The most common failure mode I notice in cheap consumer electronics is failure of capacitors. A rough rule electrolytic capacitor life is it doubles for every ten degrees below the rated temperature. If the capacitor does not handle large AC currents causing self-heating, the effects of temperature predominate over factors like how close to that rated voltage it is used. When I have to select an electrolytic cap, I take the life of the product and temperature at the capacitor’s location and select a part with at least that much life. Capacitors are not a big part of the BOM cost, so I can usually afford some headroom.
I decided to do a survey of capacitor prices versus life to understand how much longer capacitor life really costs.
Note:
Since I’m posting this on a Newark website, I chose Newark’s parametric search to find parts to compare. I find the element14 blog and forums to be a bit slow and clunky. It has been steadily improving, but I still think it would be better using some open-source system. Newark’s parametric part search, however, was stable and reasonably quick. It has more search flexibility than other online distributors’ sites. I will be using again when I need to narrow down parts parametrically.
For the survey I looked at 100uF, 25V, Radial, 20% tolerance, caps with ESR > 1 ohm: Your basic low-cost bulk decoupling cap.
| Rated Life | Rate Temperature | Life at 50C [hours] | Life at 50C [years] | Cost at 6k [USD] | Estimated (See formula blow) Cost [USD] |
|---|---|---|---|---|---|
| 1000 | 85 | 11,314 | 1.3 | $0.03 | $0.04 |
| 1000 | 85 | 11,314 | 1.3 | $0.04 | $0.04 |
2000 | 85 | 22,627 | 2.6 | $0.13 | $0.06 |
| 1000 | 105 | 45,255 | 5.2 | $0.06 | $0.10 |
| 1000 | 105 | 45,255 | 5.2 | $0.05 | $0.10 |
| 1000 | 105 | 45,255 | 5.2 | $0.09 | $0.10 |
| 1000 | 105 | 45,255 | 5.2 | $0.05 | $0.10 |
| 2000 | 105 | 90,510 | 10.3 | $0.18 | $0.19 |
| 2000 | 105 | 90,510 | 10.3 | $0.10 | $0.19 |
Here is a graph of cost versus life at 50C:
Drawing a line through this cluser plot and finding the equation for the line, I find Estimated Cost = $0.017*(Life in yrs) + $0.014. That means you pay roughly an extra 1.7 cents for every additional year of life.
From my experience, very cheap consumer electronics’ performances drops after only a few years, suggesting the manufacturers use the cheapest possible, 1000hrs at 85C, which is seven years at 25C. Add some self-heating if some of the caps are handing current and consider many caps whose failure probability distribution function is centered on seven years, and you get problems after only a few years.
Sometimes failed caps physically expand or leak, and you identify which part to replace. If the failed part is not obvious, it’s not worthwhile, even if we attempted to account for the cost of the e-waste on the environment, to replace every capacitor on the board.
The good news is I don’t encounter this problem on more expensive electronics. My Sony ICF-2010 shortwave receiver works always has for the past 20 years.
Should manufacturers have to specify their products’ MTBF? Should their be some sort of tax on short-life “disposable” electronics that contribute unnecessarily to e-waste?