Fake caps are bad, but I never thought that bad

[Electr0nicus]

Nearly two years ago I bought a pack of 50 1500uF 10V "nichicon" branded caps at my local electronics store. I actually thought that caps were genuine nichicon ones, so I expected them to work quite well. I then used those caps in some of my projects, I made back then. One of those projects was a 320x240px color TFT screen, with a SSD1906 display controller and a AVR. It is powered by a 8-15V plug- pack. On the PCB the 3.3V and 5V rails are generated with standard 78** linear dropout regulators. On the output side of the 5V and the 3.3V regulator, I placed one of these caps for each output.
So far so good. Two years back the circuit worked fine with those caps. Since then, I never powered this circuit up again. Until today. I plugged in the plug pack and turned away from the PCB for just 4-5 seconds, until i heard a high pitch noise from a capacitor venting. I immediately disconnected the power and checked what was wrong. In the first moment I thought I accidentally used a plug pack with the false output polarity, but then I remembered that this wouldn't have been a problem because I designed in a diode for reverse polarity protection. So after a short time investigating the problem it was clear that the capacitors were the problem.
I then randomly picked one of the nearly 40 left over caps of that type out of my parts bin and connected 5V to them via my lab power supply. Instantly after I connected the leads the cap pulled nearly 2A  from the PSU and heated up quickly. 
I then tested another 10 caps, all were absolutely dead. even at 3V they pulled 800mA. I'm a bit surprised, I've heard of fake caps having a very short life span, but I've actually never heard of caps being bad at the first time voltage is applied to them.
 

[PedroDaGr8]

Caps which have been sitting for a while also can have issues. The aluminum oxide dieletric layer breaks down which basically results in a short between plates.

Cheap caps can have compounds in the electrolyte which break down this layer quicker.

[Legit-Design]

You could try building up the oxide layer by slowly charging them. I think I saw some instructions for this somewhere, would be interesting to see it in practice.

here ya go:
http://www.epcos.com/web/generator/Web/Sections/Components/Page,locale=en,r=247996,a=490592,principals=none_21none_21_20_21_20_21_20,print=1.html
So according to that the leakage current could last several minutes until the oxide layer will start to build up again.

http://forum.talkingmachine.info/viewtopic.php?f=7&t=262

This process of restoring old electrolytics is called "reforming" the capacitor. It has been known and done for decades. It probably started from people being too cheap to replace the tired capacitors. But it has new relevance today as people wish to restore old electronics and retain as much of the original equipment as possible. The basic method is to bring up the voltage on the electrolytic slowly and monitor the current in the circuit to keep it at a low level of a few milliamps or so. Ideally, a good electrolytic shouldn't be leaky at all and there will be almost no current flow thru it at full rated operating voltage. If the capacitor is capable of being reformed, the current thru it will gradually decrease. You then increase the voltage across it and continue to monitor the current. If the capacitor is still capable, it will be possible to eventually place full rated operating voltage across it with very little or no current flow. At this point, you may want to actually measure the capacitance to see if it still has some or if it has merely burned out and become an open circuit.

[Psi]

Yep, some of the bad caps are truly bad.

We ordered a small custom run of computer motherboards from china 3 years back at great expense.
They were for use in xray machines where an ISA slot is required as well as other things, (only a few makes/models of boards worked and none were made any more).

One of the new boards was put into use when they arrived and all was fine for about 6 months before it started crashing. (bulging caps).
When we grabbed a new board off the shelf we found it also had a few budging caps and so did all the others spare boards.
These were boards that hadn't ever been used, they got put on the shelf when they arrived!

[Spunky]

I was selling PC motherboards in the 90s when it was really bad, some boxes had 17 out of 20 boards DOA, same for monitors. Cost me a lot of money just in lost time and shipping costs.

Then there was all the dodgy hard drives after Seagate bought out Connor, and the S3 graphics cards with interference all over them, the overheating 3DFX cards, and then the sound cards where the designers hadn't actually drawn the track connecting the sound chip to the amp circuit (unbelievable I know).

I don't think it's so bad now but at least I'm out of the IT business.

[tocsa120ls]

This happens more times than it should, but still bugs me when I repair something. This was in a Chicony-made chinese charger for an Acer Aspire One netbook. Unbeknownst to the owner, it shared an outlet with an arc welder
Old cap blown, replacement cap more, than 20% bigger with a lesser voltage rating. Someone is lying or someone is overtly honest? I will never know.

[tom66]

Or Ltec has better technology than Antel. Ltec are quite a big company (they were involved in the "stolen electrolyte" story that gets passed around and may or may not be true.)

I have used a 100uF 400V Rubycon that was half the size of its replacement CapXon.

You should not use a 400V capacitor where a 420V has been spec'd, because most likely, in a laptop charger, you have an active PFC circuit which will boost the output to 400V +/-10V ripple, which could exceed the rating of the 400V cap.

[Kjelt]

You should not use a 400V capacitor where a 420V has been spec'd, because most likely, in a laptop charger, you have an active PFC circuit which will boost the output to 400V +/-10V ripple, which could exceed the rating of the 400V cap.

Good point, but that would also exceed the 420V cap    Use a 450V instead.

[tocsa120ls]

You should not use a 400V capacitor where a 420V has been spec'd

Believe me, I did not want to fix it this way, but that was the only I could get my hands on in reasonable time. Here in Budapest we have like 4-5 shops who sell components, 2 of them had 68uF caps, and one had this 400V one. Didn't have time to order a 420/450V rated. Did some Vpp measurements, the cap is well within its comfort zone (max value was ~320V). I did load the thing all the way to its max 2.1 amps, did not fluctuate more, than 5% up-down between no load and max load.

[mariush]

Nearly two years ago I bought a pack of 50 1500uF 10V "nichicon" branded caps at my local electronics store. I actually thought that caps were genuine nichicon ones, so I expected them to work quite well. I then used those caps in some of my projects, I made back then. One of those projects was a 320x240px color TFT screen, with a SSD1906 display controller and a AVR. It is powered by a 8-15V plug- pack. On the PCB the 3.3V and 5V rails are generated with standard 78** linear dropout regulators. On the output side of the 5V and the 3.3V regulator, I placed one of these caps for each output.

Could it be that the plug-pack gives more than 10v under low load, therefore exceeding the voltage rating of the nichicon capacitors?  Usually good brand capacitors have no problems lasting long time with voltage very close to their maximum rating, but everyone recommends you leave about 20-40% margin (ex use 10v rated caps for 5v, 16v rated caps with 9-12v dc input etc)

[amyk]

Capacitor size may also differ because one is lower ESR than the other. 420V is a bit odd of a voltage rating, I usually see 400 and 450 more.

Nearly two years ago I bought a pack of 50 1500uF 10V "nichicon" branded caps at my local electronics store. I actually thought that caps were genuine nichicon ones, so I expected them to work quite well. I then used those caps in some of my projects, I made back then. One of those projects was a 320x240px color TFT screen, with a SSD1906 display controller and a AVR. It is powered by a 8-15V plug- pack. On the PCB the 3.3V and 5V rails are generated with standard 78** linear dropout regulators. On the output side of the 5V and the 3.3V regulator, I placed one of these caps for each output.

Could it be that the plug-pack gives more than 10v under low load, therefore exceeding the voltage rating of the nichicon capacitors?  Usually good brand capacitors have no problems lasting long time with voltage very close to their maximum rating, but everyone recommends you leave about 20-40% margin (ex use 10v rated caps for 5v, 16v rated caps with 9-12v dc input etc)

The caps were on the regulator output, so they never saw more than 5v.

[GreyWoolfe]

I tend to stick to Panasonic or Illinois Capacitor products.  Even after sitting for a couple of years, there is no issue with them.  I used a lot of the I C caps for repairing Dell 15" monitors a few years ago that had the same infamous "CrapCom" caps that were also in the OptiPlex GX-270 motherboards.  Still have a bunch and am using them when I have a need for the same values.   I am willing to pay a bit more for something that won't give me aggravation down the road.

Tom, NW0LF