16

Recently I had an explosion on my desk. A 220uF 25V electrolytic capacitor just blown up shortly after attaching a load. enter image description here

The high-level connection was: Phocos CA08 (solar charger controller) -> 12V-3.8V step-down -> uBlox Leon GSM modem.

Because the variable load generated by the GSM modem was causing the charger controller errors (it indicated an overload and cut off the voltage at the output socket) I put a 220uF 25V electrolytic capacitor on its exit and it worked for a week or so with no issues, also no overload indications were occurring anymore. The connection was not firm, it was just attached into the contacts of controller's output. There was an external power supply attached to the the solar cell so that to charge the lead-acid battery. The explosion occurred after ~5 seconds since attaching the K1 to the charger controller's output.

The schematic of a power supply for the GSM modem is the following: enter image description here

The modem itself can generate a significant load when registering to the network (right after power up): enter image description here

It came to me as a surprise that the capacitor like this exploded this way, especially when it is not the only capacitor on the path between the supply and modem, there are plenty of them.

Now I wonder what happened and what might have caused the explosion of this type. I did not change the polarization of the capacitor, I'd bet that its leads did not met each other. Do you have any ideas?

Ricardo
  • 6,134
  • 19
  • 52
  • 85
eltomek
  • 921
  • 1
  • 9
  • 18
  • 7
    Typically an explosion like that is caused by either over voltage, reverse polarity, or exceeding the ripple current rating. What't the part number for the capacitor? – John D Jun 02 '14 at 19:22
  • @JohnD sorry, that was 220uF 25V. Unable to read part number right now though... – eltomek Jun 02 '14 at 19:31
  • If you're sure the polarity was correct and there's no way it was exposed to over voltage, you likely exceeded the ripple current rating on the part. – John D Jun 02 '14 at 23:42
  • 2
    From the photo I'm guessing it's Nichicon UPM1E221MPD6TD (size 10mm x 12.5mm) or pretty similar. If so then 220uF 25V is rated 670mArms ripple (105C / 10kHz to 200kHz), impedance 0.13 ohm. From the graph it looks like you need to supply 2A current in less than 100usec, accurate to within the GSM modem's supply tolerance... Ceramic doesn't require as much voltage derating as other types, so I'd try some ceramic MLCC 220uF/4V, since all of that transient current has to come from these additional output capacitors. Taiyo-Yuden AMK325ABJ227MM-T (mouser.com). – MarkU Jun 03 '14 at 03:20
  • 4
    What a pretty explosion and picture! It could be a picture for wikipedia! :) – Kamil Jun 05 '14 at 05:33

2 Answers2

2

There are plenty of counterfeit caps coming out of China. Can you tell if there was a smaller cap in the larger package? If you buy bag-o-cap type parts on Dx or eBay it is a good idea to cut one open from each batch and make sure there isn't some little totally wrong cap on the inside.

C. Towne Springer
  • 2,141
  • 11
  • 14
0

Maybe your PWM charge controller might have looked a little bit like AC with a DC bias and the AC would have been passing straight through the capacitor causing it to heat up and fail?

BillyB
  • 1
  • 2