Hi, ive just got my budget DIY cnc router table build to the stage where i can start to cut demo Gcode files.  I have a chinese breakoutboard controller connected to the printer port on a windows 7 PC. im using DM542 drivers and nema 23 3.2a stepper motors.  I first noticed my cuts were stepped on two sides (cutting out a basic square i drew in vectric vcarve ) i was cutting out of 12mm mdf in 6 passes. each pass steps out a little further.

The bit is at 90 degrees to the table, i thought i could be losing steps so have tried running the program with the bit raised away from the job then returning to zero, its off by the same amount each time (approx 2mm on both the -X and -Y axis) if i run the program again its out a further 2mm etc.

After some googling i have tried changing the direction and step pulses in the config menu from 0 to 5, this made no difference. I also tried unchecking the DirLowActive box for each motor (then swapping a wire on each driver to correct the jog direction) this had the effect of sending the error the opposite way (2mm +X and +Y)

ive checked for looses drive cogs and belt tension etc but the distance seems too consistent so i think its something software/pc related.

Does anyone have any ideas what this could be?

Much appreciated!

not sure if i can post the full chat in here but im currently working through this with google AI and think it may be the board itself at fault..

Ive been running a 'dead end' gcode test over and over and my x axis carriage is working its way furthere from my DTI

Since the error is a consistent 2 mm per run and occurs with the cutter lifted, I’d first check your steps-per-unit calibration and whether Mach3 is actually commanding the expected distance, then verify the DM542 microstep settings match the software before chasing lost steps.

Hi thanks for the reply, I spent a few hours trying things suggested by google ai. This included changing the step pulse and dir pulse timing all the way up to 15, changing the DirLowActive etc (this made the error go in the opposite direction) as it  was suspected it’s the pulse timing. I set up a DTI on the spindle body then ran this code..

%
G21 (Set units to Metric - change to G20 if you use Inches)
G91 (Set to Incremental Mode)
G01 X20.0 F1000 (Move right 20 units)
G01 X-20.0 F1000 (Move left 20 units)
M47 (Rewind to line 1 and repeat indefinitely)
%

I was losing approx 0.12mm each time the spindle returned to the DTI

Ran driver test which came back with good results, ordered a new breakout board from amazon in case it was this but it was just the same.  Changed the 5v power supply from the PC to a good phone charger in case of voltage drop but still the same.

I happened to read another thread on here with a similar issue where someone suggested trying sherline mode (apparently this changes the step delays to 40 (either pico or microseconds?) when mach3 settings only go as high as 15. This has sorted the error when running the loop program and it would return to the same point every time.  I re ran my program with the square inside the circle and the square now cuts out perfectly no steps on the cut at all. The circle has no steps but doesn’t look completely round though? Not sure if that’s my program or still the same error due to two axis running at once.  I’m going to order a good quality branded motion controller today as I want to build a decent machine as I will use this for work so don’t mind investing in it.  I’m looking at either an AXBB-E or a UC400-ETH

I would still like to know why my setup doesn’t seem to work with the Chinese board as lots of people seem to use them without issue on the YouTube videos I have been watching throughout my build!

Regards