A little update on this topic.
The encoder of that motor is mechanically perfect, no problem on that side.
Tested feeding 5V with the PSU instead of the servo drive, but the problem is still there, after a while working I get the error in the drive. Changed the other diferential driver, same issue.
Created a HAL component that takes axis position and generates the UVW signals. So far it does what it has to do. My main concern was what could happen after powering up the machine if it's left out of machine home, but with the position_file to remember where is left does work well. Unfortunatelly I have not been able to test it with the servo because during the tests I crashed the glass scale to it's limit. Broke the reading head guide and the cables that go from the interpolator PCB to the LEDs and photodiodes. Already repaired both but is refusing to give me values. If I induce noise on the photodiodes, the interpolator IC gives me measurements so I will have to do some checking with the oscilloscope.
In the meantime I asked the servo vendor if they could get the encoder, but the manufacturer said no. Found another chinese encoder with the same ABZUVW signals at a decent price to give it a chance. Soldering to the old encoder connector was a pain because the new encoder cables were too short, but it did work. This new encoder logic is reversed and just had to change one of the motor phases. To couple this encoder to the servo I wired the oscilloscope to one phase of the motor and the U signal and left the encoder loose so it would slip while rotating the saft.
So far the lathe is working again, that was my main concern. I will probably get a couple of the new encoders to repair my other servos, even If I don't have them in use for now. I will continue exploring the way to over pass the encoder with the help of LinuxCNC as I think that it will be usefull for somebody.
