Switching power supply question.

[kvresto]

Hi.

I’ve been following TI’s design examples, and plugging in numbers into WEBENCH (Vin=9V to 20V, and Vout=15V@1A), and have settled on a SEPIC design using the LM3478, but I have a couple of question I was hoping to get the answers to.

The data sheet writes that since the presence of an inductor at the input of the SEPIC converter, the input current waveform is continuous and triangular. WEBENCH simulation seems to indicate that the input voltage does in fact have a couple of millivolts Vripple pp. My question is  can I filter this so as not to interfere with other cct’s connected to the input, in this case the battery on a vehicle. Will a ferrite and cap do the job? or other? Seems a pissy amount of ripple??

The data sheet for the LM3478 also recommends for applications that require very low output voltage ripple, a second stage LC filter often is a good solution. Apart from sizing this correctly, my question is the frequency of the ripple I will be filtering against is obviously the regulators switching frequency, but is there a trap here? or is that it?

Thanks
kv.

[T3sl4co1l]

How does Webench simulate the source?  An ideal voltage source allows zero ripple by definition.

A typical automotive equivalent (LISN: line impedance stabilization network) is an ideal source, followed by an inductor terminated to 50 ohms.  The cutoff frequency is modest (I forget if it's 500kHz or a bit over 1MHz?).  This may yield quite a bit more 'spiky' ripple than you are expecting.

Yes, the usual solution is additional LC filtering.  Pick a cutoff frequency more than a few times below the switching frequency, and an impedance comparable to the ESR spec of your capacitors, so the filter is well damped by itself.

Tim

[kvresto]

Hi T3sl4co1l, how do you think I should go about filtering the front end of my power supply in general, or is it worth the trouble?

kv