I was assuming you'd already have an inverter for 120v AC. Wouldn't it be less loss to convert at 400v, step that down to both other voltages, and convert back

I was assuming you'd already have an inverter for 120v AC. Wouldn't it be less loss to convert at 400v, step that down to both other voltages, and convert back to DC at 40v than to have a buck convertor for 40v and an inverter for 120v separately? Of course then you'd draw fewer amps to just go ahead and plug 120v tools in and not convert back. If you're just skipping the NEMA wall outlet and going to 40v DC that's different. You might want too go with a non-isolated switched mode there for efficiency. Still, an inverter above 20kHz and transformer at high frequency in an RCC or resonant forward configuration shouldn't have terrible losses and the transformers at high frequencies can be small and multiple for different output voltages if necessary. I think the design goals regarding space, weight, cost, flexibility, and reliability really come into play here. About that buck convertor... I'm just a hobbyist when it comes to power supplies but since a buck convertor tends to use semiconductors I'm thinking a 400v input might be an unusual design. The highest input voltage rating I find for a "high voltage" DC buck convertor is around 140 volts. If you tap the battery a quarter of the way through you could do that, but then you're loading that portion of the cells more than the rest. I can find 700 volt "AC bucks", but there again you're talking AC power. Everything I can find that bills itself as a 400v capable down-stepping DC to DC buck convertor has an isolated portion labelled, which means it in fact has at least a 1:1 transformer somewhere connected to the buck. If you're talking about flipping mechanical relays quickly instead of transistors and diodes, you're going to have a lot of noise, slop, and reliability issues aren't you?