VESC with propeller, not reaching max performance

Hi all, I've been using the forum a lot to troubleshoot but this is my first post :)

I'm an undergrad engineer - 

We've had a lot of problems trying to replicate T motor test data in FOC and BLDC mode

So I've been wanting to use the VESC for use in UAVs due to the potential efficiency benefits as well as wanting to be able to use reverse thrust and record data.

We've been testing these 3 motors in a wind tunnel: AT3520 (550kv, 1000W, 14x7e), AT4125(540kv, 2000W,16x8e) & AT5220 (220kv, 3000W, 19x10e).

We're running 6s , 6s and 12s respectively.

https://store.tmotor.com/goods.php?id=794

https://store.tmotor.com/goods.php?id=827

https://store.tmotor.com/goods.php?id=948

we're using the FLIPSKY 60amp 6.6 VESC

for the AT5220:

In BLDC mode as I understand, we should just be able to get max thrust with no tuning, there's phase advance which can affect max RPM - i've tried adjusting this to no avail - still getting a maximum of ~90,000 erpm. (instead of the expected ~110,000)

In FOC mode I've messed around with it a lot and it runs nicely up to 90,000erpm but no further.

We've also tested this motor with the smaller props and still failed to reach the correct rpm

For the AT3520:

We manage to almost get maximum thrust in BLDC. (this is the smallest motor) but ~10,000 less erpm on FOC. I tried field weakening and this seemed to have little effect

For the AT4125:

pretty meh performance all round.

I've probably spent 24+ hours messing around with the VESC now - I've pretty vastly underpredicted how easy it would be to get right.

Is there something i've missed? 

https://vesc-project.com/node/356 - in this forum post someone shows the VESC vastly underperforms compared to other ESCs, both FOC and BLDC. Apparently this is due to the motor "saturating". For us the high end performance is very critical for short periods of time (take off) so we don't care much about efficiency in these regions. 

We've purchased a "normal" ESC instead which supports reverse thrust but won't be able to measure anything without further equipment.

I've left out lots of details, but if anyone's willing to help We'd greatly appreciate this and I can send any details / data over

massive power consumption by the VESC compared to T motor data:

However on the smaller motor, we do get much better agreement (on BLDC):