serious error of position

salve a tutti esiste un difetto di base di linuxcnc e' presente in tutte le versioni da emc2 fino ad oggi . la mia macchina e' una pick and place smd i motori sono step il rapporto di riduzione e' 1:1 cioe' il motore e' diretto con la cinchia ed in questa configurazione si nota il difetto nelle macchine con un rapporto di riduzione alto non si nota perche' viene diluito nella riduzione ma esiste comunque , sono state fatte tutte le prove possibili sono stati montati motori in continua con encoder e motoriduttore 150 : 1 . ed introduce lo stesso errore attualmente la parte di controllo e' una mesa 7i43 , escludendo la mesa 7i43 e pilotando i driver direttamente da parallela non cambia niente , il difetto e' questo :
si azzera il plotter in basso a sinistra x0 y0 poi si va ad una posizione qualsiasi ed il movimento e' corretto ritornando in posizione home la posizione risulta corretta , praticamente da qualsiasi posizione mi trovo posso ritornare in home in maniera perfetta senza errori e fino qui tutto bene , quindi da home vado ad una posizione qualsiasi ad esempio x200 y200 da qui riparto e vado a x400 y400 ora ritorno alla posizione x200 e y200 e come si vede la posizione non e' la stessa ma differisce di 0,9 millimetri , ritornando in home la posizione di zero e' perfetta e si azzera anche l'errore .Questo errore non varia e non e' dipendente dal tipo di configurazione ne da nessun parametro del gcode.

Hello everyone , there is a basic flaw of LinuxCNC and ' present in all versions from emc2 to date. my car and ' a pick and place SMD step motors are the reduction ratio and ' 1:1 that ' the engine ' with the direct cable neck and in this configuration you will notice the defect in the machines with a high gear ratio not you can see why ' is diluted in reducing but still exists, have been made all possible tests were mounted DC motors with encoder and gear 150: 1. and introduces the same error now control part and ' a mesa 7i43 , excluding the mesa 7i43 and piloting drivers directly from parallel changes nothing , the fault and ' this :
resets the plotter in the lower left x0 y0 then you go to any position and movement and ' proper returning to the home position is the correct position , from virtually any position I'm in I can go back home in a perfect manner and without error up here okay , so I'm going to be home any position such as x200 Y200 allotment from here and now I'm going to x400 Y400 and Y200 x200 return to the position and how you see the position and not ' the same but differ by 0.9 mm , returning home in the zero position and ' perfect and it also clears the error. This error does not vary and is not ' dependent on the type of configuration it by any parameter of the gcode .

See the video

can you read:

axis.0.motor-pos-cmd
axis.1.motor-pos-cmd

with halmeter when the position is in error?

This is to see if the commanded position is in error
or its a feedback/hardware issue that will need to be traced further

Also posting your hal and ini files may help locate the problem

ok thanks hal interest , positions do not correspond with the positions of axis
ypos-fb = 182.7883 axes 182 624
ypos-cmd 182.7883
xpos-fb = 366.4099 axes 366 582
xpos-cmd 366.4099

are you saying tha the axis display is
182.624 but ypos-cmd is 182.7883?

is exact

Nice camera.

Can you please make a video where you start at any point you want, call it X0 Y0 (if you have a scale below (that the camera can focus), something that you can measure what the machine really moved would be perfect), then:

ask for moving to (one line at each time):
Y0.5
Y1
Y0.5
Y0
Y-0.5
Y-1
Y0

Backlash?

ok, I'll show you an-y-axis only say 'cause both axes behave the same way. I can show you that the axes do not move in a linear fashion but is introduced by a software error

That is backlash.
If you go forward always in one direction you do not notice, if you go backwards it is present.
The best way to see it is like you did. When you traveled from 0 to 200 then 300, it was all good, when you changed direction, there it is.
Backlash!

I almost forgot. Linuxcnc can emulate the backlash and try to compensate that error. Please see a bit more on the documentation.

are not in agreement, the game mechanic does not exist in this machine transmission and 'belt, there are no gears, if you look carefully the video when I walk 5mm at some point makes the most of the steps' courts, then after a while returns to 5mm. the problem 'in the software.