Shapeoko 3 (not XXL or Pro)

Purchased this Shapeoko 3 about 5 years, but never had the opportunity to fire it up. Its factory setup and mounted with a GRBL v2.2 controller. I think its running .09 version firmware.

I am attempting to mill my first piece, a material holder, that I designed in Fusion 360. In Fusion I have the mill configured to use the carbide.cps in Post Processor.

The issue is the drawing looks fine, however the cut does not match. For example, in the drawing the width between the two edges do not match. See image.

Also, it does not travel to the bottom of the material.

I put a mark and zero’ed out my axis and jogged the steppers and made another mark. In Carbide Motion it shows I moved 49.50mm. If I measure the distance (with calipers) it shows it only moved 24.53mm.

I must be missing something simple. Any help appreciated.

Can you share your Fusion file?
Can you share your Grbl settings?

1 Like

I seem to remember that early controllers had DIP switches on the board near the stepper motor connectors, and when the distance traveled was half/double the expected one, you would toggle the switch between 4x and 8x on the axis.

24.53mm is not quite half of 49.5mm, but very close and the rest may be due to uncalibrated belt stretch
I would not be surprised if the 4x vs 8x switch setup addresses your problem.

If your board does not have stepper switches then it may still be incorrect steps per mm parameters in GRBL (like Neil hinted at)

1 Like

Here is a link to the F3d file. tools v4

Can you advise how I would capture my GRBL settings to upload? Not sure where that is located.

Here is a picture of the controller. I’m not seeing any DIP switches to adjust.

Not sure how or where to adjust the GRBL settings.

This is caused by the defaults in Carbide Motion 3 being for 4x micro-stepping

Please go to the MDI and send:

$100 = 40
$101 = 40
$102 = 40

That should work with the default 8x micro-stepping on new boards.

You may want to update to Grbl 1.1 so that you can use CM5 — but you’ll need homing switches which will either require rolling your own w/ the existing Molex connectors or replacing the controller (so that you can also upgrade to the proximity switches — might also want to go for a Z-Plus or HDZ which include those).

Or, use the machine as is w/ Grbl 0.9 and CM3 (I rather liked its appearance and minimalism and kind of miss it) or use a 3rd party communication/control tool which doesn’t require homing switches.

Will give the MDI settings a try and advise.

Is there an advantage to upgrading the Controller for use w/ C5? Do you have a link to a suitable upgrade board?

Newer boards have better EMI filtering and additional connectors for the newer accessories, and come in a nifty low-profile enclosure:

The boards are on a hidden page in our shop:

Okay, we’re definitely getting closer with those MDI settings. This time the part cut out as designed, however there is some variance in the tolerances.
When measuring the design across (Y axis), the design is 50mm, however with calipers measuring the part is 50.33mm.
When measuring the design across (X axis), the design is 90.076mm, however with calipers measuring the part is 91.72mm.
When measuring the design across (Z axis), the design is 3.70mm, however with calipers measuring the part is 7.91mm.

Any insight on how to get this dialed in better? Also, do I need to make those MDI changes each time I run a job?

Please calibrate for belt stretch:

Would I have an issue with belt stretch on a never used Shapeoko? I has been sitting, never powered on, since purchase 5 years ago.

It’s something which I do as part of commissioning a new machine or changing belts.

Do you have steel core belts? If so, contact

Okay, performed the calibration test.

I created a job shown in the illustration; 4 circles spaced 60mm a part. After running the job I measured the distances. Distance between X axis circles was 58.08 and distance between Y axis was 57.15.

So I performed the math for the X axis (shown in illustration) but than the gaps was even greater. Should I be dividing the actual by the assumed instead?

I’ve gotten the math wrong before (but had thought we had gotten it corrected on that page) — if it’s changed in the wrong direction, then yes, reversing the operators or the operation should get it right.

You want to multiply that ratio (1.033) by the current steps per mm (40).

60 (expected) / 58.08 (actual) * 40 (current step/mm) = 41.322312
$100 = 41.32

That said, I wouldn’t use your example design. Can you just use a V-bit and measure the jog?
Or cut a shallow pocket and measure.

In your example, the if the calibration is off a little, it might not show.


I don’t yet have a V Bit.

As you.mentioned above, would a simple 60 by 60 pocket suffice?

Yes. Keep the passes shallow to minimize any deflection. And it’d be best to run a finishing contour pass after the pocket is roughed out.

1 Like