So how are you doing that? Did you consider what I did write a number of posts back about speed of change at zero-crossing and at the peaks?

You mentioned a sampling rate of 1024 samples/period earlier.

There are 360 degrees in a full period.
So each sample represents 360/1024 = 0.35 degrees.

And how does that matter?

sin 0.35 = 0.0061
ln 0.0061 / ln 2 = -7.35
So you get an idea how fast the voltage changes between two samples as worst - during your "zero crossing".

sin 90+0.35 = 0.99998
ln (1-0.9998) / ln 2 = -15.7
So you get an idea how fast the voltage changes between two samples as best - at the peaks.

So if you just look for highest/lowest value, the worst-case change between two samples requires a 16-bit ADC to capture.
Actually 17.5 bit if you are going to throw away 1.5 bit by having your input signal just span a small part of the full ADC range.

So the question then is - do you really need a sampling rate corresponding to 1024 samples/full period? Given that you have an ADC capable of 8, 10 or 12 bits - and you only plan to get access to 6.5, 8.5 or 10.5 bits.

Exactly what conclusions have you made?

By the way - are you going to control equipment with a peak power of 180 kW? The diagrams in the linked document shows a current of 3 Amp with the volage at 60 kV.