Battery-powered linear-regulator-filted variable switchmode powersupply

[Recyclojunk64]

G'day

I'm planning on building a portable (as in battery powered) variable power supply. I've got several kilograms of unwanted lithium ion batteries I can use for it so that shouldn't be a problem. Output power is probably going to be in the order of several hundred watts or so. Maybe around 32 to 48 volts max output.

I want it to have a low-noise output that's well regulated, so that probably rules out switchmode powersupplys somewhat. However, I didn't want to just use a horribly inefficient linear supply, that would just waste heaps of power and require a very large heatsink.

Would it be possible to use a regulated switchmode buck step-down converter (battery voltage will always be higher than output voltage), set to a couple of volts above the desired output voltage, a few LC filter stages, and then a linear regulator on the end to produce a stable, low-noise output? Probably some large mosfets for the both switchmode and the linear regulator.

Also, any ideas about a charging solution for maybe up to 24 cells [parallel groups of Li-ion] in series? I don't care about the constant-voltage stage of charging (since it only gives 10% or so of extra charge). Perhaps maybe an isolated mosfet-based constant-current regulator for each cell or something?

[Psi]

yes, this is totally possible and called a tracking pre-regulator

You should check out the eevblog videos on daves usb powersupply operation.
It does exactly this but at lower power levels and even has lithium battery charging as well.

[peter.mitchell]

With regards to the batteries, you're looking at least 15 cells in series for 48v, because they vary between 4.2v at full charge and say... 3.4v at cut-off, plus some voltage sag under load, and some voltage drop in your non 100% efficient regulator.

Charging that many cells in series without adequate control circuitry is a significant financial loss waiting to happen.

If you went for 32v max you could get away with 8S3P maybe, and there are some pretty neat "hobby" chargers designed specifically for this. Additionally, if you intend to use raw cells or unprotected packs, you will need to make a very good cut-off and monitoring circuit; otherwise you will quickly destroy cells and possibly cause significant financial loss aswell.

http://www.hobbyking.com/hobbyking/store/__6609__iCharger_1010Bplus_300W_10s_Balance_Charger.html This is a "hobby" charger capable of safely charging and balancing 10 cells. It uses a main charging connector, connecting to the ends of the series string, plus a wire for each cells terminals this allows it to "shunt" individual cells.

[Recyclojunk64]

Thanks for the input. I'd probably go with a shunt regulator across each cell then. Good idea, never thought of that. I guess a few One-Hung-Lo branded 4.2 volt zener diodes aren't gonna cut it. What about one based on a tl431 adjustable voltage regulator and probably a low-input-voltage-threshold mosfet for the switching part. Should'nt be too hard to build 20 or so of them out of junk parts. Constant current charging should be a piece of cake to implement across the lot then.

Actually the reverse diode across the mosfet could be a problem though, right? Is the only way around it adding a diode in series? I don't really want to have to use BJTs (cause they're less common in suitable size packages).
Edit: Of course not. That's what I get for making a post while I'm half asleep.

However, I was wondering whether it might be possible to instead use the battery packs as they are currently (they are infact old laptop batteries) and somehow interface with the internal charging circuitry. They are, BTW all identical model thin (as in not containing 18650 cylindrical cells). Yeah, they're only 11 volts each, so it'll need to be a boost converter if they were all in parallel. Don't think they'ed work in series because of the floating point ground between battery packs i'm guessing. Is this a ridiculous idea or is there a chance it may work better than a DIY charging solution such as a shunt circuit on each cell?