Fixed Gantry CNC Mill for cutting Aluminium and softer materials

Hi everybody,

So, this is literally my first post on this wonderful forum, so please be nice to me (English is also not my first language) :)

This post is intended as a build log (even if it's going to take a year to complete) and to gather experience from you, adjusting my design based on your input.

I am planning on building a small fixed gantry/moving table Mill for milling aluminium with reasonable precision, between 0.01 and 0.005 mm would be ideal if I manage with my low budget.
Work area will be 500x500x150mm, made with aluminium T-slot extrusions (I plan on filling them with sand or epoxy granite later to improve vibration damping).
I have already purchased a complete set of 20 mm square linear rails and ball screws (the chinese set, 1000, 700, 400mm , just like this one). No other purchases were made.
I am planning on using plain nema 23 motors 1,8Nm/254 Oz-in with TB6560 driver.

Gantry is 700mm wide, I plan on using a 80x160mm extrusion. Base is will be 1000mm long, made from 80x80 and 40x160mm extrusions, and 40x80mm traverses.
Z Axis design is still pending. Please let me know what you think.

Thanks in advance!

If you haven't bought that TB6560 driver board, don't. If you have, throw it out and start looking for something better. Look up the inductance on those motors - less is better. Too much, and you'll need to spend more money on drivers that can handle enough voltage to run them at top speed.

Your design is pretty basic, but I think it will be easier to align if all the Y-axis rails were in the same plane, instead of one on top and the other on the front.

You will not reach your desired tolerance (0.01mm) with your design and intended hardware.

That level of tolerance requires a lot to achieve. Linear scales, servos with high encoder counts, ground screws, very rigid frame, temperature controlled environment.

Quote:

---

Originally Posted by pippin88

You will not reach your desired tolerance (0.01mm) with your design and intended hardware.

That level of tolerance requires a lot to achieve. Linear scales, servos with high encoder counts, ground screws, very rigid frame, temperature controlled environment.

---

Yes, you are absolutely right, I don't know what I was thinking. 5-10 Microns accuracy .. I wanted to rival a 300k machine :)
Back with our feet on the ground, If I will be able to get 0.05-0.08 accuracy for a hobby machine, it will still satisfy my needs. Heck, I am able to achieve 0.1 mm accuracy with my home made FDM 3d printer, why shouldn't be possible with a home made mill?

I was first thinking about building the base out of UHPC but the thought that I could never be able to move that thing from place to place and out of the basement without help or specialized tools kind of scares me. Plus, building it from heavy duty aluminium extrusions and then filling them for better dampening, will give me more flexibility in adjustments and later on for transport if needed. I hope i will not regret this decision.

Now, I have another question for those that have already built such machines using aluminium t-slot. I want to bolt them together directly using rounded head hex screws like in the picture attached. Basically just drill a hole through the profile just to reach the screw with a allen wrench and bolt them together every 10 cm or so. Would this cause any issues ? I don't see how else could I bolt those 1m long extrusion together lengthwise.
https://www.cnczone.com/forums/attac...d=405318&stc=1


awerby: no, I haven't bought anything else except the linear rails and the ballscrews. I still have buy and build the aluminium structure until I will purchase the motors. But I am looking to see what else is there at the basic level, and all I can see are 3.5 mH phase inductance motors, and a cheap driver would be DQ542MA, DM556 or EMA2-070D56. I honestly don't know which one would be better, they all look like black boxes made in china, so I always suppose that the internals will be almost identical. Should I try to invest more in closed loop steppers ? The price difference is not that astronomically high.. maybe 30-40% more.

Thanks again for your input.

Yes, you can put extrusions together that way, with bolts and nuts. Make sure to drill out the holes on the nut side big enough to clear the socket, and use a thread-locker compound to discourage them from working loose with vibration. But there's no particular advantage to aluminum extrusion for a fixed bridge. It's not a moving part, so it can be as heavy as it needs to be - and heavier is usually better. If you got a section of heavy-section steel box tube Blanchard ground, it would be flatter than the extrusion and stiffer as well. It's a good idea to build the frame before starting to purchase electronics and motors; this gives you a more realistic idea of the forces required, and also gives you more time to research the various alternatives. You might look into an lightly-used automation actuator for your Z axis - they often have a couple of high-quality linear rails and ball screw, mounted solidly in an extrusion, for no more than the price of the slides.

I usually recommend Geckodrives, since they're proven to work well and come with good support - something that's lacking in most Chinese offerings - but I'm not sure what's available in Germany. Closed-loop stepper systems are worth considering, but they aren't the same as true servo systems. They can save your part in a catastrophe by shutting down, but when they wake up they'll be lost, and need to be re-zeroed.

awerby: I don't think that I will easily find those ground steel tubes in my area, but your advice made me rethink my design. I have a supplier of heavy duty steel tubes in walking distance from my home, and also a CNC manufacturing business where I could mill the grooves for mounting the rails at a relatively low cost. I will bolt everything together with screws, at least 3 for every joint, contact surface should be enough. I don't have a welder, and even if I had, my welding skills are non existent. I hope this will be rigid enough.

Regarding the Z axis, having already bought the entire set which has the linear rails and ball screws for the Z axis as well, I am going to use those. I am trying to keep costs down (I have mouths to feed :) ).

So, I came up with a second design. The gantry is made from a 200x100 mm 8mm thick stell box tube with 200x100x5mm legs. The base is made of 80x80x5mm where the rails sit and with 100x40x4mm traverses.
A static stress test shows almost half of the deflection for steel tubes with a 5mm thickness in comparison to a standard aluminium extrusion of the same external dimension. So I get a lot more stiffness and with less money as well.

One more thing: regarding the closed loop steppers. I knew that if they loose steps, they will compensate and recover, but how does this "shut down" work. They will give feedback to the motion controller when the lost steps reach a certain amount and cannot be recovered in time?
Re-zeroing with high precision to be able to recover a part isn't that complicated if done manually with dial indicators.. am I right?

Please let me know what you think.

https://www.cnczone.com/forums/attac...d=405596&stc=1

Quote:

---

Originally Posted by unromeo21

awerby: I don't think that I will easily find those ground steel tubes in my area, but your advice made me rethink my design. I have a supplier of heavy duty steel tubes in walking distance from my home, and also a CNC manufacturing business where I could mill the grooves for mounting the rails at a relatively low cost. I will bolt everything together with screws, at least 3 for every joint, contact surface should be enough. I don't have a welder, and even if I had, my welding skills are non existent. I hope this will be rigid enough.

[Bolts will work, if you use enough of them and seal the threads so they don't loosen up. In some ways they're better than welding, since there's no heat to distort the steel. Milled channels will work for mounting rails if there's enough thickness for them; you don't want to have the steel warp from uneven stress relief. See if your CNC shop offers grinding services as well. The other way to go would be to bolt some steel onto the tubes and mill the channels in it for attachment to your rails.]


Regarding the Z axis, having already bought the entire set which has the linear rails and ball screws for the Z axis as well, I am going to use those. I am trying to keep costs down (I have mouths to feed :) ).

[It sounded like you hadn't decided what to do, but sure, that should work.]

So, I came up with a second design. The gantry is made from a 200x100 mm 8mm thick stell box tube with 200x100x5mm legs. The base is made of 80x80x5mm where the rails sit and with 100x40x4mm traverses.
A static stress test shows almost half of the deflection for steel tubes with a 5mm thickness in comparison to a standard aluminium extrusion of the same external dimension. So I get a lot more stiffness and with less money as well.

[Right; steel is a lot stiffer and cheaper, if weight isn't a factor. You can also beef it up by filling it, either with epoxy-granite or fitted spacers bolted in place.]

One more thing: regarding the closed loop steppers. I knew that if they loose steps, they will compensate and recover, but how does this "shut down" work. They will give feedback to the motion controller when the lost steps reach a certain amount and cannot be recovered in time?
Re-zeroing with high precision to be able to recover a part isn't that complicated if done manually with dial indicators.. am I right?

Please let me know what you think.
IMG]https://www.cnczone.com/forums/attachment.php?attachmentid=405596&stc=1[/IMG]

---

[I've had experience with some hybrid systems, but each one is somewhat different in how it works, so you'd need to specify which particular one you're going with. In general, there's a band of steps within which the system can adjust discrepancies between actual and commanded position. But once that's exceeded, a shut-down error will be triggered. If you've got extremely accurate home switches, you can re-zero with them, otherwise you need to have a previously recorded home position you can re-etablish with your dial indicator, and an accurate idea of where it was in the program when the fault occurred to resume milling without discontinuity.]

[

Been working on finishing the CAD design (well, not quite finished, the steppers are missing). I made some extra support for the gantry legs, I don't know if this will improve rigidity much, but like this I can bolt the legs from both sides.

I am not 100% sure about the Z axis design, with the fixed rails versus moving rails and fixed carriages. What would be the most rigid setup? My Z travel is 160mm.

https://www.cnczone.com/forums/attac...d=405864&stc=1

What do you guys think?

Thanks!

you probably want to bring your Y rails in a little closer, you don't want the trucks all the way at the edge of the bed. The stiffness of the table is going to be greatest right where the it is bolted to the truck, so by putting them on the edge you make a longer weaker span in the middle, and throw away half of your stiffness off the edge where the table ends. You shouldn't be seeing so much load leveraging on the table that the wider stance is necessary.

Agree with extent.

Rails should be spaced more like:
1/4 table width - rail - 1/2 table width - rail - 1/4 table width

Yes, good point on that. I changed the two steel tubes from 80x80x5 to 140x80x5 so this gives me around 60mm on each side to move the rails inwards. So now I have around 120-260-120 mm between the carriages. Now, I know it's not quite the quarter-half-quarter rule yet, but close, and since I was going to use a 30mm thick aluminium plate for the bed, I don't think I will have much deflection to worry about on that 260mm span.
https://www.cnczone.com/forums/attac...d=406068&stc=1

Quote:

---

Originally Posted by unromeo21

Yes, good point on that. I changed the two steel tubes from 80x80x5 to 140x80x5 so this gives me around 60mm on each side to move the rails inwards. So now I have around 120-260-120 mm between the carriages. Now, I know it's not quite the quarter-half-quarter rule yet, but close, and since I was going to use a 30mm thick aluminium plate for the bed, I don't think I will have much deflection to worry about on that 260mm span.
https://www.cnczone.com/forums/attac...d=406068&stc=1

---

You look like you have a good cad program does it have FEA this would let you check the table as to how much deflection it would have, don't forget when testing if you can have the Ballscrew attached as this to will change the deflection of the plate

30mm plate is a good thickness, you most likely could of left the rails at the outside for this size plate, I don't believe that there is any such 1/4- 1/2 rule, good design does not have rules, just calculated load points that are a little more than what is needed before any deflection can occur, for the job that is being done, the greatest force you will ever have is when drilling holes, milling does not have very high down force, if there is you are doing something wrong

In your case you could calculate how much force that can be placed on the cutter before it would stall the spindle, when you do this you will see that there is not much force acting on the table, of cause you could design it to with stand the full force of what the Z axes motor and screw can apply, which is what would be the ideal number to work with, Table strength over support every time

A table like this can be made even thicker than what you have chosen and pockets machined out in the under side leaving ribs and pockets,which then could be filled with epoxy granite mix to help for a better machine build , I have been experimenting with this, to build a table that is a complete aluminum epoxy granite sandwich, which is one thick aluminum plate and a aluminum base plate, it may end up in the end more expensive than a cast-iron table but would be lighter and have the same damping affect as cast-iron would have

So I did some simulations (I have Fusion360) and for a 100Kg down force in the middle of the bed (in the first configuration, with the carriages at the edge of the bed) I get only a 5 micron deflection. This would be one and a half times my weight pushing down on a drill bit right in the middle. I would say, no reasons to worry at all.
Same 100kg force pushing the bit upwards into the spindle will result in a 0.05mm maximum deflection (here were the ballscrews not yet attached, so it might be less at the end).
I know, in real life we have much more play and the real deflection might be 50% more, but I am still happy with this result. I don't see any reasons to spend more for the wider tubes at the bottom.
I will go ahead and order the steel tubes cut to size. And a drill press as well.. .
https://www.cnczone.com/forums/attac...d=406412&stc=1

I would think that if the rail position was enough alone to make or break your bed deflection spec that your bed just wasn't thick enough full stop. But my thinking was more that no matter how thick and stiff your bed is, putting the rails not on the edge should let it respond better to vibration, rather than letting the entire middle of the bed ring. Could all just be made up.

With those numbers on the gantry you're just over 100k lb/in stiffness, which should be good for this type of machine. That's the guideline I used designing my current build and that's worked out really well.

Quote:

---

Originally Posted by unromeo21

So I did some simulations (I have Fusion360) and for a 100Kg down force in the middle of the bed (in the first configuration, with the carriages at the edge of the bed) I get only a 5 micron deflection. This would be one and a half times my weight pushing down on a drill bit right in the middle. I would say, no reasons to worry at all.
Same 100kg force pushing the bit upwards into the spindle will result in a 0.05mm maximum deflection (here were the ballscrews not yet attached, so it might be less at the end).
I know, in real life we have much more play and the real deflection might be 50% more, but I am still happy with this result. I don't see any reasons to spend more for the wider tubes at the bottom.
I will go ahead and order the steel tubes cut to size. And a drill press as well.

---

That was a worth while check 100Kg is more than your machine will ever see, for example if you could or used a 12mm drill in your machine, most likely never will this would require 72 Kgs of thrust force to cut aluminum,

You could add some aluminum bar stock like 16 x 38 under the table this would help to change the resonance in the aluminum plate

Allow to fill your steel tubes with sand or a epoxy sand mix when you have completed the build

Nice simple design.

The one thing you might find is the steel likes to ring. You might think of some kind of fill for some of the tubes. Even the cans of expanding foam will reduce ringing a lot.

Make sure you leave a way to tram the mill.

One small update on this project. I have changed my mind and bought a TIG welder.
I will first build a workbench and a small welding table to somehow improve my welding skills and then will move onto the frame for the CNC machine. But for now, just "playing" with some scraps, I am quite confident I am going to pull this out.
I will need to make some small adjustments to the design to make it more welding friendly and easier to take apart and transport the pieces to and from the CNC Center where I will get the surfaces flat and true. I want to have the legs completely removable from the base and gantry, with welded mounting plates which will bolted together.
Any thoughts ?

Regards