need to pick your brains on the makelangelo math

[Anonymous]

I’m working with the maslow project (http://www.maslowcnc.com, https://www.youtube.com/watch?v=y60q6U7NjTQ&lc=z13axtnwqpzrzv1ka23ujjbx5yi2dxc3l.1495696910660983) which is very similar to this in design, except it’s scaled up to run a 2.5HP router across a 4’x8′ piece of plywood to cut things out.

we are fighting some systematic error in our positioning (we are off by ~1/16 of an inch when cutting a 7′ long horizontal groove, bowed up in the center) and since you have a similar mechanical design, I was hoping to be able to compare notes and see if you can help us.

unfortunantly, maslow is GPLv3 while you folks are GPLv2 so direct copying of code isn’t possible, but I hope that figuring out the math aspects of things can be done.

On our side, the firmware file that contains the nasty math is at https://github.com/MaslowCNC/Firmware/blob/master/cnc_ctrl_v1/Kinematics.cpp

I tried to take a quick look through your code, but wasn’t able to spot where you do the equivalent.

David Lang

[Dan]

Hi!

I’ve had a look at the code. I’m immediately put off by the naming of major variable “l”, which looks exactly the same as the number “1” in most system fonts. But let’s put that aside.

1.5875mm (1/16″ is pretty small. If you are not missing steps and the math is good then there’s some extra variable for which you have failed to account. Does the position of the tool head change the contact point of the chain on the sprocket? Remember it’s not an actual point as it is probably treated in the math.

I also found that making flexible arms https://youtu.be/FJe5MkoUVIA?t=155 increased the precision as the line from the sproket to the tool stayed the length I expected. I didn’t have to account for the relative difference between the tool position and the chain mounting points on the anti-vibration disc. It also simplified our math.

Speaking of which, we used basic trigonometry and KISS: https://github.com/MarginallyClever/Makelangelo-firmware/blob/master/makelangeloFirmwareRumba/makelangeloFirmwareRumba.ino#L150
I don’t account for the sprocket size because our drawings are small enough that it doesn’t matter.

Does your machine auto-home? Are you certain you’re starting at the right position? A tiny error in homing can lead to egg-shaped or bowed results. That’s why the Makelangelo 5 has homing switches and gets great results every time.

Top center has the most tension on the chains. Is it bending the support arms that hold the motors?

I can relate to the entrepreneurial spirit of the Maslow team. I would never put a cutting head on a polargraph like this. I’ve seen to many accidents with polargraphs and routers to combine the two. I picture the router vibration shaking a critical screw loose, one chain detaching, and now you have a swinging router of death. OSHA field day. What I’m trying to say is please please please be safe, make sure your liability is air tight, and have a plan how to react when John & Suzy chair maker post their disaster story. Nobody wants a Streisand Effect, right?

[Anonymous]

I agree with you on the variable naming, but I didn’t write that code :-/

I agree there is probably some variable that we aren’t accounting for, I was hoping to be able to look at a differing implementation to try and figure out what.

We don’t have the lines always pointing directly at the center on pivoting arms, we have them attached to a fixed sled, that makes the math much more complicated (but the machine much simpler to build at this scale). It looks like this simplified your match enough for it to not help me, oh well.

The machine measures out a known amount of chain from each side for the home position, so it’s accurate to the limit of someone being able to set the sprocket with one tooth straight up (since it can measure movement of 0.044 degrees, there is going to be some minor variation, but it should be <1mm)

we’re aiming for 1/64″ (~0.5mm) in total error in the final project.

I’m not too worried about something getting loose, the router is attached to a large/heavy sled, and chain doesn’t twist well. Most of the time if it gets loose it will fall to the ground and stay put, it’s only if it’s working near the motor that doesn’t come loose that it will swing, and even then it will probably tangle in the chains rather than swing freely.

[Author]

Posts