IHSV60 integrated servo moves when removing power or disabling it

Hello everybody,
I'm almost on my 3 years mark with my IHSV60-30-40-48 integrated servos, they always served me well, lot of torque and power, awesome finishes and huge MMR (for a CNC router obviously). 

Only one thing has been bothering me a bit, whenever i remove power, the machine goes in alarm or i push the E-STOP, the two servos that move the gantry X axis jerk a bit, each in a different direction from the other. This doesen't happen on the Y or X axis(Correction, it did happen on the Z axis a few times, causing the endmill to be plunged in the material for a few mm, this confirms that the movement is done under servo power somehow, and not by gravity, as the amount of plunged mm seems to have been always the same in the few occasions that this happend).

Now, i've always been concerned on how LinuxCNC manages two servos on one axis, expecially for homing, as during the homing procedure, they seem to move differently from one another.

I've checked the gantry, and it's square to the axis, so no twist that could put the servos under load when homing, tension that is released when the servos gets unpowered. Another confirmation of this is that sometimes, the X axis doesen't move when removing power.

My thought is that it could be caused by bad tuning maybe, but i'm now sure if it's plausible or not.

Do someone have any thoughts?

Are you using the drive enable signals to disable the drives
when there is a fault or LinuxCNC is not running?
 

The drives gets 5V constantly when disabled. To enable them I have to remove power from the enable pin. For me this logic is counter intuitive, meaning that if a wire breaks, the drive enables(the opposite of the logic used in E-Stops).

I'm looking for a way to invert that logic, but drive wise I couldn't find anything

EDIT: the drive manual indicates the following: "SW5 dial code switch sets input edge, off means rising edge and on means falling edge."
I don't know if this is going to change things, either on the enable, or both the enable and STEP/DIR(changing from falling to rising and vice versa shouldn't change anything about step/dir, am I right?)

I guess one way is to use a normally closed relay/SSR
to power the drives disable inputs until LinuxCNC energizes the
relay with say motion.motion-enable. Its still awkward as it requires
control power be present to disable the drives.

I know this is common in step drives as it allows the enable input to be
ignored so simplifies wiring and probably results if fewer technical
support costs, but it is really counter-intuitive

I wonder if the setup software can change the enable polarity
over the RS-232 port?

I've looked into the software's options, and an option to switch enable polarity is not listed anywhere. I'll try to switch position on the Switch 5 and keep you posted

Update: I've tryied not disabling the servos on E-Stop, they jerk anyway. My guess is that the tuning is probably constantly applying some torque and it gets released as soon as the loop stops