Johansense.

Because life is supposed to make sense

Long story short I need to build a few boost power factor converters to run induction heaters, telsa coils and what not.
I'm not interested in getting 100% power factor, so the three phase single or dual switch ac to dc conversion is good enough.
(to get 100% power factor across all three phases requires at least three independently controlled switches, and for continuous conduction mode, a uProc to control them)
The attached schematic also lends itself to resonant snubbing for the Boost IGBTs and the boost Diode D5 and D6. That being said for a high frequency design, it would be run in discontinuous mode, and a resonant turn off snubber only requires two more diodes rated the same as boost diode D5 and D6 (and some pulse capacitors that can handle the full boost current)
--that being said, if the boost switches are replaced with mosfets, resonant snubbers are not required, but to get a 4:1 step up ratio, IGBTs are likely required.
Turns out there are some incredibly cheap film capacitors on ebay, on the order of 60$ for (2) 100uF 400vdc and (2) 3.1uF 600vac capacitors, rated at 150 amps for the 3.1 uf cap, and 500amps for the 100uF capacitors.
100uf for Cbus1 and 2,
3.1uf for the transformer resonant cap and Ca,Cb,Cc
100uF for Cbus 1 and 2 on the other end of your transmission line.

1000 vdc 80 amp fast diodes are on the order of 5$ each new. --8 required.
600 volt 60 amp IGBTs rated the same are about the same price. -- 6 required
For a low frequency design, employing the inductance of a wind turbine generator as the inductors, and the removal of capacitors Ca Cb, Cc, would probably require a resonant snubber to recover or soften the boost diode recovery charge for emi purposes (but perhaps not if CCM can be achieved below 5Khz.
one additional modification is to split the current between a mosfet and an IGBT for the boost switches. both turn on at the same time, but the IGBT turns off first, thus the mosfet eats up all the turn off losses. this doesn't increase the cost much btw.
Not shown in the schematic is a resonant inductor in series with the step down transformer. power control into a battery or other load is then controlled by the frequency, and turn off and turn on losses are very low.

so while i'm estimating the actual parts cost and the ratings of these components
(actually, designing the entire thing around maxing out the available capacitors)
a 16KW power supply run from 100 vac to 800 vdc, and then back down to 48 volts (56) would cost 276$
plus the following components:
A 16 KW transformer (i'm guessing i could do this for 50$ upper limit at 100Khx. 3 high frequency boost inductors.
At 100Khz and buying EE80 cores at 17 bucks each, it may be possible to just use three of these cores. --17$ x 3 . i'm estimating they would need to be liquid cooled.--for the price of the core, that's 3 pounds of copper wire.
6.5 uH required for 6KW, 100 vdc to 400 vdc at 100Khz. peak inductor current of 120 amps.
i'd just as soon use one pound of copper in an air core. 8$ x 3.
both boost switches turn on at the same time, so only two isolation transformers are required per switch, one to supply 12 volts to run a 3$ gate drive chip.. the other to supply the signal.
10$x 2 i'm estimating.
for the dc-dc converter side, 4 are required, but the switches are 1/10th the size, run at a fixed duty cycle of 50%, so all they need is a transformer. 5$ or less in parts.

to control the dc-dc converter, basically take an existing resonant dc-dc converter from the nearest plasma tv in a dumpster, and hook the output of the control chip to a gate driver and feed it into the gate drive transformers instead of directly into the mosfets. then scale the voltage accordingly, and add some additional components to regulate the power flow instead of keeping a constant voltage.
so.. i'm estimating about 400$ worth of components.
and that's to run 16 KW from 100vac to 800 vac, then back down to 48 volts.
any thoughts?
is running 800 volts dc a problem?
cutting the voltage in half (+/- 200) doesn't really change the cost much at all.
to reduce all the components down to a more reasonable 8KW design, the 120$ worth of high ripple film caps can be replaced with 20-40$ worth of film caps at the expense of making 200-800 solder joints.. --probably not worth it.
the boost switch cost is cut in half. the Diode cost is cut in half, not much else is.
air core boost inductors are on the order of 20$ in copper. so, this becomes 10$. same labor cost.
so, probably 300$ to run 8 KW from 100vac to 400 and back down to 56. ===================
is this at all worth it?
things that would be required: intelligent turbine control. getting the boost circuit self operational is easy, the dump load can be mounted on the tower.
--sending the turbine RPM down the tower to the dc-dc converter is required.
--setting the boost PFC system up to have a variable output voltage instead of a fixed voltage is an option.. this also reduces losses considerably.-- but it complicates the step down converter. --- actually, its not that bad, because as the bus voltage falls so does the available power, a 5:1 voltage variation is probably doable.
from here: http://scholar.lib.vt.edu/theses/available/etd-08142002-075617/unrestricted/Barbosa_ETD.pdf page 160