supercap leaking acid

[martin_nga]

Hi @all,

short question:

we're having some serious trouble with super capacitors (3F) leaking acid after ~2 years of operation in the field.
Is there anybody out there who had a similar problem (and maybe a solution)?

not so short question and further details:

We have a building control PLC that runs Linux OS. The use case for this device is to run 24/7, however, when power fails, we need to shut down and store some data on the flash. To achieve this, we have an on board ups that buffers power for ~15 s after power fail is detected.

The ups consists of four super caps, 3 F each, in series, giving us a voltage of 10 V fully charged. The caps are charged with constant current until charge end voltage is reached, then we keep the voltage constant. The caps also have a voltage balancing circuit.
Charging and balancing circuit is verified to work as expected, keeping the voltage across each capacitor within tolerance.

The device is in the field for ~3 years now. When doing maintenance work at a customer,  our men discovered that on >75% of the controllers, the capacitors have leaked acid onto the pcb. On some boards, the acid already damaged traces and vias (see attached images). I seems that this starts after ~2 years operation.

I've done some search on the net, and the only occurrence of such a case I've found is this one here: https://www.eevblog.com/forum/testgear/fluke-189-with-leaking-surface-mount-coin-cell/225/. Of course, Fluke replaces failing meters...

We've also contacted our distributor, which actually admitted that there is a problem (called "wetting"). However this only occurred the first time a year ago or so, and that nobody really knows what causes it to happen. He also told us that this is not a manufacturer specific problem but industry wide.
We have asked other vendors and distributors and -what a surprise- none of them ever heard of this.

So we don't know who to believe...

- Did anybody of you have a similar problem? Maybe even know what causes leaking?
- We've been told that reducing temperature and voltage will increase lifetime significantly. But our problem is not EOL because of capacitance reduction but because of acid leaking. Any ideas on that?
- Any other Ideas?

Thank you for your help!

[HighVoltage]

The only time I had a similar problem was when the voltage came close to or right at the maximum allowed voltage on super caps.
I have learned to stay about 5% away from the max voltage
On 2.70V supercaps, I go to a maximum of 2.55 V for longevity reasons.
 

[SeanB]

Like Highvoltage says, add an extra capacitor in series, so any one only has 2V across it, or drop the supply to 8V and accept the shorter runtime. one requies a board respin and the other likely only a component change, depending on your design.

As well it would probably be best to make a small daughterboard  with the capacitors and balancing circuitry on it, and connect to the main board with a small cable, so it makes the damage less on these that do leak. I would also suggest placing a reverse biased Shottky diode across each capacitor, so that the run down does not reverse bias the one capacitor, and cause it to fail as it degrades and drops in capacitance. Current rating a little more than your charge or discharge current, whichever is larger. I would also suggest a fuse if making the daugherboard, so a shorted cable does not blow up traces.

Supercaps are rated for a voltage, but if you read the datasheet you will find they are not going to be usable at this voltage over the full operating temperature range, so a derate is in order.

[edavid]

Use an extra cap. Commonly seen EDLCs are designed to operate at 2.5V or 2.7V per cell, not 3.3V. If I were you I would use at least 4 of them, and decrease 10V charging voltage by 0.5~1V.

Look again, there are 4 caps on the board (as the OP stated).

[martin_nga]

Thank you for your answers!

We're charging each cap to 2.5 V, rating is 2.7 V. There are four of them on the PCB.  We are also within spec regarding temperature (max. ~50°C, some devices much less).
We already know that we have to reduce voltage in order to increase (regular) lifetime.

Before starting this thread, we have already discussed the issue in house and came up with the same changes (reduce voltage/add a cap, place caps on daughter board, change manufacturer). I left those out to maybe get some fresh ideas 
But we don't know how far reducing the voltage will increase lifetime, and the manufacturer can't/doesn't want to tell us. If this gives us six years instead of two, we only have postponed the problem to 2022, only difference is that more devices will be sold  . If that is the case, we have to remove them from the board.


@blueskull, regarding pcbs
We had some discussion with our pcb manufacturers some time ago, we ended up with them (hopefully) knowing their process the best...

[Richard Head]

Probably of no value but we had a PSU board that had three output capacitors on the edge of the PC board just like yours. We found in production that the caps would often get bumped skew thereby pulling one of the leads out of the can a little. It caused control loop issues as the ESR was affected. Once we started paying more attention to the bumping issue the problem went away.
In your case perhaps some lead pulling is causing the seal to fail prematurely.
Also, have you checked to see if the soldering temp isn't being exceeded? Maybe they are temp sensitive.

[NiHaoMike]

Maybe form the leads so that the caps sit parallel to the board, with some electronics grade silicone to hold them down? That should help reduce stresses on the seals. It also looks like it could be done without redesigning the board.

[MagicSmoker]

This sure sounds similar to the "capacitor plague" problem from a few years ago, which, IIRC, was the result of a bad batch of electrolyte. This issue affected many different capacitor manufacturers because they all used the same electrolyte manufacturer. It was a real mess... literally.

If something similar has happened here then the only economical option would be to replace the capacitors with a different brand (and, ideally, a different type - there are several different ways to make a "supercapacitor").

The only reason I would continue using the same supercap is if I found a mechanical cause - like strained leads, as pointed out by Richard Head, or some other kind of physical abuse during assembly or installation.

 

[Ian.M]

Interesting thesis: http://www2.uwstout.edu/content/lib/thesis/2011/2011nerisond.pdf

It looks like there may be rogue brands with systemic defects in their manufacturing process.

There's also the mechanical stress issue - a redesign to cluster the caps so they can mutually supported by gluing, and to move them further from the board edge may be justified.  If there is a lot of existing unpopulated PCB stock, NiHaoMike's suggestion of lead forming, laying flat and gluing down may be worth looking into, as may SeanB's daughterboard suggestion, which would also aid refitting units currently in service.

[martin_nga]

Thanks for your answers!

We used multiple batches of capacitors, with date codes that are month apart. So hopefully not an electrolyte problem...

The thesis is interesting, however it is lacking the negative test (why wait for months and not put some of the old type in the climatic chamber as well, just to verify that they're failing during the test  ).

[amyk]

we're having some serious trouble with super capacitors (3F) leaking acid after ~2 years of operation in the field.

Are they these? http://www.nesscap.com/common/download.jsp?dir=product&sfn=BKBTGVMASWXYNND.pdf

"Shelf Life: 2 years" doesn't sound that great for your application.

I would go with a small SLA or NiMH, since it seems you don't need much capacity.

[jh15]

Really nice you want your product to be durable. Wow, more than 2 years?

The fruit named fashion product  company would love you to give them these part numbers for their 18 month cycle.

They might even be able to not bother with software death items.

How many tons of hardware in schools and homes is rendered for the trash heap because a 10.6 is no longer secure or usable? And disabled to put Linux on?

[LabSpokane]

How many caps are failing on the board?  Just one, or more than one per board?

[martin_nga]

"Shelf Life: 2 years" doesn't sound that great for your application.

this is uncharged storage life. In our application, it's charged all the time.

I would go with a small SLA or NiMH, since it seems you don't need much capacity.

They also need maintenance every few years, so no advantage compared to a separate board with caps on it. Except the chance for them to last longer 

Really nice you want your product to be durable. Wow, more than 2 years?

We're not doing consumer electronics. For us, warranty by law is five years. The expected lifetime of a building is at least 20 years...

How many caps are failing on the board?  Just one, or more than one per board?

All of them, however we do not know if one starts and causes the others to fail (chemically or electrically)


Our distributor recommended us to switch to Nesscap XP series caps (e.g. http://www.nesscap.com/common/download.jsp?dir=product&sfn=XVEAFMTNYUGMPYE.pdf&ofn=2.7V-6F%20XP%20Datasheet_REV3_20160404.pdf) According to the datasheet, they have improved sealing to reduce leaking effects. However, they do not guarantee that they will last any longer. On the sample caps we received, one can see that they added some glue to the leads. The other changes are not visible to the eye (except the pink color).

[R005T3r]

If I were you, I would use batteries instead of capacitors: there are some pcb lithium batteries with various size and capacity.
Also, if you say that the caps already have a voltage stabilizer, switching to them require minimum effort, and as you said before, the caps are charged all the time, so they will last longer. If you are concerned about maintenance every few years, you can opt for a pcb replaceable battery tray. Something like this:

[LabSpokane]

It looks like you have a MOSFET-based balancing circuit, is that correct?  If so, are you sure that there isn't some type of unpredicted behavior there?  Have you taken an old, non-leaking cap from a field unit and placed it into a new board to see how the balancing circuit behaves in the presence of an aged super cap?

[Yansi]

Interesting thesis: http://www2.uwstout.edu/content/lib/thesis/2011/2011nerisond.pdf

That was one hell of a long read  Interesting and just to say wow, that medical appliance uses 5V without balancing across two 2.5V caps in series. No engineer there seem to use common sense in design. 
Doing so seems like pure stupidity for me, a lonely forgotten hobby designer.

[martin_nga]

It looks like you have a MOSFET-based balancing circuit, is that correct?

We're using a shunt regulator parallel to a resistive divider. Shunt does overvoltage, resistors do balancing. So the balancing works even with higher capacitance differences.

uses 5V without balancing across two 2.5V caps in series

you can buy this part off the shelf http://www.digikey.com/product-search/en/capacitors/electric-double-layer-capacitors-supercaps/131084?k=supercap&k=&pkeyword=supercap&pv13=623&FV=38001e%2C380029%2C3802da%2Cfff40002%2Cfff8000c&mnonly=0&newproducts=0&ColumnSort=0&page=1&quantity=0&ptm=0&fid=0&pageSize=25
They do not contain any balancing. We bought one to see how the manufacturers themselve do the balancing 

[MagicSmoker]

Is the current drain too high to consider going with board mount rechargeable lithium cells? Supercaps seem to combine the worst traits of batteries and electrolytic capacitors and often are the proverbial "solution in search of a problem".

[Kilrah]

If you are concerned about maintenance every few years, you can opt for a pcb replaceable battery tray. Something like this:

I'd go with that solution too, slap on a couple of lithium-type AA or AAAs as required and you're good for 10-20 years, and can ditch the charging / balancing circuit.

https://www.amazon.com/Energizer-Ultimate-Batteries-Longest-Lasting-High-Tech/dp/B0002DUQDQ

[R005T3r]

Also, if the previous solution is not suitable something more compact may be a button cell type:

[RES]

Try one of the supercap brands Maxwell, Ioxus, Nesscap.

[Owen]

Try one of the supercap brands Maxwell, Ioxus, Nesscap.

They've used Nesscap. Leakage is something that you have to be aware of. Nesscap even state this in the datasheets like the one above: "protective coating of components nearby", "cells may not be completely hermertic during lifetime", "spacing 1 mm or more should be provided". I wouldn't even call these capacitors "fresh" considering two years at 50°C and 2.5V for a C rated 1500 hours @2.7V, 65°C ... A rule of thumb: lowering by 0.1 V from VR doubles your lifetime. Lowering the temperature about 10°C doubles the lifetime as well. So let's say it's now 12000 hours "rated". 24/7 in 2 years are round about 17500 hours. You can calculate this for your five years warranty your own.

[jitter]

Hi @all,

short question:

we're having some serious trouble with super capacitors (3F) leaking acid after ~2 years of operation in the field.
Is there anybody out there who had a similar problem (and maybe a solution)?

The company I work for produces a control board for a rescue tool that the firefighters use to cut people out of crashed cars ("jaws of life"). It has a NEC/Tokin 0.47 F supercap on it. We have been making these boards for at least 10 years, and we haven't had a single complaint about the supercaps. Mind you, this is not a 24/7 application, and the boards are potted after testing, so this may help any leakage from being contained.



Myself I had a bad experience with a Pioneer PDR-555RW with a leaky supercap. After ~6 years it had leaked and eaten numerous vias on the board. It turns out my unit wasn't the only one susceptible to supercap leakage.

We've also contacted our distributor, which actually admitted that there is a problem (called "wetting"). However this only occurred the first time a year ago or so, and that nobody really knows what causes it to happen. He also told us that this is not a manufacturer specific problem but industry wide.
We have asked other vendors and distributors and -what a surprise- none of them ever heard of this.

So we don't know who to believe...

Manufacturers will only admit to having done something wrong if you rub extremely hard evidence in their face repeatedly. After much stalling for time they may eventually and reluctantly admit they made an error... If they can weasel their way out, they will.

[Mikec.Vina]

Sorry I have just found this forum and this subject regarding the EDLC wetting effect.
Our company has been investigating this phenomenon which affects all EDLC capacitor manufacturers, including coin cells which are subjected to higher than recommended temperature and humidity. The effect has been seen in most industry sectors.
We have only seen this wetting effect with the 2.7v products and some manufacturers take longer to leak than others.
The VINATech 3v series has not been affected and offers 3x longer life when used at 2.7v
Is the leaking only on the (-) pin?
If you would like more details I can share this with you. Long term tests have shown that VINATech has over come the solution and a patent has been applied, its more complex than just changing the rubber bung. VINATech, the major manufacturer of EDLCs from 0.5F to 500F is ready to launch a new leaded series called NEO that combats the effects of extreme environments.
The 3v & 6v modules are available at Conrad, Farnell & Rapid online sites if you wish to try them.