Were those voltage readings taken with an averaging meter? If so, try measuring with a true RMS meter. If the phase-to-phase voltages get significantly closer to each other when using an RMS meter, that probably indicates that there is significant harmonic content (for example third harmonic). The problem is that the diode bridge rectifier in a VFD is responsive to peak voltages, not RMS or average voltages. However, if both RMS and average measurements show balanced voltages, then it's likely that there is low harmonic content and the peak voltages should also be balanced.
It should be possible to get the peak voltages more balanced in the presence of harmonics with, for example, buck boost transformers. But especially if we are dealing with an open delta service, harmonics could even make the peak voltage across the phases where 225V was measured even higher than where 250V was measured. So in this case you'd want a buck boost to reduce the voltage where 225 was measured, as counterintuitive as that sounds.
I believe this is why there was only current flowing on phases B and C on the feed to a VFD in this previous post, even though the measured B-C voltage was less than the A-C and A-B voltages:
[QUOTE="synchro, post: 3044798, mem...-high-leg.2587440/page-4#post-3019099[/QUOTE]
It was not a true RMS. It was a Kline amp/combo meter. About $100 I think. I don’t know the model number. I do have a true RMS meter at my shop. If you think I should, I’ll test with it.