Electrolytic capacitors in power supplies that are stored for long periods

[treez]

Hello,
A company that I know of are building 200W Offline SMPS's. They have no customers as yet, and so are just storing the SMPS's in a big warehouse. Each SMPS has four 400V 47uF electrolytics in them.
These are DC Bus capacitors which come after the PFC stage.
After how long will these stored power supplies be rendered useless due to age assisted deformation of the electrolytics?

 

[KlausST]

Hi,

Afaik the lifetime of electrolytic capacitors depend mainly ot their quality and temperature.
So its hard to say how long they are good..

Klaus

 

[BradtheRad]

According to an article I saw in a trade journal several years ago, electrolytics need to have their chemical structure re-formed after they have been stored several years.

If this is not done they may be ruined when voltage is applied suddenly, due to unimpeded current flow.

The article recommended that the capacitor be re-formed before it is put into serviced, by installing a high resistance in series, so that it charges very slowly, until operating voltage is reached.

 

[FvM]

Epcos application literature says, after two years there may be a problem with circuits that don't tolerate higher leakage currents (See paragraph 3.7.5 Reforming and 3.7.6). A power supply circuit usually would. They don't discuss possible damage when applying rated voltage without current limit. I also won't expect it with modern capacitors and moderate storage times (e.g. < 10 or 20 years).

B.t.w., you asked the same question before https://www.edaboard.com/threads/304746/

 

[treez]

thanks, yes I wrote before but that time it was about loose capacitors, and not ones soldered into power supplies and stored in there.
Anyway, FvM, on page 18 of your excellent article it states that after 2 years of storage, the power supply should be powered up and run for an hour, before being put back into storage.
On page 18, it also states that after 2 years of storage, the leakage current in the capacitor can be 100 times as much. Considering the 47uF , 400V nichicon capacitor here....
http://nichicon-us.com/english/products/pdfs/e-vr.pdf

..that means that 85mA of leakage could flow...and since this cap is for use on a 400V DC bus, that's 34W of dissipation!

...looks to me like building SMPS's for storage is not a good idea if they have electrolytics in them...do you agree?

 

[D.A.(Tony)Stewart]

Aluminum Electrolytic capacitors have always had an aging effect on leakage after long storage which can be eliminated to prevent self-heating on first use with high voltage by using a large resistor in series to monitor current decay before being used.

Back in the days when we used 60V 150k uF caps for stereo amps and bought these mainframe computer grade caps from surplus houses, this was how we ensured they wouldn't blow up.

I used 1MOhm to 100K in series with rated voltage on the ground side to monitor current. The leakage would reduce back to normal levels within a few hours. These days depending on series RC time constant and size, I would expect a few minutes is possible to restore low leakage performance.

I always wondered why they call direct AC to DC power supplies as "offline" when they only run when ONLINE.

I would advise the same to any company storing "offline" units to test them with a high series resistance in a non-destructive but perhaps, non-functional test to ensure this does not affect reliability from self-heating. ( Start with an AQL type audit before deciding to retest all units or none of the rest.

 

[treez]

The thing is SunnySkyGuy, is that there is no series resistor within the power supply circuit (as you'd expect)....so it wouldn't be possible to do a reforming test with a series resistor.
This is why I am asking again about this case of electrolytics that are inside smps's that are being stored. It seems to me that this is bad news?
SMPS's should not be stored for more than 2 years..surely?....and also, how would you know that the storage people are keeping the aircon on and keeping the storage temperature below 40 degC?

 

[FvM]

I didn't experience power supply capacitors being damaged when powering them after a long time. If we see factor 100 leakage currents, it starts to decay within seconds and usually doesn't seem to overheat the capacitor. That's also my reading of the Epcos application note. You can rather expect problems when the capacitor is charged with low current limit, e.g. in a timing circuit.

I can't assure you that power supply filter capacitors will never fail after years of storage, I just say it seems unlikely to me and I never saw it happen.

 

[treez]

for me page 18 of the Epcos app note says it all....they say that if stored for over two years the power supply would need powering up, then put back into storage......so if your storage people forget to do this , then you are already in violation of the application note.....to me this I saying...making thousands of PSU units and storing them is a bad idea.
Epcos don't say that the factor 100 leakage currents quickly die out..Epcos say they go on for minutes...minutes at 34W is going to blow our 400V caps.

 

[pebe]

You are assuming that the power used is 85mA x 400V, but it is not. If 85mA flowed into the cap its voltage would rise, but it doesn't - it stays substantially constant.

Part of the energy going into the cap is expended as heat in the electrolyte (ESR), and part is changed to chemical energy when the anodic film is being reformed. IIRC the leakage current decays exponentially.

 

[KlausST]

Hi,

I didn't experience power supply capacitors being damaged when powering them after a long time.

I know about damaged capacitors. But mostely the were very old from tube radios. I think this is a problem of drying out the electrolyte.

I hope nowadays capacitors are better. So besides the increased leakage current during the frist powering up, i hope they have a lifetime/storage time of more than 10 years...


This is my hope, but not my knowledge...

Klaus

 

[treez]

You are assuming that the power used is 85mA x 400V, but it is not. If 85mA flowed into the cap its voltage would rise, but it doesn't - it stays substantially constant.

...sorry but leakage current doesn't raise the voltage of the capacitor.
In this case the voltage across the caps is regulated to 400V by the PFC controller.

FvM's article on page 18 makes it very clear, that for a few minutes after 2 years of storage, you could see 100x the leakage current flowing in the caps....for us, that means Kaboom!

 

[pebe]

...sorry but leakage current doesn't raise the voltage of the capacitor......

That's precisely what I said!

I think the power during reforming is 85mA x 85mA x ESR, which is very little.

 

[treez]

the 85mA of leakage flows through the 400V.

 

[FvM]

The calculated 34 W losses are correct. The question is however, how fast does the excessive leakage current decay? My experience so far is that it decays very fast, I expect that the capacitors will be reformed before it heats up critically.

Unfortunately I'm not sure if it happens every single case. A practical conclusion could be to perform a shipment test with power supplies that have been stored two years and longer.

 

[D.A.(Tony)Stewart]

The thing is SunnySkyGuy, is that there is no series resistor within the power supply circuit (as you'd expect).


This is what PANASONIC says

6. Long Term Storage
(1) Leakage current of a capacitor increases with long storage times. The aluminum oxide film deteriorates as a function of temperature and time.
If used without reconditioning, an abnormally high current will be required to restore the oxide film.
This surge current could cause the circuit or the capacitor to fail.
Storage period is one year. When storage period is over 12 months, a capacitor should be reconditioned by applying the rated voltage in series with a 1000 Ω current limiting resistor for a time period of 30 minutes.

For storage condition, keep room temperature (5 °C to 35 °C) and humidity (45% to 85%) where direct sunshine doesn’t reach.

I would simply choose a suitable 1K power series resistor to limit the current on the primary AC side which normally rectifies into the large voltage electrolytics

 

[pebe]

the 85mA of leakage flows through the 400V.

Not so. If the applied voltage is equal to the forming voltage then the leakage current will be zero. That is true whether the capacitor is rated at 50V or 350V. In either case increasing the applied voltage will increase the leakage current. So the increase in leakage current is a function of the voltage differential between applied and formed voltage – not on the voltage rating of the cap.

Many years ago I worked in the lab of the now defunct Plessey company which manufactured high quality electrolytics. Part of the accelerated life tests on sample caps was to alternately freeze them at –10degC and bake them at 70degC for 1 week each, with no polarising voltage applied. After each cycle the caps were left at room temperature to recover, then tested for capacitance, power factor at 50Hz, and leakage.

Freezing had little effect on leakage current but the high temperature did. Part of the anodising film was absorbed into the electrolyte and when working voltage was reapplied, the leakage current had increased. The current was proportional to the surface area of the anode – ie. its capacity.

Part of the test was to time how long it took the leakage current to normalise. I cannot recall how long the average time was, but it was required to be within 3 minutes. I never knew of one that overheated as a result of it.

 

[FvM]

Not so. If the applied voltage is equal to the forming voltage then the leakage current will be zero. That is true whether the capacitor is rated at 50V or 350V. In either case increasing the applied voltage will increase the leakage current. So the increase in leakage current is a function of the voltage differential between applied and formed voltage – not on the voltage rating of the cap.

Whatever leakage current has been specified in your previous work, leakage currents in capacitor datasheets are currents flowing into the capacitor terminals and the loss power is applied voltage times actual leakage current. Leakage currents are always specified as worst case values and it's difficult to tell what typical values are. Defining realistic values for initial leakage current after long periods of voltage-free storage seems even more complicated.

Referring to the quoted Panasonic recommendation:

If used without reconditioning, an abnormally high current will be required to restore the oxide film. This surge current could cause the circuit or the capacitor to fail.

Storage period is one year. When storage period is over 12 months, a capacitor should be reconditioned by applying the rated voltage in series with a 1000 Ω current limiting resistor for a time period of 30 minutes.

Who believes that the recommendation is obeyed e.g. in the typical life-cycle of a SMPS module:
- voltage applied for the last time in final production test
- stocked at the manufacture for some time
- stocked at distributors site
- purchased and stocked again by the instrument integrator
- assembled and undergoing instrument test

What's the expectable interval between first and last step? Does the SMPS provide a means for "soft" capacitor reforming?

 

[treez]

I would simply choose a suitable 1K power series resistor to limit the current on the primary AC side which normally rectifies into the large voltage electrolytics

Sunnyskyguy.....I see what you mean,but that's quite a fiddle for the storage people to have to rig a jig up wth a 1k resistor, connectors either end of it, and run it for the allotted time.

I never knew of one that overheated as a result of it.

..I understand, but that test didn't involve testing of smps's with electrolytics in them that have been stored for 2 years or more...which is the subject here....though it is of interest.

Sunnyskyguy...your Panasonic quote says it all....how 'lytics cannot be stored for over 1 year without reforming activity being needed.....

I think we have concluded that we cannot be sure that 'lytics stored for over 2 years will not fail if powered up without the reforming process being done.

I am seeking the article linked in by Sunnyskyguy....please tell of the link if you have it.

 

[pebe]

... leakage currents in capacitor datasheets are currents flowing into the capacitor terminals and the loss power is applied voltage times actual leakage current.

It is impossible for leakage currents to be flowing into the capacitor. The voltage across the capacitor would rise in accordance with V = I x T / C. Work it out.