Awl--
What I found out today was pretty flabbergasting, about commercially-built rpcs. In a word, they effingSuck. In more words, they seem to suck gratuitously, even negligently.
I have two commercial units, one by Steelman (extremely well built), and a 5 hp Phaseamatic, which 5 or more years ago cost almost $800. Both produce a wild leg of about 275 V with 240 input, which fried everything it could.
The reason for this is--despite their bragging about "custom made" motors by Baldor, etc--is that they load the unit up with big-assed start capacitors
*only* between L1 and L3 (L3= generated), which are then *never switched out* or redistributed between L1-L3 and L2-L3, resulting in very high, very uneven voltages among the legs.My first inkling of this occured when I powered up a regular Baldor 10 hp motor with the two "oh-ficial" rpc's, and after switching off the rpc's, noticed I had *much better power* from the Baldor, all by it's lonesome, no caps at all.
After I cannibalized my Steelman rpc for the caps (4 caps at 50 microfarads each), and putting 100 microfarads on both L1-L3 *and* L2-L3, the voltage was much better, and became even better with another 5 hp motor added. The two motors yielded 239, 232, and 231 volts leg-to-leg, with 240 V input. That's about 3%--not bad!
The more caps you add, the higher the voltage gets. It seems to be a bit of an art to juggle the right hp with the right cap values. AND, it seems peculiarities of the motor design may affect this as well.
The site below gives rules of thumb for cap values: Start caps: 50-100 microfarads per hp; Run caps: 12-16 microfarads per hp.
Altho adding in caps "as you go", from a switching/breadboard-type ditty, may seem attractive ito voltage regulation for various loads, you get some fearsome sparks, and one commercial site advises that this is a no-no for sensitive loads, like pyooters, cnc, etc.
So unless you come up with some scheme to make adding/subtracting capacitances less electrically traumatic, whilst a load connected, it's likely best to anticipate these values and connect them before the load is added, and not disconnect them until the load is removed.
Some commercial units stabilize voltage by capacitor adding/shedding, but this is dicey, as above.
The following site has a very nice summary of diy rpc's, and points out two little known factoids:
120/240 V systems provide an inherently hobbled 3-ph, as two legs are not 120 deg out of phase, but *180 deg*. Which makes the other two phase angles 90 deg!They also point out that a number of small motors will give as much capacity as one large one, AND are more versatile, easier to start up, but require more wiring logistics. And, I suspect, might give an inherently more stable voltage. There is no such thing as too much rpc capacity--altho a lot of unnecessary capacity will spin your electric meter--ergo the beauty of staged smaller idlers.
Googling diy rotary phase converters yields many schematics for automatically switching out start caps, and switching in run caps--but it's proly not that inconvenient to switch them in/out manually. I think cnczone and practicalmachinist have dedicated threads for this, with lots of schematics, and some parts lists/sources.
I would advise being careful about putting rpc-idlers together in parallel, as two generated legs can be out of phase, if I grokked one of my many dazzling short-circuits correctly. Check the voltage diff between them, before connecting them up. If out of phase, maybe switch legs for the input.
Inyway, to sum up, I am dumbstruck that I can kluge something together that is *far* better than something I would otherwise pay thousands of $$ for, and which in some contexts is altogether unusable. And I WOULD have payed the money, if the stuff just wasn't so derelictly designed.
Props to Ig, for his good, albeit super-terse, advice.