High-Power Split Rail DC-DC Options

[Evan.Cornell]

I have power source defined as 36-48V. I need to produce +-60V rails from this source, each rail capable of providing ~2A current.

My initial thoughts are a standard boost controller (external FET power stage) for the positive rail (LM5122 perhaps), and a buck-boost controller (external FET power stage) for the negative rail (LTC3863 perhaps).

F_sw must be >500kHz and disabling of light-load pulse skipping mode is required (so FPWM for LM5122, and Forced Continuous mode for LTC3863).

Any clever suggestions beyond this proposed topology, or am I already on the right track?

[T3sl4co1l]

Minimal control method is, use a coupled inductor to do combined SEPIC and Cuk in one fell swoop.  Gives good matching and cross-regulation (doesn't depend on load currents so much; the regulated output still needs enough minimum load to keep the other side going).

Downside is, there aren't many coupled inductors available above 1515 sized SMT shielded-bobbin style parts; you'd probably need more than a few of those to get to 240W without melting anything.  And the bigger ones that are available, tend to be more lossy, or poorly coupled (e.g. two windings on opposite sides of a toroid).  So it may need a custom design.  I think some 3637 PQ style parts are available though, which could do it.

Mind, good luck finding 3+ winding inductors -- but you can always use a pair of duals and wire the primaries in parallel.  Use double the design inductance, since they reduce in parallel (two uncoupled inductors in parallel).

Otherwise, yeah, boost and invert is about as good as anything.  Or multiple flybacks, which is basically the SEPIC-Cuk thing but worse (but it can be isolated, if need be).  Or forwards.

Tim

[Evan.Cornell]

That reminded me of another constraint - avoiding custom magnetics. This isn't a high-volume thing, and definitely would prefer to avoid that route. Boost + buck-boost it is!