Invertor required

[Voicecoil]

All these VFD's will be rectifying the AC input to DC, and then doing stuff with that to make the 3 phase O/P. For low power (and low cost) you can use a capacitor/diode voltage doubler on the input to generate ~650V DC, alternatively for larger power you can use a switchmode Power factor Corrector (PFC) stage to do the same and also keep you legal for harmonic currents. This will need just one (or a few) small decent inductors, if the switching frequency is high it's quite possible to get 3 KW through a couple of cores maybe 35 or 40mm across, larger powers pro-rata, so you don't need massive transformers.

[Neale]

I use an "inverter" that works on the voltage-doubler principle - it was a bit of a bodge by a now-out-of-business company who grafted a voltage-doubler to the front of what is quite a nice ABB 380V VFD but it does mean single-phase 240V in and 400V or so 3-phase output. It works fine until you put a heavy load on it when the effectively half-mains frequency input means that the DC voltage sags too much between cycles and the VFD trips. I've been kicking around the idea of finding a 3KW 240-415V transformer and go back to the original ABB VFD input arrangement - and a cheap transformer has not found its way to my workshop yet - or try one of those Ecogoo 240-380V boxes (I linked to this company in an earlier post on this thread). They seem to deliver a lot of performance for the price, if they work as well as they claim...

My lathe has a mechanical variable-speed mechanism so I have no experience with dual-speed motors and VFDs. In fact, I run the VFD at 50Hz output, with a switch that drops this to 25Hz for a quick low-speed option.

[Voicecoil]

I use an "inverter" that works on the voltage-doubler principle - it was a bit of a bodge by a now-out-of-business company who grafted a voltage-doubler to the front of what is quite a nice ABB 380V VFD but it does mean single-phase 240V in and 400V or so 3-phase output. It works fine until you put a heavy load on it when the effectively half-mains frequency input means that the DC voltage sags too much between cycles and the VFD trips.

Sounds like they missed a trick and only used a half wave voltage doubler and/or got the capacitor sizes too small. The full wave voltage doubler circuit works pretty well if you do the sums for the output power and choose the right size caps, but it does have twice the component count of the simple circuit.