It would, indeed be helpful if we had a more complete description of the battery circuit. From what the O/P has provided, I guess we're allowed to assume a large-capacity (e.g. 1^560 AH) ideal 12.0000000... volt battery that is going to be replaced and charged separately when needed. Since the total current to be drawn from the battery is to be 330 mA, that won't be necessary often. I think those modules suggested in Erik's post may be just a bit large, however. Further, if they're to be applied in accordance with Erik's suggestion, i.e. a battery the voltage of which rises as high as 14.4 volts and is allowed to drop as low as 6 volts, then a SEPIC switcher is not required, and these modules do not, I believe, fit that model. However, the majority of them won't work from 12 volts.

They cost and weigh quite a bit more than the somewhat under $2 that a simple circuit capable of adequate power and substantially lower noise would cost, particularly since there would have to be two of them.

Hopefully about 100 mA on each of the 5-volt supplies would do the job, as that can easily be managed with a 1-watt zener with a generous factor of safety. The 3V3 could similarly be controlled with a zener capacity of 1/2 watt. If one (a) were able to tolerate the noise from a switcher, and (b) had to deal with the widely fluctuating input supply voltage as Erik suggests, then that battery could be pregeulated with a SEPIC regulator costing perhaps a dollar or two, to the somewhat more ideal voltage of 10 volts, in which case a rail-splitter would provide the necessary pseudo-ground and the complementary 5-volt outputs. Somehow, that makes more sense to me ... I could be wrong, of course.

RE