Commmand gear torque A2

How did you configure the robot to be ceiling mounted? Can you show machine data?

Is the robot absolute accurate?

Has the robot a counter balancing system?

Both these conditions would mean the robot cannot be mounted on the ceiling, because internal models (dynamics model, absolute accuracy model) would be calculating results not fitting your ceiling robot which would lead to command gear torque errors. For these models you would need a totally different set of configuration data and not only gravity direction needs to be flipped.

Think about the counter balancing system instead of pull it would in ceiling configuration need to push. Since its A2 my guess is you have a counter balancing system.

Fubini

there is no counterbalance on Agilus line of robot.

also there is no oil in Agilus axes - lubricant is grease. (on some robots where lubricant is oil, quantities need to change when robot mounting is different) .

are you using correct MADA for ceiling robot or did you compare it to floor mounted MADA?

Quote from EmilijaS

And maybe load parameters are not ok?

No it should not. Load data is given with respect to the flange system and the flange system does not change when putting a robot on the ceiling. Only $Robroot changes (or in case of a KL involved $ersysroot).

As Panic suggested I would make a diff between floor mounted and ceiling mounted madas. If available possibly also for not WorkVisual created machine data of your robot but for data inside the machine data tree of an KUKA installation image. This would help to identify whether for your robot type some additional changes might be necessary.

Fubini

that file is a part of one file set (robot MADA contains also $robcor.dat).

manually reading and parsing of long files is a possibility but that is tedious and error prone. as already suggested, it is much better to use a tool to find the differences between two sets of MADA. suitable tool is software such as Beyond Compare, WinMerge, KDiff etc.