Build your own 1 phase to 3 phase rotary converter. You will have your true real 3 phase sine wave, but, of course, you will hear an AC motor spinning all the time: Note that other youtube videos include schematics and tests and what to avoid...
A rotary converter
would certainly work, however, for the power level that the OP needs it is going to have to be one seriously big honkin' motor, especially to handle the horrible power factor that is likely to be presented by a bunch of capacitor-input rectifiers in all those power supplies without the wave tops getting so smashed that it looks more like a square wave.
Edit: After more thought, I don't believe a rotary converter would ever be able to be made to work correctly in this case, especially since the loads are not static. There would great imbalance between the phases to begin with that would be unlikely to be able to be compensated for with the capacitor arrangement, certainly not always stable with varying loads on each phase. I do believe a motor-generator set would work, if appropriately sized, but I think the single-motor converter is out.That will not fulfill his "
low noise, somewhat sinusoidal" requirement.
The motor will also make some interesting noises with power being drawn only at the peaks of the wave, especially when the phases are going to be randomly imbalanced and the imbalance is going to be changing during operation.
Remember, we're talking about needing about 20 amps per phase, all being drawn through DC supplies with (presumably non-power factor correcting) rectifier inputs. I expect the peak currents to be very high, and only at the tops of the waves and that is probably why it has a 30 amp supply, even though the
average power stays down below 20 amps.
<story mode on>My first appreciation of just how much power factor actually meant in practice was a similar issue 15 or 20 years ago that I had just really never thought through, even though I had built dozens of different and all manner of power supplies and such before in my lifetime. It took sticking my scope on there to see what was going on before the light bulb above my head finally came on.
We had installed a (for the time) fancy new inverter/charger system at a friend's off-grid cabin so there could be power without having a noisy generator running. (A Trace Engineering DR series with 6 x GC2 batteries, IIRC). The original generator that was at the cabin was the Honda (EX?)3500 workhorse that had been there for years. The inverter supplier had indicated to my friend's dad that charging would be slow with such a generator, which it was, so he had already planned to buy a larger unit.
The next spring, they took up a new 6500 Honda but, of course, it didn't charge any faster either. (Humorously enough, I calculated later that the old one was actually faster, and they are even in the example charge rates for various generators in the Trace manual, though Trace didn't do a very good job of explaining why, power factor, etc. and basically just said "you will need a much larger generator than you think") The cabin owners were a bit dismayed, of course, but knew I would be coming out in the summer anyway.
At Trace's suggestion, I brought a nice big motor-run capacitor and wired it right across the generator output to help with the PF. I also parallelled in a second 10ga cable to the inverter/charger from the generator shed to reduce the impedance of the feed line. I had even brought a little PC-based scope with me (fortuitous choice) just in case and did various basic tests before I finally built a danger-divider using some resistors I happened to have in my laptop bag and put the scope right on the mains (scope and laptop floating, just in case I did something stupid) to see what was really going on. When I saw just how bad the waveform really was, I knew exactly what the problem was. Well, I already knew what the problem was but not the
magnitude of the problem! The charger was only drawing current from the part of the wave that was above the equivalent battery level since it was a smart, but relatively standard, non-PFC charger circuit. The average voltage was being held correctly by the poor Honda but there is just
no way it could supply all that power just on the peaks. I knew it would be a bit clipped, that was expected, but I wasn't expecting the wave to me
THAT mashed.
The next morning I was sitting on the steps of the generator shed, pondering the manual for the Honda when I came across the schematic and another lightbulb-over-the-head moment struck me. Eureka! Another friend (who is an electrical engineer) stumbled out that way, probably to re-tap the keg for a breakfast brew (we're at the fishing cabin remember, don't judge
) and I said, "Hmmm, Dös, take a look at this, what if I take this winding, flip it over and jam it on this one, that should work, right?" He basically said, "Hmm, yeah, that should work, just don't get it backwards and burn out the windings. LOL"
The 6500 generator has two separate windings which make up the two sides of the 240 volt output. The 120 plugs are just distributed half on each side, so in effect our charger was only running on a 3250 watt generator winding. The old 3500 Honda had a
switch for 120/240 volt which used both halves when in 120 mode to actually supply more current than the 6500 could.
Anyway, I opened up the control panel, took off the winding wires, re-jiggered it so they were both in phase and connected together, buttoned it back up and fired her up. The voltage gauge now only read 120, of course, but it only took a moment before Dös was out there sayin' "Dude, you gotta come check this out!" I went into the cabin to the inverter and, sure enough, the inverter/charger was hummin' like it had never hummed before. Dös said "Wow, that's impressive," when he felt the cables. He was working for Nortel at the time and they have some giant honkin' 48 volt rectifiers and battery banks but he was impressed that our 00 or 000 cable running to the batteries was actually getting warm.
Essentially, I had (more than) doubled our charge rate by doubling up the 120 at the expense of the ability of the generator to do 240 volts. (Since we had only ever used 240 once before, to run my giant compressor during a cabin expansion the previous year, that wasn't an issue.)
Charge rate was now solved but I was still never able to properly equalize the batteries with that set-up. You just couldn't get the battery voltage high enough to really do it, even after many hours of running at full charge. PFC control chips were just starting to become a thing so I started drawing up plans to make a PFC'd battery charger but in the years since, they ended up added a solar panel setup which keeps the batteries fully topped up, even when nobody is there for 95% of the year so it has become less of an issue and I never built the PFC charger.
</story mode off>My point is, you're likely to need a
WAY bigger motor than you expect if you go this route. That is where an electronic solution may well become more practical than the mechanical one. Phase imbalance becomes irrelevant and any design with feedback will at least do its best to keep the waveshape correct within the limitations of the wiring impedance, etc.