Trading off lifetime vs. cost in LED light capacitor selection

Ceramic caps have a dark side, too, and it's not just cost. At line frequencies (e.g., 120Hz), their ESR skyrockets. The reason is their piezoelectric nature. They actually mechanically flex under applied voltage. As long as the applied frequency is either d-c or well about resonance, it's not a problem, but at low frequencies, with large voltage swings, ceramics will fry. If you want to use ceramics as the front-end bulk storage, you must derate the voltage at least 2x, preferably 3x. That means 500V caps for rectified 120VAC, not 250V.
Which makes them cost even more.

Presuming the lytic cap is a BR filter, while larger caps are better overall, don't forget an important point: The larger the capacitance the larger charging current pulses fed back onto the mains, which increases stress on the BR diodes and generates harmonics (lowers power factor)

Using a higher voltage capacitor only implies a longer lifetime for the part. Voltage stress ratio is reduced and leakage (internal temperature rise) is lowered. Capacitance value should be maximized to reduce ripple current which is probably a lot larger piece of power dissipation. The elastomer seal is critical for electrolytic capacitor lifetime. The component manufacturer processes are key here. Ceramic capacitors, as mentioned herein, are probably BME electrodes which can have the same leakage failure unless the stress ratio is over 50%. Hermetic wet Tantalums could resolve the issue but, their cost is as high as the selling price of the entire assembly. Solid Tantalums could be considered but, then the stress ratio is a problem again since, at best, they are rated for 60V. The best solution for commercial applications is probably still Al electrolytics but, source control should be vigorously exercised to control quality.

If it was only as simple as "lifetime vs cost".
There are so many additional variables that are being ignored in this comparison.
- environment (operating temperature/installation considerations)
(lamp will last for 100,000 hrs and only cost $1, but only if you keep it in a fixture that cools it to 15C)
- consumer pricing expectations.
- consumer performance expectations.
- profit expectations - greed - (making it good and cheap, doesn't always equate to better pricing for consumer)
....all mentioned by others...
Not mentioned (but eluded to) ..
- Quality is NOT always related to pricing. Often related to competence of anyone involved in bringing the product to market.
aka "only as strong as weakest link".
And so .. the market has a lot of sorting out to do when competing with a technology that has been around for so long (incandescent lighting)

It isn't management demanding low prices, it's the customers. No one is going to sell LED dulbs with 2$ capacitors inside, as Ed suggested, since the final price to the consumer is unacceptable. The actual cost at the checkout includes tax, retail markup, internal distribution, import, export, transport, packaging, advertising &NRE, and factory markup, so the total BOM cost is going to be about 10% of the sticker price. Now figure out the cost of the capacitors (add in 2 cents each component assembly cost).

MLCCs are available from a large variety of vendors now, the technology has improved greatly over the past few years, and the prices continue to drop. For example, you can now obtain a TDK 3.3uF 250V 20% X7T MLCC cap. for only ~$2 each (quantities of 1,000) at DigiKey...! And it is small: only 6mm x 5mm x 5.5mm! Prices will definitely be lower through distributors. I hate to see premature product failure due to over-stressed electrolytics. LED light companies should compete (with bragging rights) for the _longest_ lasting products! I would definitely pay a few dollars more for an LED light which is guaranteed to last the longest!
hxxp://search.digikey.com/us/en/products/CKG57NX7T2E335M/445-7108-2-ND/2674910

If you want to see products failure due to bad caps, look at my albums in photobucket.com under budm, even MAC used bad caps.

OK, I can see it better now. It's a 4R7 Ohm. The Vishay PR01-PR03 series is an example of a "fusible" product.
www.vishayDOTcom/docs/28729/28729.pdf

I can't read the color code but I believe the component believed to be a fuse is actually a fusible resistor or a resistor with a "defined interruption characteristic." Thus it likely serves as both an inrush limiter and fuse.
Regarding electrolytics in harsh environments, what about the electrolytics in my outdoor SmartMeter? If you peak into the Landis-Gyr AX300 you'll find several large-value leaded electrolytics. The AX300 has a claimed service life of "20+ years." I don't believe that when it's having to go through Texas summers in direct sunlight.
As someone who has replaced thousands of failed electrolytics in audio consoles that dried out do to excessive heat, I don't think you can get that kind of life out of electrolytics in a SmartMeter or LED lamp.
FWIW I have an entire house full of LED lighting and so far no failures in one year. Knock on wood.

MLCCs are available from a large variety of vendors now, the technology has improved greatly over the past few years, and the prices continue to drop. For example, you can now obtain a TDK 3.3uF 250V 20% X7T MLCC cap. for only ~$2 each (quantities of 1,000) at DigiKey...! And it is small: only 6mm x 5mm x 5.5mm! Prices will definitely be lower through distributors. I hate to see premature product failure due to over-stressed electrolytics. LED light companies should compete (with bragging rights) for the _longest_ lasting products! I would definitely pay a few dollars more for an LED light which is guaranteed to last the longest!

I'm with Matt on this one. Contrary to the teachings of Harvard Business School, mindless pre-occupation with price will kill us all ... not to mention the hypocrisy of intentionally designing products to fail early and often to feed corporate profit (and what a "friend" China is to the world - ask any Syrian) versus our pretense of caring about polluting the world with all our cheap, non-repairable products. MBA ("mindless bottom-line analysts") are the enemy of honest, ethical business. The penny-pinching on these lamps is just one example. For that matter, why any high-valued capacitors, 'lytics or otherwise, at all in these designs? Do we really care if the output is modulated at 120 Hz? Seems like phosphors with a longer persistence would be a good way to reduce flicker if that's a concern.

The big problem I see is the design rating of these bulbs is done with the bulb in a table lamp facing up.
Most bulbs in homes today are in a ceiling mount facing down. This is Much hotter positioning and orientation. Worst case is also often seen, bulbs in recessed can fixtures. this traps the heat on the bulb raising temp significantly. LED bulbs produce little heat at the bulb but some in the power conversion. This heat must be allowed for in the component specs, as noted many times in the blog. Stop going CHEAP on the parts. I'm paying good money for the bulb and I will return it when it fails. And I will make your vendor unhappy when I do. And I will not be the only one :-|

Does every other lighting company out there want to make yet another gaping product opportunity for Apple to seize on? Apple products are not cheap, but they work and they please their customers.
LED lights today look like where the MP3 player market was before Apple marched in. Lots of mediocre products wide open to "intelligent design"

"Does saving a few pennies on the BOM of a $30 light bulb really matter?"
No, but getting you to buy a new $30 light bulb in 2 years instead of 10 years DOES matter, a LOT (to the manufacturer).

A factor to consider in any electronic manufacturing is that the management of the production line/procurement is always under pressure to reduce product costs each year and it may actually be the principal measurement of his/her performance review. Working in Japan for a while at a large company, this was not the case and our product produced generally exceeded similar ones made in the US or other parts of Asia. I attribute that to the concept of caring about the product over cost.

Margery, the component that you designate a fuse in the AC line *is* actually a resistor, from the yellow-purple-gold-gold-red marking this should be a 4.7 Ohm resistor, but it may be a *safety* resistor, which is a type that is guaranteed to fail open and thus has a secondary function as a fuse. I have seen that type very often in the power supply line of larger consumer electronics devices, where you want a failing module not to take the whole device down, but only one function - also greatly improving troubleshooting.
From its size I am guessing that it is a 1 Watt resistor, so it should blow with a current over approx 0.5Amp.

And how do we know any of the parts are not counterfeit or salvaged parts? They are going to buy the right part number from the lowest cost supplier with the best delivery schedule. There is a real good chance that that supplier will be in Shenzhen -- somebody is buying all the counterfeits they crank out and most are NOT going into military electronics. Buyer beware!

The problem is that purchasing will almost always go for the cheaper components, and many times the management simply does not care that the product will fail. Temperature rating certainly does matter a lot. I have radios built for WW2 that still work very well, with capacitors and resistors rated for a lot more heat than they get. Possibly 70 years is more life than we should expect from an LED lamp, but lasting ten years would be a good start.

Are any of the vendors looking at 2 part assemblies seperating bulb from power electronics? This should allow us to change only the part that fails and allows improvements to the support circuits or 'upgraded' support modules for some applications.

The question is how hot the cap is running at. if it is at 75C, a 125V 2000hour cap will have 64,000 hour life. if it is at 65C, it will have 128,000 hour life. big difference here.

I for one hate paying premium dollar for something and then find out that some weenie tried to save a few cents by spec'ing a cheap part. However, given the design space, an electrolytic is probably the only option to get the voltage and capacitance where they need it. The CV product for aluminum electrolytics is much more favorable than even MLCC's at the higher capacitance values. I spec'ed some 7565 postage stamp sized ceramic capacitors (4.7uF at 250V) for a military job but those babies were about $40 each.

Electrolytic caps (even low ESR, 105 degree) fail with regularity on every consumer device. But besides that, the second most often failure is the heat-induced oxidation and resultant residue conduction of the glue you see holding the parts to the board in the photo. The Orientals have known about this type of adhesive since the 1970's and are still using it, so it must be intentional. As the CEO of EDS-Inc (maker of the CapAnalyzer) and 40+ years experience in consumer electronic service, I can tell you that product life is calculated to fail short of expectations as part of doing business.

Well, most 'thermally insulating goop' will be a better thermal conductor than air for convection cooling. Unless it's fan cooled.
Margery re temp ratings on caps. This is important, almost as much as lifetime in hours at rated volts and current. Lifetime can vary between 1K and 10K hours, a big difference, and can mean the difference of more than a few years in actual practice, especially if the parts are derated greatly.
Tim

My experience with electrolytic capacitors shows the lowest voltage rated parts are least reliable. Probably because of thinner plate separation and less total electrolyte. These things are being pounded on by switching supply applications. Give them a break. Derate temperature, ac current and voltage substantially. Don't overminiaturize. Smaller is not better with these caps.

On the other hand, thermally CONDUCTIVE goop around the caps would be good if it helped to dump heat to the pcb and case. So is the white stuff thermal epoxy or plain old RTV silicone?

What are the chances you will find MTBF listed along with lumens, power factor, color temp and power consumption on lamp packages?

Does saving a few pennies on the BOM of a $30 light bulb really matter? I'd rather have a bulb that is guaranteed to last its 25 year advertised lifetime. Picture a world where things cost more, but last much, much longer. That's where we need to be headed if we're serious about the environment.

What does "kbsbt" mean? Is it rude and should it be here?

Doing a MTBF using MIL217D on an assembly shows electrolytics as having the shortest life of any component. Careful choice and positioning to reduce temperature can improve this short life.

kbsbt

You can derate all you want with a blindfold, pencil, and paper, but as soon as the packaging engineer decides to put thermally insulating goop around your caps, it's all out the window and you're back to cooking caps again.