110428-1758 EDT
Davebone:
Are you sure the problem is caused by a sag vs maybe a high voltage transient?
You say an instant. Is this 2 cycles or 60 to 180 cycles, 32 milliseconds vs 1 to 3 seconds? Assuming the spindles and servos derive power from a DC bus with substantial capacitance, then a short perturbation of even 3 seconds may not cause their failure.
Therefore I assume that the control part of the system, meaning the computer that controls the spindles and servos, is probably at fault. This might be transient susceptibility in some fashion, not likely the power supply. Again because there is substantial capacitance storage in the power supply.
Maybe MOVs and filtering of the AC to just the control power supply might help. But the noise could get thru in some other fashion.
It can be tough to solve your problem but you need more information. For example supply voltage waveforms resolved to maybe 0.1 millisecond from just before and just after the disturbance. Also what fault messages do the controls display? I am fairly familiar with HAAS machines. These machines trip out nicely on loss of phase with no damage. In the great northeast blackout we had some very large and moderately long voltage changes, both low and high, and mostly lost some fuses in the HAAS machines. No part or machine damage.
Are you drip feeding when these problems occur, or are you using an internally stored program?
.