The GRMC/MZ servo was suspected to have issue, but it was found that GRMC works well in elog2741. This indicates the problem comes from MZ servo.
I asked Pierre for some suggestions. Especially, I found the enable signal from MZ servo is always around zero. Pierre suggested me to check the resistance of MZ servo while MZ servo is not powered. I did such check and found there is a resistance of 67kOhm. At least there is no short cut.
I did similar check as elog2713, I found that the signal coming from GRMC ref PD has no signal. Then I checked the PD and found it is not powered! This is very strange because the switch of this PD is very hard to reach. But anyway, by switching on GRMC ref PD, I could lock GRMC and MZ again. Now there is no problem at all for GRMC and MZ.
Although this is just a stupid issue, but during the check of GRMC, I found out it is very necessary to record the level of signals for each control servo. These values will be very essential for trouble shooting in the future. I plan to make a list of all useful channels of these servos soon. On the other hand, elog2741 is a standard way to check servo and can be useful for future check as well.