E-stops can be a pain to implement.
I opt for limits forming part of the physical E-stop circuit, so if that circuit gets broken, the main power to the drives/spindle/anything else mains powered i.e. coolant pump gets removed.
I then have the KFlop monitoring the e-stop, and should it get triggered, it activates inputs to the drives that trigger them to stop. However, the KFlop has the benefit of being programmed to do this within milliseconds, so it's not reliant on asking the computer/mach what to do.
I know Hood likes to use Safety Relays, that when triggered, will trigger the stop input on drives, then after a small time delay kill all power. This has the benefit the servo drives should remain powered long enough to do a fast controller stop, before power is removed.
I suspect the issue dude1 is trying to mention, is the feature of DC powered drives where they dump excess power back into the DC supply during heavy deceleration, which can cause a voltage spike resulting in over-voltage damage. This doesn't apply to the type of servo drives you're using (it would apply to regenerative drives), but it could cause the drive to trip out with a DC-bus overvoltage error, as they do dump power back into the internal capacitors. If the error did get triggered, all that should happen, is the drive will error out and let the servo freewheel to a stop.