What capacitors are proper to use for input of AC DC converter?

Question

I am trying to build an 120 AC to 15 DC @ 200 mA converter using a Power Integrations LNK3206P. Using the full bridge rectifier setup on Table 10 from application note AN-70, Cin1 and Cin2 should each be >2uF/Wout * 3W = 6uF rated at 400V.

It appears that at least Cin2 should be of a low ESR to reduce the ripple voltage on the input of the converter and reduce the ripple current on the electrolytic capacitor Cin1.

Using ceramic capacitors such as this 2220 250V 2.2uF X7T, after de-rating due to the ~170 DC bias of the rectified 120 AC as seen in the right hand side middle graph of its details page from TDK, it becomes approximately 0.8uF.

This would mean that I would need at least 8 if these capacitors in parallel, for a total cost of about $30 USD.

Looking into other options such as ceramic capacitors with higher voltage ratings still results an an unreasonably high total cost. Similar thing with film capacitors.

With AC DC SMPS being so common and cheap, this does not make sense. What am I not thinking about or considering.

Answer 1

Note the symbols on that schematic: both Cin1 and Cin2 are electrolytic caps. So, even for high voltage ones, you can get them at reasonable price (10uF-100uF, 400V electrolytic caps are fairly common in SMPS).

Note also that in that application note there is a photo where it appears that they actually use electrolytic caps: pag.10, figure 3, the caps on the right appear to be 4.7uF,400V electrolytic caps.

Answer 2

I think what people usually do is specify the capacitors to be low enough so that the worst case ripple of the output of the full-bridge doesn't come close to bringing the 170V peak down to 15V. Since 15V is pretty low, I'm sure you could get away with much less capacitance, since the buck converter will regulate down your voltage the 15V output.

I would make sure to check if your buck converter has good line ripple attenuation; if you have something like 100V of ripple on your 170V (so the minimum voltage is 70V), the feedback might not have a low enough pole to totally attenuate that and so you might get a few 100mV of ripple on your output. Whatever you can live with will be what sets your capacitance.

I would simulate the circuit, since this catalogue has an inductor in the filter as well and your design will probably have a large ripple.