| Electronics > Projects, Designs, and Technical Stuff |
| "MLCCs ... do not require any voltage derating" |
| << < (4/5) > >> |
| Someone:
--- Quote from: T3sl4co1l on May 15, 2020, 11:32:29 pm ---AFAIK, ceramic failures are more commonly due to cracking, which can be poor soldering profile, improper handling, strong vibration, etc. One product (I work on from time to time), the customer has had consistent problems with a few large capacitors around the edge of the board; probably they're getting stressed during assembly? (Fortunately, they just announced a re-design campaign! I can get my hands all over this finally, instead of the small maintenance fixes I've done over the years.) It's often standard in automotive to use caps in series, so that your 12V, 30A+ circuit doesn't unload through a puny cracked capacitor. There are also types with floating electrode design (effectively two caps in series, in one part), longer terminations (so the crack propagating from the solder joint tends not to cross the electrode stack), and soft terminations (so the crack propagates through the metallization, avoiding the chip). --- End quote --- Yes, cracking is far more commonplace than wear out/age related failure. But I've seen the evidence of capacitors with otherwise similarly excellent mechanical handling/use fail at elevated rates with increasing voltage. Ceramic caps are so reliable on the wear-out end of the curve you need to have an otherwise very reliable design and/or large volumes to notice. Cracks usually show up as an open or short, not always caught by a functional test. But I bet there would also be partial shorts with elevated leakage or not-quite short. --- Quote from: wraper on May 15, 2020, 11:36:18 pm --- --- Quote from: Someone on May 15, 2020, 11:17:03 pm --- --- Quote from: thm_w on May 15, 2020, 10:21:21 pm ---I could never get any ceramic capacitors to fail on my ~60v power supply, now I know why, its not even close. --- End quote --- When they do fail it can be spectacular. Lots of energy in a tiny volume = explosion and/or flames. --- End quote --- More like dead short and nothing spectacular happens. As it's dead short, not a lot of power to dissipate. And ceramic does not burn. If something will burn spectacularly due to this short, it will probably be something else. --- End quote --- Play with some capacitors to failure, they aren't a sharp transition straight to 0 ohms. Likely some thermal runaway positive-feedback effects too. An example of abusing a ceramic cap: https://www.eevblog.com/forum/projects/mlcc-discharge-curves/msg735041/#msg735041 That was done with the expectation that it could detonate, and suitable physical precautions were in place despite the supply being power limited. |
| tggzzz:
--- Quote from: 741 on May 15, 2020, 11:58:03 am ---I'm looking at a Multicomp (Farnell/CPC) branded item, MCCA000593, 'X5R', CPC code CA08032. --- End quote --- Multicomp is a Farnell invention - they buy whatever they can that meets their spec. Hence you might need to re-test each batch you buy. |
| Siwastaja:
--- Quote from: wraper on May 15, 2020, 11:36:18 pm --- --- Quote from: Someone on May 15, 2020, 11:17:03 pm --- --- Quote from: thm_w on May 15, 2020, 10:21:21 pm ---I could never get any ceramic capacitors to fail on my ~60v power supply, now I know why, its not even close. --- End quote --- When they do fail it can be spectacular. Lots of energy in a tiny volume = explosion and/or flames. --- End quote --- More like dead short and nothing spectacular happens. As it's dead short, not a lot of power to dissipate. And ceramic does not burn. If something will burn spectacularly due to this short, it will probably be something else. --- End quote --- I have had a 1210 X7R ceramic capacitor (Murata) fail as a few ohms short, not blowing the fuse, and actually glow bright orange, converting everything in vicinity to smoke. I guess surface tension kept it in place even if the solder melted. I happened to be nearby and cut power in some 5-10 seconds. Eventually it would have evaporated the PCB traces, but the results are not nice. Yes, the reason was me failing at correctly soldering the parts. Take the appnotes about cracking seriously. Not only board flex, but also soldering. - Voltage derating isn't an answer, though. Although you could see a small reliability improvement by luck, the chances are, you see a decrease in reliability if you have to choose a larger package because of the voltage derating policy; larger parts crack more easily. The only real solution is to perfect the mechanical design and assembly (soldering) and handling processes so that the cracks simply don't happen. Use soft termination parts as an extra layer of safety, then. |
| wraper:
--- Quote from: Siwastaja on May 16, 2020, 09:35:56 am ---Voltage derating isn't an answer, though. Although you could see a small reliability improvement by luck, the chances are, you see a decrease in reliability if you have to choose a larger package because of the voltage derating policy; larger parts crack more easily. The only real solution is to perfect the mechanical design and assembly (soldering) and handling processes so that the cracks simply don't happen. Use soft termination parts as an extra layer of safety, then. --- End quote --- It's not that simple. Parts which push spec boundaries for the size and have a lot of very thin layers are way less reliable. |
| Siwastaja:
Yes, the point exactly was that it's not that simple. Soft terminations certainly help. Following the guidelines for stress-free mounting definitely helps. Everything else, not that simple; too many factors. Voltage derating for example - no guarantee of increased reliability. Layers are thicker, but cracking is still dangerous: a shear of, say, 100um, still shorts them out. In my example (glowing bright orange), voltage rating was 50V and applied voltage was 10V. No mechanical stresses, just soldering process violations. |
| Navigation |
| Message Index |
| Next page |
| Previous page |