Gantry goes down 1.5cm further than up.

Hi!

I am currently building a new CNC Router. My current problem is the configuration of the Z-Axis.
I have a IHSV57 Servo with a 1:1 connection to a 1RPM/5mm Ballscrew. 

1: Although it seems to be configured correctly (which apparently it is not), the gantry goes down only 13,5cm instead of around 220cm.

• Steps on the Servo are configured to 3200 Steps Per revolution
• Ballscrew is 5mm Per revolution
• The Servo is connected via 1:1 belt to the ball screw

2:My second issue is that every time I go down it looses 1.5cm! So after 3 tries it's  4,5cm below the point where it homed. From my understanding, the Servo should have an Encoder in it, making "losing steps" impossible. No?

I already removed the Z-Brake on this axis as I thought it might be an issue. However, upon removing and Motor disabling the Gantry goes down slowly but surely moved by gravity.

I made a video of the process. (unable to upload it here since its bigger than 1.6mb)
1: Homed state, going down 3 times, re-homing.
share.icloud.com/photos/03dJTYtek7poTiNn1YLhaQ0uQ
​​​​​​​

Can anyone help me?

 

Buts its weird. As soon as the motor is off the gantry goes down because it moves so lighty (without brake). Also, it does not move down fully but only to 13.5cm. If it looses steps going up, I would expect it to go down without any losses (especially as it almost moves without any force).

I already reduced the acceleration and speed to veeery little, still the same problem with exactly the same distances. My Scope unfortunately doesnt have a Pulse counter, otherwise I would have counted the pulses for its way down.
Could it be that there is an issue in my config?

First, make certain that there is no slippage between the screw/coupler/motor connections. (Maybe mark a line on each.)

Perhaps you can observe what the drive is doing using its set up software. Use the software to monitor the commanded position vs the real position. Your goal is to try to determine if the servo is receiving all of the commanded steps from Linuxcnc. If it isn't then you need to check things like step space timing, filter settings, pulse voltage levels, and noise.

I found the error!

After two days of checking, measuring with the oscilloscope, calculating number of pulses by measuring the frequency and pulse duration,  I was questioning my sanity.
I also throughly tested the other axis which are fine, so I compared the axis in all the files and found my error.which should have been set to
I changed the value and had to invert the axis. Everything is running smoothly now.