Why is charger over-ripple-current'ing its electrolytic capacitors?

[treez]

Hello,
We have a 3kW Electric car charger which we bought. It often unfortunately shuts down when in use, for unknown reasons.
We opened it up to have a look, and noticed that it is a Half Bridge LLC resonant converter running at 100khz, preceeded by a 3kw Boost PFC.

The DC link capacitance between the PFC output and LLC input is just 810uF. This is comprised of three pieces of 270uF, 450V Electrolytic capacitors (Rubycon VXH series). These capacitors only have 1.68 Amps of ripple current rating at 120Hz. (so that’s a total ripple rating of just 5.4A RMS.

The ripple from the PFC stage alone is 8.1 Amps RMS. When the LLC is factored in, the total ripple current in the DC link capacitance is 9.1 Amps RMS.

= [sqrt (ripple_boost^2 + ripple_LLC^2)]

Why have the designers of this charger so badly undersized the DC link capacitance?

VXH series capacitor datasheet:
https://www.rubycon.co.jp/en/catalog/e_pdfs/aluminum/e_VXH.pdf

 

[Easy peasy]

Easy - they have no concept of the real ripple current at full power - or assume full power only happens for 5% of the time - or pruchasing put in the cheapest ones they could source as the Epcos ones were way too expensive...

We have seen loads of hi power converters (mainly forklift chargers) where the electro's are seriously under-rated and last for perhaps 1 - 2 years before failing to low/no capacitance, splitting open due to internal temp...

 

[treez]

Thanks, our boss toys with the idea that its ok though , and he likes the cheapness, he says that because these electrolytics have some gap pad on top of them, which connects up to the metal enclosure, that this makes it ok.

- - - Updated - - -

Another Engineer tells me that he used to work for some company that regularly used to use over-ripple-currented electrolytics in base station power supplies that were in use 24hrs per day, and he reckons they were ok.
I told them that if you go above the recommended ripple current rating then none of the lifetime equations (if supplied) are applicable. But they wont listen to this.

 

[dick_freebird]

Well, certainly the manufacturer's warranty is out the
window (if there was any). Field failures are the engineer's
professor. Too bad management is only auditing the class.

 

[D.A.(Tony)Stewart]

VXH is rated for Load Life= 5000 hours. @105℃ which is least 3x that of cheap light bulbs (1500 h).

I guess they were calculating it would last at least the warranty period.

Was it interleaved PWM controlled? or constant duty cycle single phase? or?
Did you measure that accurately? or guess.

Also ripple rating increases with switch frequency somewhat to 1.5x

 

[Easy peasy]

yep, you can assume 20 C between the middle of the cap and the can (no airflow), if running at rated current, so for 105 C hot spot in the middle, the can must be 85 or less.

For every 10 C below 85 (surface) you run them the life time will double, so short term overloads are acceptable, however if they run up near 70 -85C surface all the time 5000 hrs is all you can expect, 7 months.

 

[treez]

Was it interleaved PWM controlled? or constant duty cycle single phase? or?
Did you measure that accurately? or guess.

The PFC is two PFC's, (each at 45khz) one runs for a mains half cycle of 10ms, then shuts off, then the other one runs for the next mains half cycle, etc etc.
The figure of 5.4A is from calculation and simulation to check.

The input ripple from the 3KW , 100khz LLC alone is 4.17A RMS.

The ripple at 50Hz from the PFC stage is 7.82 AC RMS into the output cap.
As you know, the total ripple is the Square root of the sum of the squares of these ripples.

 

[D.A.(Tony)Stewart]

well at least interleaved doubles lifetime.

Yes and RMS of a pulse is proportional to square root of duty cycle.

 

[Easy peasy]

the interleaved boosters will be running into the caps, I don't quite see how the current into the caps is reduced? the 100/120 Hz current will be the same - there will be no reduction in ripple current (of the boosters) if there are two sections running on opposite mains half cycles... then there is the ripple drawn by the converter as well...

 

[D.A.(Tony)Stewart]

By continuously supplying current in alternate phases, the RMS ripple is cut by 50%, hence I²ESR reduced, temperature reduced and MTBF increased.

 

[Easy peasy]

sorry, the boosters aren't interleaved in the usual meaning, they operate on alternate half cycles (of mains), hence no ripple reduction...

 

[treez]

Thanks, though in this case the two Parallel PFC's are not continuously supplying current together simultaneously....one supplies it all for 10ms, then switches off, then the other one does the next 10ms...etc etc
Easy Peasy I think we both replied to post #10 simultaneously