Anchal, Paco and I agreed to follow the polarity conventions below for suspension damping loops.
Some of the polarity/gains were changed to homogenize all the suspensions to the convention.
All the suspensions are homogenized except for MC1 (which have all - in sensor inputs and all - in coil outputs) and AS1 (SDCOIL_GAIN have the same sign as LL).

*SEN_GAIN to be +1
INMATRIX to be the following, and calibrated so that SUSPOS/SIDE_IN will be um and SUSPIT/YAW_IN1 will be urad (calibration to be done later)
+ + + + *
+ + - - *
+ - - + *
* * * * +
SUS*_GAINS to be +
TO_COIL gains to be the following
+1 +1 +1  *
+1 +1 -1  *
+1 -1 +1  *
+1 -1 -1  *
 *  *  * +4
*COIL_GAIN to be the following or flipped one so that SUS*_GAINS will be +
+
-
+
-
+

To do this, the following changes were made

For BS, ITMX, ITMY, PRM, SRM, ETMX, ETMY, MC1, MC2 and MC3 ("old" suspensions), TO_COIL_5_4_GAIN (for side) is changed from +1 to +4 and SUSSIDE_GAINs are divided by 4 accordingly.
For ETMX, the sign of SUSSIDE_GAIN is flipped to +, and SDCOIL_GAIN to be -1 (it was +1).
For MC1, *SEN_GAINs are - (not following the convension; see 40m/16894). The sign of INMATRIX_4_5 (for side) is flipped to +, SUSPOS/PIT/YAW_GAIN are flipped to +, and *COIL_GAIN are flipped to - (not following the convension). IMC WFS output matrix components for MC1 were also flipped.