I noticed twice in the last few days that the Y coordinate changed after homing the machine.
When I went back to X0Y0 and started the next job, the bit moved about a millimeter towards the front or back of the machine. The X coordinate seemed to be correct or at least very reproducible.
I read this post on the forum and I think it may apply. It makes sense that two dueling stepper motors could get out of whack without the same tension.
I have seen similar things occur with my Voron v2. That is always fun to tune – four motors, one in each corner of the floating gantry.
Like Will says, start the machine up with the gantry pulled to front or back and ‘square’, you may have seen in one of my posts that I 3D printed some little adjustable endstops to help me get repeatable square.
Assuming your belt tension, pulleys etc. are all good then you’re down to the homing switch repeatability.
Which machine are you running? Specifically does it have the old mechanical homing switches or the inductive ones?
I have found that if I let the machine re-home itself, after a power cycle or using STOP on a job then it will frequently come back with the X, Y a little off, in my case generally not more than 0.25mm but enough to be irritating on the wrong job.
To deal with this, I make sure that I keep a zero point on the stock or fixture that I can re-test the machine against after a re-homing and then re-zero if it’s moved. On the more complex jobs with slightly rough stock, lots of setups or bit changes I tend to machine an X, Y zero at the front left of the stock in the first operation so I have a repeatable point to go back to.
I have a Shapeoko 4 with the inductive homing switches. I assume those are pretty accurate but I’m not positive.
The Y difference is at least 1mm – I could see a ridge yesterday by just looking from the front of the machine. I will check the belt tension and pulleys as well. I have not checked them for force or frequency and have only listened to the sounds as I pluck them.
Normally I would not home between jobs but there was an error and I did an emergency stop – my button kills all power to the machine except for the Raspberry Pi running CNCjs. The other occurence a day or so before that, there was no emergency stop. I cannot recall why I homed the machine between jobs that time.
Keeping a spot on the stock for reference will also help – I try to zero off the front left corner and on my 3018 I used one of my own 3d printed jigs to do this fairly accurately.
Thanks for the input. I hope to try and tune the machine tomorrow morning.
I have the Shapeoko 4 XXL machine. I marked off 200mm from each end of the Y rails and slipped two 1/4" shanks under the belts. I measured appoximately 152 Hz on each side, I was amazed but the numbers were reproducible when switching back and forth between the belt readings. I suppose I need to reduce the tension. The belts do not seem to be very tight at all but I can try to reduce the tension on both.
When I moved the X gantry to the front of the machine, the ends touched at the same time. I did not try the rear as my machine is pushed against the wall but I can try that later to verify.
When I had to hit the emergency stop, a small piece of wood had come unstuck from the blue tape / CA glue and knocked about. It could be that this error caused some kind of temporary belt distortion? I’m not sure. That is why I rehomed the machine and tried to get back to the origin.
As I said, X was fine and Y was about a millimeter off. I did not see that there was an angle – I redid the entire contour job and it hugged the left and right sides of the pieces without much sound, but the bit totally cleared the front and dug in another millimeter on the back.