Machine moves by itself

I was referring to inverting the step pin in hal, not electrically.

Another very important thing to remember about the step timing specs specified in a drive's manual is that all of those values are minimum timings required for the drive to be able to function at all under ideal conditions. Doubling a drive's minimum timing specs is often a good idea so long as they won't interfere with your required maximum velocities.

Another detail to remember, the step signals are only pulsed on for the step length time specified for your step length, for every step. They are not turned on and off for equal lengths of time. So if you have Linuxcnc configured with a 5000ns step length and command a move that makes 2ms pulses, the drive will be sent a 5000ns pulse every 2ms not 1ms pulses. This is one of the reasons that inverting the signal in Hal can make a big difference. (Like Andy illustrated.)

I have but I couldn't attach, I sent the link.



Driver A3: drive.google.com/file/d/1SZGpqod4p6Mx70GiIygGle5xt2cJAP5x/view

 

Ok Andy and Todd, I will test.

Tks

Interesting.a quick look at the manual you linked to does not mention any specifics of step timing other to reference them as t1,t2,t3... But nowhere mentions what they should be equal to. In the parameter section it has a parameter setting (#33.0) for a filter value for the position pulse inputs. The setting of this should essentially set the minimum time value.

What settings do you have for your minimum step timings in Linuxcnc, and what do you have set for the parameter #33.0 in your drives?

I think a good conservative setting for your step timings would be 50% of your base thread time. (Assuming you're using the reset function.) And approximately 25% of that for the parameter #33.0 filter time in the drive. Also be generous with the Dir pulse setup time at least equal to, and double the base thread time would not be out of line.

Alternatively you may want to consider switching the step pulse mode from step/dir to quadrature (AB-phase). You would have to remove the reset function from your parallel port settings. This might give more reliable positioning.

There is a good wiki about determining optimal step timings for a parallel port system here.
wiki.linuxcnc.org/cgi-bin/wiki.pl?Tweaki...ftwareStepGeneration

Hello guys,

 With a lot of gratitude and happiness after a week of testing and with the help of Todd (following last tips) and friends we managed to solve it.  
The last two things I did was change the servo thread and the second I changed the Step and Dir signals to NOT (first it was both normal signals and previously it was Dir to NOT and normal Step).  
I am very grateful for everyone's effort and help.  

Thank you so much

Ingo