Designing DIY CNC Milling Machine

Hey,

I wanted to say hello to everyone since that is going to be my first topic on the forums.

I thought for a while about writing here and ask for your advice on the CNC Router (re)build I am cautiously approaching.
To give you some context: couple of years ago I've I've built an aluminium-extrusion based router (see the attachment).

It works fine, but there are couple of issues that keep bugging me:
- General lack of rigidity - chatter in aluminium (but acceptable for me), steel requires a lot of courage and hearing protection.
- Accuracy or rather lack of thereof - axes are deflecting quite a bit under load. X gantry is not stiff enough to avoid sagging in the middle, and hte Y axis is linear bearing riding on unsupported shafts - and it's a joke.

Re-build I am considering has couple of primary goals:
- Keep the working area similar (~450x220mm with significant Z travel) since I need it to keep fitting certain job,
- Have enough rigidity to do some machining in steel (very conservatively but reliably) - I've got plenty of steel at hand and would like to use it.
- If possible, reuse components from existing build.

Limiting factors:
- I don't have welding table, so welding straight enough is with my skills is impossible - that rules out welded construction.
- Neither do I have access to milling machine big enough to go over the surfaces that linear rails will bolt onto.
- No surface plate.

Current rebuild design

Materials
Without the mill to machine everything, I am currently hoping to rely on a very simple, bolted together aluminium frame primarily made of off 20x30mm aluminium bar stock that might be flat enough to accept linear rails. All critical dimensions will rely on unmodified factory surface of that bar stock.
Worst case scenario, I might need to manually try to scrape it with indicator between rails. See the photo.
This frame won't be rigid enough on its own, so I am planning to cast epoxy granite over it.

Fixed gantry vs C-frame:
So far seems like I settled on C-frame. Decisive factor was that it is difficult to have rigid Z axis with ~200mm of travel in fixed gantry setup where in C-frame deflections are basically constant in whole working area -it  is always supported in the center of the frame and the Z axis overhang is static (compared to overhanging Z-axis in fixed gantry build).

Questions.
What's bugging me the most:
- Is aluminium bar stock going to be flat and straight enough so I can build frame out of it? I am going to test that empirically by ordering the stock I guess and testing it with straight edge, but maybe you have any ideas.
- I was hoping to reuse 15mm rails (HGR15 + HGH15CA) from the previous build. Are 15mm rails strong enough? I really like to reuse these, but 20mm rails are not prohibitively expensive so maybe I can upgrade. It always seemed like even 15mm rails the machine base in hobby build is always going to be the limiting factor.
- Epoxy Granite casting. From what I read this is only material worth considering, but it's so painful to work with due to epoxy. Just wanted to double check if there aren't any other alternatives (like some kind of fiber-concrete maybe?)
- Will epoxy-granite bond to flat aluminium or do I need to add some "studs" across the length to better anchor the frame into the casting?

Let me know if you have any thoughts. Thanks!

Did you see this project?


Consider a two-column machine. Built on the basis of a surface plate or table from a large milling machines Aluminum has a large temperature expansion. Do not put it in responsible places.
You will have a better result on large rails.

Large surface plate or milling machine table is very difficult to get by and very expensive where I live so I don't think it's an option really.

About the rail size - yeah, you might be right. I will only be able to reuse one pair of 15mm rails anyways and 20mm rails will help with clearance above the ballscrews.

This video is very helpful, thanks a lot!

Ditto to what @besriworld said - skip aluminum for structural components.

Everything is about stiffness.  The stiffer you can make the assembly, the better everything will work.

Suggestions:

• Use concrete instead of epoxy granite for the bulk of the frame and structural members.
  • Specifically, steel or basalt fiber-reinforced CSA-cement based concrete or mortar (Calcium sulfoaluminate cements)
  • Much less expensive than epoxy granite on a per-volume basis, and easy to work with.
  • Usually available locally.  (CTS Rapidset in the US)
  • Zero shrinkage, extremely high strength.  Very high stiffness compared to 'traditional' aggregate-based epoxy granite or aluminum.
  • Steel tubes or hollow weldments filled with reinforced CSA are extremely robust, especially if the frame component connections are designed to place the members in compression rather than tension the surface. i.e. through-bolting rather than threaded holes in the component's steel skin.
• Your frame components and connecting surfaces need not be perfectly flat after fabrication.  You can achieve the required alignment precision by using epoxy bedding to create flat/straight surfaces after the components are fabricated.
  • Using a filled epoxy product made for the process, you use a flat reference surface to cast a thin precision mating surface on the underlying imprecise component.  i.e. blubber up epoxy on a steel beam, and press a precision straightedge in to the curing epoxy.  Assuming you applied mold release to the straightedge, your warped, chunky steel beam now has a nice flat surface for a linear rail (or other precision component).
• Wherever possible, purchase precision rather than attempting to create it from scratch.
  • See if there is a granite countertop or memorial fabricator local to you that can handle custom cuts or thicknesses.  Assembling perfectly flat granite chunks together with epoxy & through-bolts makes for a stiff, well-dampened structure.  
  • Use wide-series linear rails instead of regular ones
    • All the manufacturers' literature I've read indicates wide series rails do not require clamping to a rigid reference surface to maintain straightness.  The base must be flat, but no need to have perfect side surfaces.  This reduces the number of challenging precision surfaces you must create in the frame components.
  • Use oversized linear rails & trucks
    • If you think (based on the rail/truck stiffness values you research) you need 15 or 20-series rails... go up a size or two.  35mm rails will add stiffness to the frame assembly rather than completely relying on the frame for support.
  • Space rails as far apart as you can stand - use geometry to reduce cantilevered components
  • Use more than two trucks per rail, and use more than two rails per axis.
    • You must be careful not to over-constrain components, but there's nothing wrong with using multiple rails other than increased friction (depending on preload) and cost.  Remember - balls screws need not be perfectly centered on an axis to work just fine as long as the axis components don't 'rack' and bend the ball screw.
• On a C-frame or gantry-type configuration, the Z-axis is usually the least stiff - either the column twists/bends, or the mounting plate/component the spindle is bolted to flexes under cutting forces.  You can build something insanely stiff, but if the mechanism that lowers the spindle to the work can flex/twist it'll still suck.
  • Give strong thought to a horizontal mill layout.  If that doesn't work for your intended uses, a sliding table, fixed dual-column, rising gantry configuration is about the stiffest thing I can think of that still has a 'conventional' placement of the work.

Do you have any specific resources to read about DIY CSA-based concrete machine bases? Never heard about it.

No 'official' resources I've found (like peer-reviewed journal studies) specifically on DIY machine building.

There has been discussion of CSA properties, as well as a couple of build logs of moderately informative content, on CNCZone.  The epoxy-granite and DIY sub-forums are littered with lots of back-and-forth debates, but there were a few threads in the past couple of years which included discussion of CSA concrete and attempts to quantify improvements in stiffness.

A number of concrete-only, or concrete-filled machines have been built by people over the years - there are even some YT vids available.  I think the fundamental problem with using concrete is that Portland-cement based concrete shrinks and cracks.  Using it to fill structures doesn't work well (far as I know) as the shrinkage decouples it from the outer structure and it just becomes a weight without adding any strength or stiffness to the assembly.  Even if cast around internal steel reinforcement it still cracks and shrinks.  Weight is certainly good, but it tends to be a by-product of a massive structure rather than the end goal.

Epoxy doesn't appreciably shrink when it cures, allowing it to stay in contact with the structure (or remain stable if it is the entire structure).   But epoxy isn't stiff - compared to metals, ceramics, or rocks.  So you add stiff fillers (which also reduce the cost) and, in theory, you get an improved bulk stiffness for the whole casting.  Problem is that without some trial and error and fairly tight control of the mix and placement, you will wind up with a heavy, but not extremely stiff casting.  Plus casting in really thick sections is a challenge without fairly specialized epoxy (i.e. overheating).

The deal with CSA cement is that it doesn't shrink like Portland cement.  It can also achieve extremely high compressive strengths very rapidly.  So can Portland-based UHPC, but those take rather a long time to achieve the high strengths and they still have the shrinkage problem.

CSA isn't extremely stiff (compared to steel), but adding stiff fibers to it improves things dramatically.  Fibers are typically added to UHPC for crack control, but a side benefit is that steel or basalt fibers increase the average stiffness... something desirable for a machine tool frame.

So the 'ideal' frame uses a high tensile strength outer skin, and a high compression strength internal filler... and design the geometry so that the skin is in tension and the core is in compression.  Then make sure the connection between the skin and core is intimate and will not dis-bond (or is mechanically coupled).  Make sure both the filler and skin are very stiff and you've got a solid assembly.

Not looking for any 'official' resources, this is hobby project after all.

This is extremely valuable insight, thank you so much. I know very little about _engineering_ cements and concretes, need to read more about this.

Here in Poland it's not really possible to purchase RapidSet CEMENT, only RapidSet CEMENT ALL which is a grout, so cement mixed with some mineral filler if I understand correctly. Not sure how much difference this is going to make.

Another question is whether CSA with steel or basalt filler will have enough tensile strength for concrete-only base with internal steel frame (with major structural connections working in tension, so tower will be bolted to base with 4 long through-bolts).

If you can get RapidSet cement-all that's fine.  It's grout, which makes it easier to handle, mix, and pour/place.

I don't think fibers add much to tensile strength, but I'm not an expert. If I understand the theory, the fibers are put in UHPC for crack control, and the added stiffness is a bonus we in the DIY-CNC world can make use of.  Steel fibers have the highest Young's modulus, but basalt aren't far behind and they're easier to deal with (lighter, wet out nicely).  Nylon or other 'crack-control' fibers would be useless in this application.

Tensile strength above and beyond the basic cement UTS comes from reinforcement - think of pre-stressed civil engineering structures.  Bridges and other structures have huge cables running inside the concrete.  Those cables are stretched prior to placing/casting the concrete and then the strain is released after the concrete is cured - putting the concrete in compression.

In the case of, say, a mill column, the same effect could be duplicated (sort-of) by casting hollow vertical tubes through the column and base.  Once assembled, steel plates are put on the top of the column (and below the base) and long rods are passed through the whole assembly and then tightened down.  The column and base would then be put in compression, but you'd need some pretty heavy-duty rods to get meaningful forces acting favorably on the assembly.

Not sure how to reinforce a whole column/base/saddle of a typical C-frame mill to withstand twisting forces, but maybe someone clever could figure it out.

I think it'd be easier, and the concrete less likely to get damaged, to use hollow steel tubes/forms and fill them.  But that's obviously dependent on one's skills and available resources.

Attached is a photo of my mill right after I filled the castings with Rapidset Cement-all.  After it cured, I bedded ~8mm plates in epoxy paste to box-in the cement and bolted the plates to the cast iron before the epoxy cured. (a 5-sided box isn't torsionally stiff, but the 6th side makes all the difference)

The result is, essentially, a monolithic lump of concrete & cast iron.  Massive difference in rigidity and vibration damping - deflection at the spindle was cut by more than 90% (sideways force applied to spindle box and deflection measured before/after the modifications).

If you (or someone else) do use Rapid-set, I highly recommend using as much fluidizer as possible, and mix with ice/ice-water to slow down the reaction.  That crap sets up FAST.