วีดีโอ

I think the wight doesn't matter too much. It's more of a size question. The movements are X300 Y200 Z200 mm.

@@tw2ka644 ok boss

Thank you! Theoretically you can hit sub 0.01 mm accuracy with the machine but in the real world the tolerance zone is propably 0.05 mm. During testing I have measured a aluminium plate with holes cut in to it and the locations of these holes were off by 0.02 mm.

​@@tw2ka644 Impressive! Thanks for your reply

That really sucks, I haven't experienced similar issues with my A8. Hopefully you'll be able to sort it out!

It's propably a little too weak to actually machine with. Maybe something like wood with light passes could work.

Thank you!

Outside measures 300x110x80mm and vise jaws open up 120mm.

Thank you I appreciate it! I trammed the Z-axis by placing a 90 degree angle gauge on a vise in a way that one side was facing up. I first zeroed the gauge with a dial indicator so one side was in line with the Y-axis and then attached a dial indicator to spindle and moved Z up and down dialing against the gauge. This way I found the tilt on the YZ-plane. Next I did the same thing but zeroed the gauge on the X-axis and then dialed the Z movement on the XZ plane. I corrected the tilt error of the Z-axis column with little metal shims. Then I went ahead and trammed the spindle by placing a long end mill on the collet. I again moved the Z-axis up and down and measured the angular error from the front and side and corrected the spindle. Hope you understood this and yes, I am a mechanical engineer who specialized in automation and robotics!

@@tw2ka644 Great, yes I understood. The shim is what I was after. I didn't know if you went with that or a metal filled epoxy or some other method. I'm an ME also, but most of my time has been designing plastic parts for injection molding. Here is my cnc design playlist if you are interested. I'm still in CAD, slow going, ha ha. th-cam.com/play/PL_e9tdItiJstrB6WyruH1jfdonHBbIJdD.html

@@nickp4793 Nice to hear, I will check that out!

Yes, I was worried about it. I tried to shim the welded frame so it would lie flat against the supporting pieces before clamping down. I think there still was some error in it. I wasn't too worried about it because I later went and measured the surface and then used a scraper to scrape the surface flat. 3 points of contact is the way to go if you can get that kind of a setup to work.

Thanks!

Thank you, appreciate it!

Shaft could maybe be printed but it would result in a weaker joint.

It was ebay/amazon. Search for HTD pulley and you should find similar products.

@@tw2ka644 I'm also struggling to find this component

Great to hear, thank you!

It's just a push fit. The tube only acts as axis and doesn't transfer torque. The actual pulley is bolted to the robot arm, tube is just a pivot point.

Great to hear, you're welcome!

Great to hear! I have a video coming on the finished mill soon where I've gathered some thoughts and details about the project. I hope it also helps.

Yeah! The cnc is fully operational now and I've been learning it over the last month or so. I'm making a video about the completed mill right now and it will be published once I finish editing.

@@tw2ka644 excited, thanks

Thank you it's great to hear all that! It is a really rewarding project building something of this magnitude. I'm in the process of making a new video as the mill is now complete and fully operational. Will be posting soon.

Hello, there was one dedicated stepper driver and the arduino cnc shield had places for five more drivers. In this video the sixth axis driver wasn't installed so only four of these slots were utilized.

I sold this project forwards a while back. It got to a point where it only needed software development and it's not my main area of interest.

Alright, thanks for the answer. Pitty to hear that we won't see it moving. But totally understandable ;-)

Yes you could build one but depends what you mean by ultra precision. My machine is accurate to around 0.02mm (on axis flatness, straightness and perpendicularity) and this kind of accuracy is achievable using conventional measuring devices. Having a known flat plate and dial indicator for checking flatness and form is a must in this case.

@@tw2ka644 .02 isn’t bad if that’s end to end or something, but yeah, I’m shooting for maximal precision, I’ll even scrape the way surfaces for the rails if I have to, I can shim them for final coplanarity if I need to, I just don’t have access to lasers and complexly metrology to make getting ultra precise measurements easy. I have dial indicators with super high resolution, surface plate, etc… I guess I need to draw something up in cad first, I want a fixed epoxy granite mill with a metal skeleton. I want to go all out on the linear motion components, insane stuff, c3 ball screws, ultra fat and rigid rails, everything super overkill but compact.

@@CryoftheProphet Yeah it's basically .02 across the rail mounting surfaces. All the axis are straight but I still need to shim the gantry to get Z-axis moving perpendicular to the XY-plane. That's the final step. Your project sounds great! Scraping the surfaces is definately the way. If you have access to a good milling machine for facing the mounting surfaces that helps a lot.

Hi! Open loop worked well for me. As long as the load on the end of the arm isn't too big, the steppers won't skip steps. At the time I went for open loop mainly for the simplicity. Nowadays there are good closed loop motor-driver combinations where the stepper drivers themselves check for the correct amount of steps taken. These systems don't require external encoders and only need step and direction signals from controller. You can see this type of system in use on my 3-Axis cnc milling machine project. If I remember correctly it is a 80 tooth gear with a 16 tooth driver so the reduction ratio is 5.

Glad to hear! I also thought the original audio fit this video best.

There was definately some luck involved, measured the 50mm length with a tape measure off the side of the blade. I'll take it tho!

That is indeed a viable idea! Would ve helpful not having to calibrate the robot. Thank you, nice to hear!

2018 version, can't remember the exact version but it was a 2018 solidworks build.

Lack of time and space was the main reason at the time. I also reached my limit in coding knowledge at the time which prevented me from further developing the software.

Thank you! The expenses are all listed out in the excel sheet found in the description.