EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: Cliff Matthews on November 19, 2016, 07:47:21 pm
-
Hey all, yesterday I bought 10 x 10,000uF 25v el-caps for $5. Back home I tested them at 20vdc and found V dropped 90% (without load) after just 10 seconds after power removal.. I looked up the brand G-Cap on badcaps.net and found nothing. Before going to bed I chained them all across a 9vdc class-2 wall wart and this morning I feel less ripped off since they're holding charge much longer.
(http://a1parts.ca/surplus/images/Power%20Supply%20Filter%20Board.JPG)
As per paragraph 6 at http://www.qsl.net/g3oou/reform.html (http://www.qsl.net/g3oou/reform.html) I now have them all at 21v in series with 10k, currently dropping 2v (200 uA) and falling ever so slowly over the last hour. So with 100,000uF, can I consider these reformed when the voltage drops to say 1v? Should I consider pushing them to 25v for a while at the end?
-
What are you planning on using the caps for?. G-cap brand doesn't instill a lot of confidence.
-
They're reformed now. The idea of reforming is just to get them to the point where they won't blow up if you apply rated voltage. It doesn't have to return them to full spec. A current of less than a milliamp isn't going to cause a problem. It's not a good idea to reform multiple capacitors in parallel. You can't tell if one of them is going bad and pulling most of the current. In this case, it worked out okay because none of them are drawing much current.
I can't find anything about G-cap, but a Rubycon data sheet that I've got (USR series) shows leakage current specification as I = 3*sqrt(CV) where C=capacitance and V=rated voltage. Current is in ua. So, for your 100K uf, leakage would be ~5ua. If you had to reform the capacitor, it could take days to get to this level. Note that the equation might not really be appropriate for other brands or models.
A 25V rating means you could put 25V across the capacitor. It's not recommended, but you could. To cover that possibility, you might want to reform them to a slightly higher voltage than rated - maybe 30V.
-
Reforming takes quite some time, especially they were stored for long time.
An example the leakage chart result from a NOS polymer cap OSCON at it's rated voltage, at the chart reading, 1 volt equivalent to -> Volt / (10M Ohm DMM internal resistance) -> 0.1uA. Other details -> HERE (https://www.eevblog.com/forum/beginners/is-there-any-practical-or-quick-n-dirty-method-to-measure-capacitor-leakage/msg722332/#msg722332).
(https://www.eevblog.com/forum/beginners/is-there-any-practical-or-quick-n-dirty-method-to-measure-capacitor-leakage/?action=dlattach;attach=163731;image)
-
They're reformed now. The idea of reforming is just to get them to the point where they won't blow up if you apply rated voltage. It doesn't have to return them to full spec. A current of less than a milliamp isn't going to cause a problem. It's not a good idea to reform multiple capacitors in parallel. You can't tell if one of them is going bad and pulling most of the current. In this case, it worked out okay because none of them are drawing much current.
I can't find anything about G-cap, but a Rubycon data sheet that I've got (USR series) shows leakage current specification as I = 3*sqrt(CV) where C=capacitance and V=rated voltage. Current is in ua. So, for your 100K uf, leakage would be ~5ua. If you had to reform the capacitor, it could take days to get to this level. Note that the equation might not really be appropriate for other brands or models.
A 25V rating means you could put 25V across the capacitor. It's not recommended, but you could. To cover that possibility, you might want to reform them to a slightly higher voltage than rated - maybe 30V.
Thanks, I'll leave these as is for another day or so. They are exactly as shown so I don't think they're too old.
-
Reforming takes quite some time, especially they were stored for long time.
An example the leakage chart result from a NOS polymer cap OSCON at it's rated voltage, at the chart reading, 1 volt equivalent to -> Volt / (10M Ohm DMM internal resistance) -> 0.1uA. Other details -> HERE (https://www.eevblog.com/forum/beginners/is-there-any-practical-or-quick-n-dirty-method-to-measure-capacitor-leakage/msg722332/#msg722332).
Thanks, good thread - I will read.. Since they're on 5 pcb's, it'll be easy to narrow-down which pair has higher leakage in a day or two.
-
What are you planning on using the caps for?. G-cap brand doesn't instill a lot of confidence.
Price was OK. I see Luxon had a G type as well, so I'll probably trust them in pairs for 5 handy little LM338 1.5-15v charger/psu's.
-
It looks like you've reformed them.
For future reference you can connect them in series with small incandescent lamp (<1W). It should briefly flash while the capacitor charges, then the current should rapidly fall to a point when the lamp won't light. It can then be connected to the power supply for a day or so.
-
OK, they're reformed, but that Rubycon 5uA at 100,000 uF number nags me. Today 21v through 10K was down to 175ua for all 10 caps, so without discharge a 1k was quickly swapped-in and after 20 mins the charge rate now sits around 450uA. So with the 10K series voltage held at 19.25v (21v-1.75v), and now with 1K series they hold at 20.6v, but the charge rate has now tripled...
I'll maintain this *charging* up to 1 week, but this is now 45uA leakage per 10,000uF cap.. (~80 times the leakage of a single 100,000uF Rubycon). Since this is 10 no-name caps paralleled and not 1 Rubycon, what think ye? Am I expecting too much?
-
Yes, you probably are expecting too much. Grab a bunch of capacitor datasheets and see what they say about leakage. Maybe the Rubycon spec is unusually tight. Maybe it only applied when the capacitor was new.
You could try separating the capacitors and see how they compare with each other. Maybe one or two have higher leakage than the others. Maybe one of the boards has a dirty spot that's causing leakage.
Half a milliamp of leakage for 10 capacitors isn't a problem. Don't worry about it.
-
Separated them with an equally stepped decrease in charge currents, so they'll stay together again to float at 22v a few more days.
FWIW, the charge current has now dropped to 36uA per cap (360uA for all 10). AFAIK no need to add more. Thanks for the help!
-
i would suspect they are bad in the first place TBH, it takes years/decades of no charge for the oxide layer to deplete which lowers the internal resistance.
i wrote this short post which was featured on hackaday about the reforming process on some 40 year old electrolytics
http://dexterslab2013.blogspot.co.uk/2015/12/reforming-electrolytic-capacitors.html (http://dexterslab2013.blogspot.co.uk/2015/12/reforming-electrolytic-capacitors.html)