than any other regulator (linear or switcher based) and there's a heck of a lot of electronics that have been running for many many years.

A lot of ucontrollers are rated to 20 year projected lifetimes for the flash memory cells.

Cheap drivers will often skimp on capacitor quality which is likely to be a likely failure point for them. Cheap drivers often have components that are being run above their rated voltage or current capabilities. Thermal paths are likely less than optimal and that also would be the case for the LED itself.

Good quality components and conservative design is not something to expect from a mass produced far east manufactured light that you purchased.

How often have folk purchased a driver or light from a far east online site and next time around the 'same' driver has changed? It's not quality that they are chasing, it is cost savings... the moment they find a cheaper driver source or cheaper component it's out with the old and in with the new.

I have drivers that have been shipping for upwards of 8 years and have gone through several revisions over the years. Revisions that were not driven by cost cutting but driven by minor changes of layout or component choice to improve quality and reliability. Those revisions are only possible from a long term history of shipping the same driver design.

Always amusing to me on CPF to see some folk happily paying a premium for a particular LED bin, spend hours modifying a light and yet shun a $20-$30 US made driver in preference for a "grab bag of $5 drivers" and then wonder why 2 out of 5 are DOA and have little or no documentation/specs.

So, when you ask what is the life of a driver you need to consider that not all drivers are designed to the meet the same goal...

cheers,
george.

 

[HarryN]

A LED driver is not much different than any other regulator (linear or switcher based) and there's a heck of a lot of electronics that have been running for many many years.

A lot of ucontrollers are rated to 20 year projected lifetimes for the flash memory cells.

Cheap drivers will often skimp on capacitor quality which is likely to be a likely failure point for them. Cheap drivers often have components that are being run above their rated voltage or current capabilities. Thermal paths are likely less than optimal and that also would be the case for the LED itself.

Good quality components and conservative design is not something to expect from a mass produced far east manufactured light that you purchased.

How often have folk purchased a driver or light from a far east online site and next time around the 'same' driver has changed? It's not quality that they are chasing, it is cost savings... the moment they find a cheaper driver source or cheaper component it's out with the old and in with the new.

I have drivers that have been shipping for upwards of 8 years and have gone through several revisions over the years. Revisions that were not driven by cost cutting but driven by minor changes of layout or component choice to improve quality and reliability. Those revisions are only possible from a long term history of shipping the same driver design.

Always amusing to me on CPF to see some folk happily paying a premium for a particular LED bin, spend hours modifying a light and yet shun a $20-$30 US made driver in preference for a "grab bag of $5 drivers" and then wonder why 2 out of 5 are DOA and have little or no documentation/specs.

So, when you ask what is the life of a driver you need to consider that not all drivers are designed to the meet the same goal...

cheers,
george.

+1 :thumbsup: