Offline flyback input capacitor ripple current

[schnell]

Hello,

I am doing an offline, isolated, flyback with:-

Vout = 36V
Vin = 85-265VAC
Primary Side DC Bus Electrolytic Capacitor = 68uF , 400V
(Panasonic EEUEE2G680)
Switching frequency = 58100KHz


DC Bus cap datasheet:

http://www.panasonic.com/industrial/components/pdf/ABA0000CE113.pdf

This cap has a 515mA ripple current rating at 120Hz …..and
-a 1470mA ripple current rating at 100KHz


From simulation, I know that the ripple current in this capacitor due to the 100Hz replenishment from the full-wave-bridge is 465mA.
(at maximum load and minimum Vin)

-this is just on the safe side of the 515mA ripple current rating of this capacitor at this low frequency.

…the thing is, the ripple current from the 58.1KHz switching stage is 429mA (AC RMS)


The thing is , how do I know whether or not I am violating the capacitors overall ripple current rating?

….as you can tell, at the high and low frequencies, I am within this capacitor’s ripple current rating, but surely, overall, I am above this capacitor’s ripple current rating?

The simulator gives the total overall ripple current in this primary DC Bus capacitor as 623mA (AC RMS)………

-So do you know whether or not I am subjecting this capacitor to too high ripple current?

The actual ripple current in this capacitor is obviously spread over a large spectrum of frequencies and the harmonic frequencies (from 100Hz to a few 100’s of KHz), and each individual current frequency will have a different dissipation factor……. In the face of such complexity, how can I calculate whether or not I am within this capacitor’s ripple current rating?

 

[DHambley]

You can model most electrolytic capacitors with a 4 or greater element model like, R1-C1-R2-C2. The first R1 is smaller and can take the 1.47Amp at 100kHz which you mentioned. If you are at 58kHz, the current is going "deeper" into the capacitor and heating up more than a 100kHz signal would. The second, R2, is larger. The physical construction of the cap makes it so. R1 will see the RMS current of both your waveforms, 632mA. R2 will see mostly just the 120Hz current, 465mA.
R2 is taking 90% of it's rating and 82% of its power and could be too hot if stress more. R1 is taking 43% of it's rating or, I^2*R is 18.5% of the power rating. I don't know the model for this particular cap but, I'm guessing that this waveform you mentioned is too much for that cap.
You can put the cap on an impedance analyzer to determine the model.

 

[itsmoked]

It all comes down to temperature rise in the cap. I'd probably design the board to take two caps or a larger cap that can handle 30 or 40% more ripple and the cap you're interested in. Run your circuit and monitor the cap's temperature and compute its rise. If the rise is too high with your selected cap you have a solder-in fall back. Meanwhile the rest of the design evaluation can proceed.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[DHambley]

Schnell,
Just curious but, is this power supply PFC or a simple rectified bridge?

 

[schnell]

its just a simple rectified bridge

 

[Warpspeed]

I am with Keith on this.
Temperature rise is really what sets ripple current limits, and capacitor life, and the expected ambient operating conditions all play a part in what may be acceptable.

There are no hard and fast rules, but measure the capacitor temperature rise, and just use your own best judgment.

 

[MagicSmoker]

Hey schnell - the simple answer, is...don't rely on the elko to handle the high frequency ripple; add a small polypropylene film capacitor in parallel, preferably as close as possible to the switch/transformer.

The ripple current rating for an elko is always given at a particular ambient temperature, usually the highest allowed for the electrolyte (e.g. - 85C or 105C). The core temperature will typically be 5C higher than ambient when passing the rated ripple current, btw, but this is mostly of academic importance.

The upshot of this is that you can handle more ripple current at a lower ambient temperature, but whether you *ought* to or not is a different story, since the reflected ripple voltage is, of course, the ripple current times the ESR. Enter the film capacitor... the elko, with its much higher capacitance (and ESR) will easily handle the ripple from rectifying the AC mains; the film capacitor, with its vanishingly low ESR but also low capacitance, will easily handle the ripple from the flyback converter's operation, keeping it away from the elko.

Finally, what usually dictates the size of the reservoir elko in an offline SMPS is the required holdup time, or how long until regulation is lost. The usual minimum requirement is 10ms, or half a cycle at 50Hz - you may want more to allow for an orderly shutdown or to ride out minor dips and blips on the mains.