Squaring a CNC Gantry Machine with Simple Tools and Math

[@BryanHoward]

It's not a fan video! but surly to blow you away. Python code shown in this video can be downloaded here: github.com/bhowiebkr/gantry-square/blob/main/main.py

[@ferriswhitehouse1476]

I'm so happy I found your channel. So many brilliant ideas that make things accessible without spending a fortune on precision equipment. Thank you!

[@BNETT21]

I think you'll enjoy this method of squaring. Look up "the 4 cut method crosscut sled". It's a way to simply take 4 straight cuts then take some measurements and do some math. It's a woodworking method for dialing in crosscut sleds to have a really accurate cut in relation to the guide slots on the table.

[@tonyrichmond9428]

This is pretty clever. I've been. Making holes at corners of 6in squares and sticking old endmills in them, then measuring those. Reading through the comments, seems like a lot of people need to review their geometry. Any parallogram with equal diagonals MUST be a rectangle. Complaints about making the tool don't seem to consider that it doesn't matter how ACCURATE the reading is, only the PRECISION, because you don't be to calculate the square, you only need to verify the equality of the diagonal. The source of accuracy improvement is in the fact that the three absolute error of the calipers doesn't change, so the PERCENT error over the much longer distance goes WAY down, assuming the tool can be placed precisely. The locator pins are about as good as you can expect. The only real complaint that holds any water is whether you've ensured that your steppers deliver the same error on each side, that tbe error is simar to the, and that the error is repeatable in both directions on all axes. Imagine am axis set to 40steps/mm vs another set to 40.5, or an axis with significant backlash. Also, whats the repeatability of the switches? It's probably reasonable to assume that all of these errors are within the margin of what you can actually measure and/or the accuracy of your calipera, but if you're claiming 0.02mm accuracy, you could probably measure and do the rest of that math to back it up. Really clever technique to use a small tool in a big way, though. I like it.

[@jonm189]

This is ingenious, I have never seen this method before. It looks really simple and effective - I'll be squaring my gantry this way when I get to that point in my build. 🍻

[@BryanHoward]

When I squared the new gantry I didn't bother to run the software and just winged it with the tool. It works super fast that way if you can easily adjust the end stops. I made a small gcode script for drilling the 4 holes which saves a huge bunch of time vs doing with sending commands manually. I'll post the gcode script in the description later today.

[@mikenewman4078]

I wish I could give this more than one thumbs up.

[@FilterYT]

This is a great tip, thanks! (3 - 4 - 5 btw)

[@ChristophLehner]

Really smart way 👌

[@giuliobuccini208]

Brilliant! I used the 3-4-5 method (Pitagora's theorem) but your is way better because you do not need a precise and long ruler.

[@BryanHoward]

Yeah this method works well, you don't even need to bother with the math and instead just keep adjusting till the difference is negligible. Writing a simple gcode program to make the dots also speeds up the process and makes for a good first gcode by hand script.

[@andreacerroni9010]

Hi bryan, I wanted to ask you how you made the holes on the spoilboard, with what type of bit (V-shaped?)

[@rafal7217]

"I have a question: Is using two limit switches a good solution? These inductive sensors have a certain measurement error tolerance, I suspect it could be even over 1mm. Wouldn't it be better to use one limit switch and synchronize two motors with each other using a belt, and calibrate them on the coupling?"

[@rafal7217]

@@wyattutz Do you have any reliable sensors from AliExpress that you know are okay and recommend? Could you drop a link?

[@toddspeck9415]

Super curious if you are self taught in solidworks? Or did you go to school for cad. I tried out Solidworks. It was a bit out of my price range. So I bought Viacad 12 pro, Moi3d, Alibre atom, qcad, turbocad. And I taught myself Blender, And I have a no save version of Rhino5. I really enjoy learning all these programs. But I always am amazed when I see what Solidworks can do. No wonder most colleges use Solidworks.

[@ferencbordacs2319]

Great idea!!!!!! Super!!!!!

[@KarlMiller]

Great tip. Are your Y rails parallel?

[@BryanHoward]

They're planar. I used a master precision level to calibrate on the right side. Then shimmed the left side so they match.

[@KarlMiller]

@@BryanHoward Sorry, I did see you establish them being coplanar with the level, but being parallel to each other. I asked because, if they were not parallel, I wondered how effective it could be to move the endstops to adjust squareness of the gantry if the frame wasn't square. I would imagine you only could compensate just so much for Y rails being out of parallel by adjusting the endstops.

[@mikenewman4078]

This extended caliper would be useful for setting rails parallel during initial calibration with a small modification to support the span. Diagonals are then the final check. Thanks everyone for the inspiration. I'm sure this would also work for squaring a column too. I love simple solutions.

[@alterstassfurter5743]

Don't get me wrong.I can't trust your way. Because you build your measuringtool by yourself. 1st point of inaccuracy, next is the count of your measurings. With every single measuring you are generating measurement errors. in summery you are going wrong remember: Anyone who measures a lot measures crap 🤣