Super Cap leakage current

[itsmoked]

I'm examining a data sheet for a supercap for a memory hold-up application.
Normally Nichicon has great products and this supercap may be another except they've got some cockamamie brief statement of what the ALL IMPORTANT leakage current is. No where do they define it clearly - to me anyway.

In the following data sheet they state:
Leakage Current | 0.5C(mA) [C:Rated Capacitance(F)]

[URL unfurl="true"]http://nichicon-us.com/english/products/pdf/evercap_en_uw.pdf[/url]

General Nichicon SuperCap data sheet.
[URL unfurl="true"]http://nichicon-us.com/english/products/pdf/evercap_en.pdf[/url]

Could they really mean that a 1Farad offering is going to have a 500uA leakage current??!! That's 50 times more leakage than the next company's offerings.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

Yep. Seems clear.

"0.5C (mA)", where C is value in Farads.

.: 1.0 F would be 0.5 * 1 in mA, or 0.5mA, or 500uA.


There's a one in a million chance that they intend mA = microamp (basically a newbie typo), but extremely unlikely.

Or this might be one of those worst case things, but no guarantee if it's that bad.

I'd 'make friends' with an Application Engineer there if possible. Emails, possibly leading to a long telephone call. Typically worth a fortune, but free.

 

[itsmoked]

Gag..

Thanks much for the confirmation VE1BLL. I'll have to change to a 76 cent cap instead of this 42 cent cap.


I've got a perfectly good hold-up using an Li 2032 coin cell that provides months of backup for about 32 cents for the battery and the holder but due to bureaucrats you cannot ship these puny coin cells on passenger aircraft period or more than 4 mounted inside products with the USPService. So I have to dump a multi-month 32 cent solution in lue of a 90 cent solution that holds up the memory for a delightful 20 hrs.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[IRstuff]

The supercapacitor performance seems to be consistent with low leakage capacitors, and possibly somewhat better. This

http://www.mouser.com/new/Kemet-Electronics/kemetT489caps/

would have about 75mA leakage current if scaled up to a 1F size, so a 0.5mA leakage is pretty good by capacitor standards.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

Of course I can. I can do anything. I can do absolutely anything. I'm an expert!

 

[Skogsgurra]

We are living in parallel universes, Keith!

The reforming, that is said to be necessary if a VFD has stayed inactive for a prolonged time, and the discussion that we had some time ago has worried me somewhat. So, I found a few old electrolytics that has been around, and never in operation, for at least ten years and did some tests on them. They were 2200 uF/16V and I connected them in series with one 10k resistor each to be able to watch the leakage current over time. I ran them at 15 V, well-regulated and without ripple.

It developed into a small test in my ongoing "The Automation Engineer and the Reality" and the conclusion is that you do not have to worry much about small capacitors at low voltage. There's simply not enough energy available to start the leakage current->high temperature->increased leakage->higher temperature avalanche that causes large capacitors to explode.

I haven't finished the text yet, but the results were interesting, I think. At least, they were not alarming.

All five capacitors had an initial leakage current that was very similar to the other's and with a mean value close to 6.0 uA.

The mean leakage current after 1 h operation was 4.2 uA and 20 h later all leakage currents had dropped to 1.1 uA.

This was at room temperature, around +22C and I haven't done any tests at elevated temperature yet.

The most interesting fact here is that:
1. these capacitors behaved quite well after more than ten years of sleep.
2. they all had very similar leakage currents

I then reversed polarity (yes, a NO-NO!) and watched leakage current when voltage was increased from zero up to 5 V. It grew überproportional and there was a tendency for an avalanche when I interrupted the experiment. Leakage current (reverse) was then all over the place with 130 uA as the highest value. What was really interesting is that the capacitors behaved quite well as long as reverse voltage was kept below around two volts.

NOW - listen carefully - I then restarted the leakage current measurements and after around 1 h, the leakage current was back to a low 1.7 uA and sinking.

These measurements take long time. I logged it all with a recorder - so I wasn't tied up for 20 hours, but still. I have gained some insight and will be doing the same test with a few other capacitors. If there's any thoughts about what to look for - please let me know.



Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[itsmoked]

Interesting point IR. Thanks for highlighting that point.

The ultimate cap I guess I'll have to use is this one:
[URL unfurl="true"]http://www.cooperindustries.com/content/dam/public/bussmann/Electronics/Resources/product-datasheets/Bus_Elx_DS_4376_HV_Series.pdf[/url]

You'll note its leakage is only 10uA @ 1.0Farads. So a lot better than the scaled version of the Kemet and way better than the O.P. one that bleeds copiously at 500uA.



Gunnar: Very interesting... What you're seeing jives closely with all the supercaps I'm looking at. They all make a big fuss about the leakage current value being viable only after some fixed time under bias. Usually at least 30 minutes. I'm very pleased to see that after reverse biasing your caps seemed to take it in stride and return to low leakage performance.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[iop995]

Supercap main advantage over battery is fast charge / release energy cycle. If your application is for memory cache RAID controller or similar, supercap is ok but must be used to provide power to RAID card for few (mili)seconds, enough to move data from volatile cache memory to a non-volatile one, when a main power failure occur or other failure. Supercap can't replace direct battery but may do same job (keep memory data) but in different way. Similar use of supercap are in EV in positive/negative accelerate phases, short brake phase charge ultacaps and then it supply current for accelerate phase, "helping" main battery.

 

[itsmoked]

iop995; This application is to keep a 500nA clock/ram chip happy for a reasonable power outage period.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

If it's 500nA, what about a non-Li cell?

Also, triple-check the bureaucrats. Usually the specs themselves are better written than that, and would provide explicit exceptions for tiny cells.

On the other hand, I've noticed that some of the batteries used in avionics, can't be shipped by air. I guess they're trying to preclude shipping an entire box of them by air, but still...

 

[iop995]

For such low current, think at this schema: if voltage for RAM/clock memory is 5V, you need about 2.5microW power; to avoid battery or ultracap, nay use remaining filter capacitors energy at power failure. Add a step down circuit (cheapest a resistor w/ zenner) or trafo dc converter if you need isolation. I made same calculation and for 1000microF caps at 100Vdc remaining voltage with 50% efficieny for dc converting, result abot 270hours or ~10days back-up time; ok maybe much less if add caps leakage current.

 

[Skogsgurra]

We have a hend-held instrument that consumes about 15 uA when in sleep mode. It is being used perhaps ten times/year and then it consumes between 15 and 20 mA for a short while. We chose AA cells (R06) and that works wonderfully. Up to three years before cusomers need to change battery. Customers usually ring and ask why the dislay is bleak (runs directly off battery) but never had a problem with lost data.

So, I agree with VE1BLL - if there is room, AA or AAA works very well - but may need to be changed every third or fourth year.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[VEBill]

50% efficieny for dc converting

I'd rethink the 50% assumption at the 500nA load point.

 

[iop995]

I think main problem is that battery are avoided in that application (maybe customer request and/or new EU rules about RoHS) or need a lot of tests/aprovals that increase cost. So, need to find a solutions w/o battery. At so small current I think it's ok with caps or ultra-caps.

 

[IRstuff]

Keith,

I seem to have missed some numbers along the way. 500nA * 1month/ 1farad = 1.3V ?? Just what is the allowable droop for your circuit? Seems to me that even the 10uA leakage is too much; I seem to get a droop of 26V after 1 month.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

Of course I can. I can do anything. I can do absolutely anything. I'm an expert!

 

[VEBill]

I have a little GPS circuit card that uses a button cell Iron Nickle rechargable battery to hold the data in volatile memory while the power is off. I don't believe that type of cell falls into any of the restrictions on shipping by air, but I'm not sure. It has to be cheap too; the entire GPS board including wee little antenna was $15 shipped.

The point: There are alternatives to lithium. If lithium is the issue.

 

[itsmoked]

VE1BLL; I looked at that. The A76 Alkali button cells are 150mAhrs which is about 70mAHrs less than the Li batts but worse are only 1.5V which requires you use two.
Battery cost = $0.36 x 2 = $0.72
Holder cost = $0.21 x 2 = $0.42
Total Solution = $0.72 + $0.42 = $1.14

(140mAHr/500nA)/24Hr/day = 11,666days. Hmmm. That could be enough.
Interesting idea iop995. This thing only has about 31V available on regular electrolytics. And the biggest problem has been the switcher it uses to isolate and drop the voltage, so at the moment another switcher is not something I'm too eager to embrace.

Gunnar; I actually switched to an AAA which fit and was about a $0.50 solution before re-reading the clock chip data sheet and seeing a minimum battery voltage of 2.0V is required. Can't fit two AAAs.
iop995; It appears transportation doesn't have any interest in non-Li batteries.

Hi IR.
I don't think I'm following you. My clock circuit needs 500nA to stay happy. Super caps only really come in 2.7V or 5.5V versions for this app. 2.7V doesn't give any headroom since it has to be charged thru a 0.6V drop of a diode. So the next cheapest supercap solution appears to be 0.33F @ 5V. $1.14 total solution.

Charged to 5V - 0.6V = 4.4V

2V is the dropout voltage. 4.4V - 2V = 2.4V of cap droop.

Equivalent resistance for 500nA = 4Mohm

-4M x 0.33F ln(2.0/4.4) =>>> 12Days (based on the assumption of constant drain which is probably wrong and means the time might actually be 1/2 that.

The supercap solution has an additional problem. The caps are rated with horrid lifetimes. Standard is 1000Hrs at 70C. Pathetic. I have to have another AL cap in this design and I can easily use a 7000hr @ 105C cap for that.
I think I'm going to opt for the two A76s. After choosing and using the Li solution, then changing to the AAA design and laying it out only to recognize it was unusable. Then designing in and laying out the 5.5V supercap.. I am now going to opt for the A76 Alkali solution.

VE1BLL; Cheapest rechargable I can find is 1.78 whoops that's Li based too. A lot of rechargeables are NiCad (horror) or Li (shipping horror) Seems a no go.

Thanks you guys for the dialog. Much appreciated.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[IRstuff]

12 days and 2.4V makes more sense; at that point, you only need 0.2 farad just for the circuit demand, so ANY leakage from the capacitor is already too much. You'd need leakage in the nA range to make it work at all. Battery is really the only practical option.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

Of course I can. I can do anything. I can do absolutely anything. I'm an expert!

 

[MacGyverS2000]

Does it really need to be charged through a 0.6V drop? I have found for many applications, I can get rid of 90%+ of that drop with a little careful planning. For example, I will often use the body diode of FETs in applications, and the drop on that is significantly lower than 0.6V. I would say that's becoming a somewhat common design method, actually. That might give you the extra headroom you need.

Dan - Owner

http://www.Hi-TecDesigns.com