switching supply inductor sizing

[Eugeneo]

Is there a simple formula for sizing switching supply inductors (step up and down) besides looking at the datasheets and provided tables? Of course it would have to include inductance resistance current frequency.

I'm using a 1.5-3.1v to 3.8 volt step up / regulator DC/DC converter (LM2621MM). Everything works find until I hit about 350 ma. At this point the voltage regulation drops a few tenths of a volt. I'm sure it's an inductor sizing problem since I've changed everything else. It's supposed to be good to 500 ma

[PCM programmer]

My suggestion is to use National's online Webench power supply
designer. Use the chip they specify, and use the parts they specify.
I tried to use your parameters with LM2621, and they didn't like it
for some reason. They want to use the LM2623 instead.
http://www.national.com/appinfo/webench/scripts/my_webench.cgi/

[Ttelmah]

The 'reason', is that as the incoming voltage falls, the effects of the internal FET resistance, and some internal drops become more significant, taking the unit outside it's design specs. The peak inductor current also shoots up. If you look at the graphs for 'efficiency versus load current', with an input voltage of 1.8v, and an output of 3.3v, the efficiency is already down to below 80%, with a 0.4A load. At 0.5A, the efficiency shoots down to about 75%, and with lower input voltages this roll off point is hit even sooner. With the specified figures, given the drops, the unit will hit the peak fet current of 2.85A, at about 400mA, and it only needs the incoming voltage to be dropping a little, for this to be happening at about 350mA. The NS design program has a good model of the unit, and stops trying, when the efficiency drops below perhaps 70%, hence it switches to the 2623.

Best Wishes