It seems that simulation is the order of the day.

Go to the LED supplier and ask them for this data.

Have a clue. How many days = 56,000 hours?

Simulation won't work.

thanks lyn,

First i need the mtbf data for all the components that i use in the assly. Driver chip,Mosfet, zener,electrolytic,zener and few resistors and inductance coil

Then i hve to estimate the mtbt of the whole assembly that would require the unit last up to 56k hours.

Some suppliers can supply me the individual component data , but if i by the components from a distributor, its quite impossible to get the data.

Calculating the MTBF (Mean Time Between Failure) of a complex circuit is problematic, but a clue would be if any component has a known MTBF less than two times your required life. Instead of simulation try Accelerated Life Testing:

http://en.wikipedia.org/wiki/Accelerated_life_testing

...more work for you, more confidence for your customer.

You can use handbook prediction values and a spreadsheet, take the failure rates for your components at the intended environment, multiply them by the quantity and a suitable quality factor and sum the results. Invert afterwards for MTBF. IMO proprietary software is a waste of money for small scale predictions, Excel for example is all you would need. This requires some surfing on your part it's not possible to summarize this in a post. A typical handbook would be Telcordia, (217 is all over the net and free) and if you use a parts count method it will give you a quick answer. You could always hire someone to help you with this.

You might also mention that his actual results may vary simply because no amount of testing can simulate random conditions that normally occur over such a long period.

Perhaps he is thinking of the kind of one of the many conditions under which other lamp types are tested. That may include a 3 hours on, 20 minutes off cycle sometimes used in other types of lighting products. Another popular "accelerated" cycle was one minute on, one minute off. At best, many of these cycles were only indicators of defective product rather than mttf.

The way the lighting companies did it was to double the price, then guarentee the product for that extended duration. Chances are the customer would lose his paperwork, move on to another job, or at worst, make a claim for which the lighting company had already been paid. Great profits in it if you can find someone to buy into it! Happy customer if he gets his second bulb "for free!" Win-Win situation!

The reliability of such an assembly (provided you have selected quality components) is more dependant on your processes than the components.

The failures that you will see will be the result of defective solder jonts or thermal profile stress in the components during the soldering process, or conduction path resistance.

Your single largest issue will be that the LED interior (post and cup) actually move during soldering, putting stress into the bond wire. These will show as infant mortality. If you observe any first pass failures of individual LEDs, then you have a problem.

I worked in a facility where we assembled around 20,000 LEDs per day.

Failures always traced back to process variation. (Thermal cycle, time exposed to heat, poor flux activity, contaminated component terminations and so on.)

Reliability data for many electronic parts, including opto, can be found in MIL-HDBK-217 (quite old)

https://assist.daps.dla.mil/quicksearch/basic_profile.cfm?ident_number=53939

For the LEDs themselves, sudden failure is not common. but over time the light output diminishes. Some people consider failure to occur when the output is down to 75% of the original, others 50%. I guess your customer must define the level.

Cheers,

Dave

Assuming the packaging of the LEDs provides a low thermal impedance path to ambient, the components that will fail first will be the electrolytic capacitors in the power supply. MBTF of the capacitors is published information. If the LEDs fail before the capacitors, there is likely a thermal problem resulting in overheated device junctions.

You will also find copious amounts of testing information in the ASTM standards for lighting. It might help to know from what country the customer is used to buying products.

High accelerated life test (HALT) and/or High accelerated stress screening (HASS)