This may be another machine with the wrong STEP signal polarity for either (or both!) axes from the factory. Some earlier problems explain what’s going on:
If getting the polarity right doesn’t fix it, then things get interesting. Successfully running a backlash test pattern will eliminate a bunch of other possibilities:
Scale this test pattern to fit the platform and run it as fast as it will go in Line mode with optimizations turned off and power set to mark cardboard: