TOP268EG Output Cap Value from PI-Expert
I just designed an 87W dual-output (40V /2.0A-pk /1.6A cont., 12V /0.6A) flyback design with PI-Expert for the TOP268EG. Operating frequency = 132kHz. The 40V output cap was specified as 470uF /100V, with a recommended obsolete UCC capacitor (KMF series) rated for 1.79Arms ripple /100kHz, and 0.06 ohms, 105C /5k hours, 100V. The design report says there is 1.45A of ripple with an estimated lifetime of 143k hours. This does not sound right; typically caps should be specified to be well beyond the maximum expected ripple, and not depend on the temp derating current multiplier to get longer life. I usually specify ripple rating at least 3x maximum continuous DC output current regardless of the cap rated temp or lifetime, as well as use the largest physical cap volume possible (to be sure the cap really can survive). This means I would need a capacitor rated at approx. 5.4A /100kHz, and that can only be achieved by a much larger capacitance, essentially multiple low impedance caps in parallel.
Can I get some feedback from PI as to the correctness of that 470uF value cap hanging directly on the diode output, eating all that ripple at maximum load and surviving.
Tod
Comments
I'm very aware of temperature and electrolytic cap lifetime issues, particularly with heating effects from ripple current combined with ambient temps. What I do not understand is how the PI-Expert SW stuck a single, rather small 470uF cap on the output of a 40V /2A flyback rectifier for the primary filter. The cap uF value may be based on it's ESR which is what really affects the output voltage ripple more than the actual capacitance at the 66kHz ripple frequency. However the ripple rating of only 1.78A /100kHz /105C is a bit low for such output current. If I get the ripple rating way up to around 3A or so with multiple caps, that will drastically drop the ESR the output sees, and if the ripple voltage flattens out that may affect the loop response and regulation.
I would kindly appreciate an explanation from a PI Engineer.
You don't want to hear this, maybe....
Likely the cap is specified because it works.
The life I get from the datasheet data and one life equation is very long... comparable to the number given... so long that exact numbers are very suspect. It's a 105C part used at possible 50 or 60C ambient, and below the 105C ripple rating. Long expected life is a given.
ESR comes out to something around 0.22 ohms, giving a half watt dissipation and maybe 15-20C rise.
If you need some special low ripple, and consequent much lower ESR, then you can use more capacitors if you like. But for life it does not seem to be needed. You can verify this for yourself.
Hi,
Thanks for contacting PI and joining the discussion.
Yes, the capacitor life time does really depend on the temperature of the capacitor. So it is the factor of the application ambient temperature, and the temperature rise of the capacitor. And the exact life time calculation for Aluminum capacitor is also variable with different manufactures.
The PI Expert software will help the design choose some reference capacitor, and make sure it will functionally correctly working. But the last confirmation will be based on the real test results, calcualtion, and the applicaiton conditions.
Thanks
Thanks Terry, I'm familiar with cap lifetime vs, ripple current and temperature. I will boost capacitance from 470uF x1 to 540uF (3x180uF) to get better ripple handing to my satisfaction. However, can you tell me: Did PI-Expert software pick out that capacitor partly based on it's impedance? The part is a UCC KMF100VB47M16X31LL, rated for 1.79A /105C /5k hours, and Z = 0.059 ohms /100kHz. I just need to know from PI what to expect as far as the Z /ESR value of the cap as it pertains to needing a certain amount of ripple voltage in the FB path.
Tod
Not a PI employee here....
But the life of a capacitor is strongly dependent on the temperature. There is a fairly standard equation for determining the expected life, findable at several capacitor manufacturer's sites. With lower than max spec temperatures, the life is greatly extended, OR the ripple current can be more than listed. It is not a linear relation.
The ripple current is not a fixed number like a fuse rating. It is a rating at a particular ambient temperature, which takes into account the I^2*R heating in the real resistance of the capacitor. If you demand a capacitor, or bank, with a rating several times the max temp current rating, *and* operate at a considerably lower temp, you might get near "shelf-life" operational lifetimes.
So unless you put in your temperatures, you may get wrong results, eitehr optimistic, or pessimistic, depending on the default. Apparently that is a 105C capacitor, so at considerably lower temps it would accept more ripple current than the 105C spec current, while giving the same life. Or, it would give a much extended life at the spec (or less as in your case) ripple current. Without running numbers I don't know the actual life extension.
This does not work the same way for film capacitors.