I’m using a Shapeoko 5 Pro with a BitZero v2.0 to flip-machine a part in Fusion 360 (I cut the stock to the exact dimensions of the part and I’m using the model stock point at the bottom-left corner on both sides. I’d attach the file, but the forum won’t let me). No matter what I try, the flipped side is always off by about 1 mm in both X and Y.
The machine is square, everything is tightened down, and I’m re-zeroing the stock after flipping. I’ve read in the forum that re-zeroing after a flip can introduce errors, but I’m not sure why, since I thought it should still be fairly accurate.
Here’s the strange part:
When I machine a smaller version of the part, the alignment is almost perfect.
When I machine the larger version (all 8 pieces in the Fusion file), the offset issue appears—always around 1 mm.
I can’t figure out what’s causing this. Any ideas on what might be going wrong?
As an interesting update…I’m I’m jogging to the zero point that the BitZero is calibrating, and the point is dead on the stock where it should be…but it’s still off…idk what’s happening here.
I just finished cutting out my tool paths and ran into a strange problem.
Before starting, I jogged to the relative zero point, and it was exactly where it should be. However, once I started cutting, the first four “slats” already had about 1 mm of X-axis drift, cutting too far into the stock from the right, with no deviations from the left.
Then, immediately after those first four slats are contoured, the last four slats are shifted by about 4 mm on the X-axis, and that 4 mm shift is consistent across all of them with no more ~1mm X-axis shift from the right (problem with the first four slats).
I’ve attached photos to show what I mean:
First photo — the first four slats, with ~1 mm of X-axis drift.
Second and third photos — the last four slats, with a consistent 4 mm X-axis drift after the contouring pass with no issues from problem #1.
Third photo - slats put together to show the vast differences
If you are getting different results each time, I would assume a machine issue.
I would use ncviewer to confirm the coordinates in the nc file.
What post did you use ? That does not show in the .f3d file.
For speeds and feeds, I’m cutting HDPE. For the Back and Front cutting operations, I’m using a 1/4" O-Flute bit (running 18,000RPM at 359 m/min, at a Cutting Speed of 2,540 mm/min - cutting depth of 6.35mm) and a 1/8" O-Flute bit (running 18,246RPM at 182 m/min, at a Cutting Speed of 1,852 mm/min - cutting depth of 3.175mm).
So the tool changes are manual, and I’m preloading the first tool. Not sure how this would be an issue, but if its of any more help, I’m using the regular “blank” bit for the BitZero v2.0 when zeroing the stock.
What’s super weird is that it’s so inconsistent with being off. For example, the wall thickness, after the Back operation runs, is supposed to be 3mm on each side. However, one side is ~2.7mm and the other is ~3.4mm. But, when I scaled this down to just one “slat” when testing, there was virtually no issues at all.
I’d set up the file and cut one slat at a time, in each position — if each slat cuts correctly, then there would seem to be a problem with the Rapid movements resulting in lost steps — if a slat has a problem cutting with similar code in a different position then perhaps something is mechanically interfering?
I ran two simple flip test cuts using two methods:
Method 1: Re-zero with the BitZero V2 after flipping.
Method 2: No re-zero after flipping.
For both tests, I used the same reference setup — 3-2-1 blocks on the Y-axis and a couple of stops on the X-axis. Both methods produced the exact same offset after flipping.
Here’s where it gets strange:
In Method 1, after probing the stock with the BitZero V2, I changed the tool to an O-Flute. When I jogged back to the just-probed point, the X was off by -0.007". I tried re-zeroing with the BitZero V2 again, but the offsets didn’t reset. back to zero. In fact, the X stayed the same (with a weird offset of -0.007) and the Y got an offset of -0.006— Carbide Motion still showed -0.007" (X) and -0.006" (Y).
Since I couldn’t manually reset those values to zero, I went ahead and made the first cut for Method 1. After the cut and flipping the stock, I jogged back to the X/Y “zero,” and it showed the same offset, and Carbide Motion still indicated no loss of steps (same -0.007" and -0.006" values). I re-zeroed with the BitZero V2, and the cut was still off (picture attached).
For Method 2, I skipped re-zeroing entirely — the result was the same offset as Method 1 (pictures attached). What’s weird about this one is that the same offset was applied -0.007 and -0.006 for X and Y, respectively. But, what’s even more weird is that I jogged back to the X, Y point that was probed and the same offset was applied.
Something strange is going on here… has anyone seen this before? I feel like if it was loosing steps, it wouldn’t be this consistent, especially considering that there seem to be different offsets being randomly applied after zeroing with the BitZero V2.
Sorry, I should’ve posted a picture of the set up IRL and in Fusion 360; I’m attaching those here. I’m using the top of the bottom left corner as the WCS for both sides, with the WCS for the second cut being the opposite corner (pictures attached). This set up is consistent with flip machining?
I see your point, but what about the very first cut operation—specifically the back-channel milling—being off on the X-axis for all slats? For example, during the first pass with the 1/4" bit, one wall measures about 2.76 mm thick while the opposite wall measures about 3.39 mm. Both should be exactly 3 mm.
This would make sense if I were seeing the difference after the second operation (opposite side) or after cutting the slat completely out, but this is happening on the very first pass. It’s as if the machine is arbitrarily offset by a random amount right from the start.
I’m attaching pictures showing:
What the cut should look like in Fusion 360 (both sides at 3 mm)
The actual cut results, with one side at ~2.7 mm and the other at ~3.83 mm
It feels like the machine is skipping steps in software, but there are no obvious mechanical issues—the chains and wires have plenty of slack, and I’m not hearing any indications of missed steps.
I’ve considered adding an encoder to the stepper motor shaft to track movement, but at that point I might just switch to closed-loop steppers. Has anyone set up a system where, if a step is missed, the machine stops the program, displays a warning, and pinpoints the exact line where the step was lost?
Inexplicable issues like this can often be traced back to insecure wiring harness connections - they can vibrate & not provide solid connections to the stepper motors.
