DIY CNC Machines

[Schietwedder]

HI Guys,

any of you built their CNC machines for kite and foil or board making?
(Hot wire foam cutters, lasercutters, plotters or milling machines?)

Thought I share my design and progess maybe there´s more feel free to share your´s

Currently designing and building a 3x1,6m CNC X,Y plotter for my DIY foil kites. Maybe upgrading it to vacuum table and lasercutter later but first I want to get it work plotting.

Status right now: I use linear rails from ebay ( HGR20 for Y-Axis and SBR16 for X Axis) built a big table from ikea linnmon plates and mounted the big Y Axis traverse and X Axis bearings. I think I´ll redo the connection of the traverse to the bearings with a more sophisticated 3d printed connector though.

Lessons learned so far:
The HGR 20 bearings run a bit smoother and have less play and can transmit the moments better without blocking if not 100%alligned, so next time I woud use them for all axes, although it is a lot more expensive.
Cutting the rails is nearly impossible without an angle grinder because they are made out of (really) hardened steel, so I could not even scratch them with a normal metal saw or multimaster tool.

To do next:
Make some 3d printed parts and order all the timing belts and stepper motors and install the rail on the Y axis.

[Schietwedder]

.

[rynhardt]

I use a router sled with profiles to shape the board. It's like a CNC, but without the CN

It's not particularly accurate (+-1mm), but it's good enough for me.

[BalsaMichel]

Hi Schietwedder,

3x1.6m is a nice size

You will need to get some tension on the timing belts. Otherwise the accuracy will be bad, especially at change of traversing direction. In my opinion this is key to built a good system with timing belts. Also I would use at least T5, AT5 or HTD5 belts with at least 16mm width. As the motor bearings will also have to withstand the tension forces I would think about motors with 8mm axis diameter or additional counter bearings in line of the motor-shaft. Maybe you could 3D-print the toothed pulleys with space for an additional bearing.

The SBR16 are not bad for light hobby-mills. The problem with them is you need to have some luck because the quality of the guiding rods varies so much. No doubt the more expensive Hiwin HGR types should be better in many attributes like accuracy, load capacity, weight, more silent...

Have fun with assembling and building

Micha

[Schietwedder]

Hi Rynhardt,
Nice machine! We did a similar thing back in the days when building skis for tapering the wood core Can you adjust the shape of side rails aka X axis or do you make new templates every time?

Hi Micha,
Thanks for sharing some knowledge! Good that I design and order it step by step so I can discuss and adjust my design especially the X and Y axis drive mechanisms will bey key as you said.

I thought about tensioning the belts either by just mounting the end of the belts to the table in such a way I can adjust the tension (for example a 90° angular fitting with an elongated hole, or a printed device with some sort of a screw mechanism to tension it on both ends of the X Axis.

Yes the noise of the cars of the SBR 16 is quite significant. I saw some advice by removing the "China lubricant" on the bearings with a solvent and either run with no lubrication or a more sophisticated one.
What I came across also is that of you leave some play in the connections between cars and traverse like on a railway car it runs way more smoothly. I will maybe as well there mount some railway car ish suspension connection to the L beam (3d printed maybe with an insert for a ball bearing that also allows angular movement,so it can twist slightly in all axes)
By then having a stiff traverse in itself which is directly driven by the belt should not result in any major loss of accuracy for the Pen as the cars just follow the traverse I hope in theory(I'm happy with +-0.5mm)
Practise will tell and I'm not affraid of picking it apart and improving it several times as I learn the most that way I guess.

What I planned up to now with the belts (I'm pretty new to all this CNC stuff so I have no clue) was a 3m HTD 9mm belt with some 5mm diameter driven 16 tooth Pullly and some improved pullies with dual bearing (not the cheap China stuff).

But maybe I go bigger as you suggest!

BR Niklas

[BalsaMichel]

...
I thought about tensioning the belts either by just mounting the end of the belts to the table in such a way I can adjust the tension (for example a 90° angular fitting with an elongated hole, or a printed device with some sort of a screw mechanism to tension it on both ends of the X Axis.

Yes the noise of the cars of the SBR 16 is quite significant. I saw some advice by removing the "China lubricant" on the bearings with a solvent and either run with no lubrication or a more sophisticated one.
What I came across also is that of you leave some play in the connections between cars and traverse like on a railway car it runs way more smoothly. I will maybe as well there mount some railway car ish suspension connection to the L beam (3d printed maybe with an insert for a ball bearing that also allows angular movement,so it can twist slightly in all axes)
By then having a stiff traverse in itself which is directly driven by the belt should not result in any major loss of accuracy for the Pen as the cars just follow the traverse I hope in theory(I'm happy with +-0.5mm)
Practise will tell and I'm not affraid of picking it apart and improving it several times as I learn the most that way I guess.

What I planned up to now with the belts (I'm pretty new to all this CNC stuff so I have no clue) was a 3m HTD 9mm belt with some 5mm diameter driven 16 tooth Pullly and some improved pullies with dual bearing (not the cheap China stuff).

But maybe I go bigger as you suggest!

BR Niklas

Alignment of the two parallel rails is a game between accuracy and stiffness. The stiffer the components between the rails are the more accurate the alignment should be. If you add a "flexible element" on one side for example you´re reducing stiffness and the risk of high forces between the rails. I found that very often it´s enough to open all rail mounting screws a bit so the rails can slide in the hole pattern, move the axis from one endstop to the other and tighten the screws again. If this doesn´t work there should be an inaccuracy of the mounting surface and/or the rails themselves are maybe not straight. Then adding flexibility on one side could help.

The 3mm HTD is only available up to 9mm width if I´m right. You will have pulleys with small diameter what will lead to a higher/better gear ratio than with higher pulley diameters (HTD5 pulleys for example). But as mentioned the overall stiffness will be better with the wider HTD5, especially the 16mm wide then. A way to escape this dilemma is to use HTD5 and add a 1:2,5 belt gear on the motor side for example. A bit more complexity of course. As you want to build this quite big you might want to think about the used mill diameters. The small 3,17mm ones might lead to very high process time (some hours). So maybe 6mm will be a better choice. You can simulate this in advance for e.g.with Fusion360 or ESTLCAM if you like.

Maybe you can have a look into some forums/user-groups about Workbee, Shapeoko, XCarve and the Mostly printed CNC (MPCNC). They´re all having timing belts and I bet the same tasks have to be solved with long axis.

Micha

...EDIT: Or just use big stepper motors

[Schietwedder]

Hi Micha,

Don´t know if we´re talking about the same machine since you write milling a lot

I only plan to lay some spinnacker cloth on the table and insert a pen with a servo (Up/Down) instead of z axis to draw down the patterns, maybe later suck the cloth to the table with a vacuum and fit a small laser module to cut the cloth.
No milling, so I actually have ony the friction and inertia of the mass in the system to work "against". So I would actually look for a fast moving transmition and if it affects accuracy go a bit slower from there. Don´t know if that makes sense or I misunderstood something of what you wrote earlier.

The tip with resetting the screws sounds good, I will try!

BR Niklas

[BalsaMichel]

Hi Micha,

Don´t know if we´re talking about the same machine since you write milling a lot

I only plan to lay some spinnacker cloth on the table and insert a pen with a servo (Up/Down) instead of z axis to draw down the patterns, maybe later suck the cloth to the table with a vacuum and fit a small laser module to cut the cloth.
No milling, so I actually have ony the friction and inertia of the mass in the system to work "against". So I would actually look for a fast moving transmition and if it affects accuracy go a bit slower from there. Don´t know if that makes sense or I misunderstood something of what you wrote earlier.

The tip with resetting the screws sounds good, I will try!

BR Niklas

Haha, you´re right. I have read the initial post in short time in the train

So the stretching problem of the timing-belt will be not as bad as I thought.

Micha

[fluidity]

I have a CNC machine I built for making kiteboards. Took me a while to find software I liked at a reasonable price (Estlcam) and by then I'd got into wingfoiling so I've done a waterline cut 2650 wing and a very special set of wavy Delta monofoil moulds. The moulds I did with straight cuts and it's here that I discovered Z axis creep. Severe enough to make my first set of moulds worthless. for my second set, I put in a stronger Nema 23 motor, limited my Z speed but i still had to do a Z axis reset about every 100mm on the X axis of y axis cuts, to keep tolerances within about 0.5mm. Very tedious but I now have a pretty radical hydrofoil to ride. I'm not sure what the answer is for the Z axis other than trying to sort a solution just for that axis with positional feedback? I'm driving it with an arduino and both Z and Y axis are ball screw. The ballscrew rubber sealing washer popped out at one stage and I got it back in but ever since then I've been careful to lube the ball screws every half hour of operation or so. It's not my favorite activity to take my laptop out to the shed and supervise long cuts, If I can do the same thing on the 3D printer I much prefer. I have a 400 mm cubed 3D printer now which opens up a lot of possibilities for bigger parts that don't need to be cut. And a resin printer and a small metal furnace with some upcoming projects.

[BalsaMichel]

...The moulds I did with straight cuts and it's here that I discovered Z axis creep. Severe enough to make my first set of moulds worthless. for my second set, I put in a stronger Nema 23 motor, limited my Z speed but i still had to do a Z axis reset about every 100mm on the X axis of y axis cuts...

Hi fluidity,

I would bet you´re having trouble with the stepper motor / stepper driver combination. There are still some really cheap, bad drivers on the market. Unfortunately you can´t see without opening them. Many of them look just the same from the outside (DMxxx). Try to use a DM542T from stepperonline for example. Don´t know if it´s available in your country. Option would be a driver with a Toshiba driver chip. They´re a bit more expensive but will normally work more reliable than cheap ones.

Two years ago I´ve switched to the stepperonline DM542T and NEMA23 motors with low inductivity. This has solved my creeping problems and has increased my XY-max. axis speed from 3.000mm/min to 12.000 mm/min without loss of steps (spindle with 10mm pitch thread). Even too much for my nerves so I went back to 7.500

Micha