TB6600 and STEP_SCALE with Mesa7i96s

Hi every one.
I'm trying to build a CNC plasma table (link to the dedicated thread here : forum.linuxcnc.org/plasma-laser/52203-th...-plasma-table#297268).

I'm at the stage where I try my X and Y axis through a base machine with linuxCNC and pncconf. I haven't configure (nor wired) anything THC or Z axis related yet.

Firstly I'm not too sure wich value I should put for the cheap chinese TB6600 drivers. I tried with :

Step on time : 5000
Step Space : 5000
Direction Hold : 20000
Direction setup : 20000

I have no clue what those values should be. I just copied something I've read on the net about those drivers.

Secondly I'm pretty sure I dont understand what STEP_SCALE should be (I did NOT set the drivers to microstep).

Given that I have a 1:10 reductor after the motor and that my pignon is 32mm in diameter, I thought that for ten turn on the motor (so 2000step) I would have one turn after the reductor, thus getting a pi*32 = 100.531 mm advance per 10 turns of the motor.
It gives me a STEP_SCALE of 2000/100.531 = 19.89

But when I enter this value my pignons move so slowly it's barely noticable. I've tried with different value and it starts to really move around 100. I think I'm missing something here and I'm not sure what. I thought maybe I had messed up with the reductor and multiplied this value by 10 and try with 198.9. It moved way better, but not at the "default linear velocity" i had chosen during the configuration (300mm/min)

Can someone also confirm that If everything's right my gantry (when "manually" moved through the keyboard) should move at the "default linear velocity" I entered during the configuration, which in this case should be 300mm/min ?

Thank you very much !

edit : I can copy and past the .ini file if needed

Are you really using full steps (Microsteps=1)?

That would typically be quite rough and
susceptible to resonance.
 

I believe I'm not microstepping, though you make me doubt and maybe I confused the ON and OFF position on the driver because I have no vibration or rougness, I will have to verify. Either way, I have either no microstep at all, or 6400 microsteps (both position where all the pins are aligned, all ON or all OFF).

If I do the same calculation with 6400 microsteps :
-10 turns of the motors gives 64 000 microsteps.
-10 turns of the motor makes one turn of the pignon, which got a perimeter of 100.531 mm

So 64 000/100.531 = 636.62step/mm

I've tried to empirically find the STEP_SCALE which would give me a 600mm/min advance (the new setting I have set for "default linear velocity) and it's around 2348...

The thing is I can't really be sure of what I'm doing. I need confirmation than when you jog the machine manually, it moves at the "default linear velocity" I have set up during the configuration. From this point it will be easier to understand what's wrong.

Thank you, I'll try that.

edit : it is basically what i've done exept I directly edited the .ini file and didn't use the calibration menu from linuxCNC. I've counted how many turn for my pignon in one minute, then edited the STEP_SCALE accordingly. This is how I ended up with a 2348 value for STEP_SCALE which moves the gantry at the speed selected on the linuxCNC slide button.

I could of course calibrate every axis this way in a more precise manner, but it feels janky and i'ts bothering me to not understand where does the discrepancy with my calcluation comes from.

Problem solved.

I was, in fact, micro-stepping (1/32 so 6400 steps).

Also, I had one of the phase of a motor which was making contact with the frame of an aviator connector, throwing weird signals everywhere through the frame (I understood it because the limit-switches went crazy when connected). Why I had these weird speed is beyond my comprehension but now that I rewired it correctly my advance matches perfeclty with the calculation and I needed a STEP_SCALE = 64 000/100.531 = 636.62.

That will teach me a lesson about buying cheap connectors.