[Mwmx54]

I am planning to build a new cnc mill, and started designing one, but started thinking there’s gotta be a better way than the typical base and tower type machine, if I’m gonna build a custom machine, why not cherry pick the good stuff for the size I want, which is mainly small parts, but still want a large capacity, like X 30” travel(42”is table” Y 18-20”, and Z 24ish”. A little backstory, I have a heavily modified zx45(rf45 clone), that I keep modifying to make better, but at this point I’m thinking just start fresh.
So the questions I pose to the hive mind,
Gantry or tower?
Horizontal(better chip evac) vertical (easier setup)
Body(castings) epoxy granite(EG), EG filled steel tube, aluminum?
Table, steel(expensive and difficult to build) aluminum(cheaper, easy to build) EG(not very familiar with it, but easy to build and cheaper-est, plus super cool..) rectangular or more square?
That’s all to start, here’s some pics of my current machine, and the initial drawing. And my current project, a switchblade. I have some videos of the build here, mind you I’m not pro video’er.
https://youtube.com/@mwmxcnc?si=8toZuGJoHcuQstCr

[peteeng]

Hi MWM - A big mill build. I watch with interest. And its a very big mill. So you need to start at the end and research how you are going to mill the rails lands flat and level. This is the crux of that part of the journey no matter what material you choose. Even well made machines have come unstuck with the finish machining so you really need to understand how to do that. It will be a major pain to get it home and find its way off the mark as some people have found. So you need to trial it at the machinists somehow....

Steel is the easiest/stiffest and cheapest build material and is friendly. EG has issues and I think you should read some of the EG threads to get an understanding of its many pit falls. The 20" 500mm Y is getting deep so towards gantry territory. But I'd go moving column to simplify the drive system. Its more complex with a gantry getting two drives to co-operate and be accurate. A horizontal is wicked but harder to set up jobs. So from what I have investigated over the years here's my 2c worth 1) all fabricated steel but you need to commit to stress relief and final machining 2) thin steel plate (laser cut to parts welded) then fill with engineering grout, stay away from EG.. but still have to final machine steel inserts. 3) I would not use steel hollows they limit your design and have various issues. I'd use thick plate bolt/ dowel machined to size and bolt it all together or weld and stress relieve. It will be much bigger then you think it should be to achieve good stiffness.... Good luck on your mission. Peter

addition - the spindle will take up considerable time to sort. Especially if you want to machine steel. So before you design the machine you need to sort the spindle as this will dictate many parts of the machine. Peter

[joeavaerage]

Hi,
steel is, for a given sum of money, is the best for machine building. It as a Youngs modulus of 205GPa.

Epoxy granite has a Youngs modulus (by peteeng's careful making of samples and having them tested) of 18GPa at best, less than one tenth that of steel.
Even an EG machine can be made stiff but the sheer volume of EG will make it too expensive.

If you are of the opinion that EG is so well damped, then be disabused. First priority in ANY machine is stiffness. Second priority is stiffness. Third prioity....see the above....then comes material damping.
Damping is nice to have but if a machine is stiff enough it won't vibrate worth a damn anyway. If it's not stiff, even if it's well damped, it will make wildly inaccurate parts of crap surface finish. Stiffness is the only
standard you need consider...everything else is down the list, way down the list.

Next best after steel is cast iron. Youngs modulus of 110GPa with nice damping and good stiffness. Not cheap, but there is a very good reason that 95% of the worlds CNC mills are still made with cast iron....its damned good.
Then there is aluminum, either wrought or cast, 70GPa. Granite has about the same stiffness...if you want to be a stone mason. Then UHPC, with credible claims of 50GPa....if you want to be a block layer.
peteeng has been investigating various composite constructions, some have more promise than others, but neither do any stand out as head and shoulders better than any other.

Be very careful about nominating the size of the machine. I know everyone wants a big and capable machine.....but.....every time you double the size the cost goes up by a factor of ten.
In short, its much easier to build a much smaller machine that is stiff and be affordable, than to try to build a big machine only to realise you'll never be able to build it with the budget you have.
I don't know where you live, but here if I build an axis say 450mm -550mm in length I'm spoiled for choice at the number of companies who could put it in their machine to mill the rail beds for instance.
If I upped the size to 1000mm then the number of companies diminishes greatly, down to several...alll charging $200NZD/hour and more. Don't fall into that trap. You are much MUCH MUCH better off
with a small, even very small machine of good stiffness and accuracy than a big machine that's as soggy as a limp noodle.

Craig

[Mwmx54]

Good to know, I was under the impression that cast iron was the best, and maybe it is in terms of rigidity and vibration damping, you are right about planning on being able to machine the frame for the linear rails, I’d plan on doing it in sections, base, then column/gantry, then bolt the two parts together with a way to either shim or scrape them in square. It sounds like steel is the way to go, especially where I will be needing to be able to weld it together. What if I did steel, then bolted a machined bar of cast iron everywhere that needs to be flat(where the linear rails mount), on a thin layer of epoxy, so that I could machine the cast, bolt it on with epoxy between the steel, then scraped it in flat for the rails… I could also scrape in the steel and skip the cast, my thoughts are cast is easier to scrape and it would also give me the ability to machine it with a lip/datum for the rail to butt against.both have pros and cons, I’ll start designing a gantry machine and see if I can figure out which might be easier in terms of getting everything square and how I would machine the areas that need machined. Hmmm… option B is to buy an old clapped out cnc machine and rebuild it my way using only the castings. But I think that might be too big as the smaller machines where I live are crazily priced no matter the condition.

[peteeng]

Hello MXM - Cast Iron is not the "best" it is the most common for a few reasons. "Best" for a bespoke build depends on a number of factors. I'd consider beryllium or high modulus carbon fibre to be the "best" material for machine builds at the moment.

Multi material designs can get you into trouble, stay in one material. No need to scrape if you are going to machine. Although a little scraping maybe required if the machining is incorrect. I suggest you look at a moving table design like the Mori M1. The size of your bed implies that the billet could be very heavy. So a configuration in which the payload stays still is an advantage as the motors then can be optimised on just moving the machine not the billet. You can look at a gantry machine (a gantry implies that it moves vs a bridge which is stationary) but it will have challenges that the moving column solves. Good Luck in your chase of the grail Peter

[joeavaerage]

Hi,
scraping is good if you have 0.02mm to 0.05mm or less material to go, any more and its a joke. The process is to mill the surface flat, and if you need that last 0.01mm then scrape.

Epoxy has a Youngs modulus of about 3GPa, and it creeps over time. It is not suitable for mounting anything on a CNC machine.

You will almost inevitably have to machine parts, whether that be a few large parts or more smaller parts is a matter of choice......that they be machined is not. Machining parts to be flat
and square requires very good quality machines run by the best and most careful of machinists......it will never be cheap. Get used to it.

Many hobby builds are designed to try to get around machining and/or thermal stress relief, and almost inevitably end up as a compromise, usually a poor one. Its time to realise that
making an accurate machine WILL require outside help, and that help is not cheap.

My own solution was to cast three identical axis beds in grey iron, 115kg each, then have them precision machined and vibratory stress relieved. This I thought to be the best possible way to
get rigid axes of the highest precision at the least cost. Not that it was cheap. The three axis beds cost $10000NZD ($6000USD) to be cast, relieved and machined. By the time I bolted in the
ground ballscrews, linear rails and fitted a 750W servo to each they cost $16000NZD ($9600USD).

These axis beds were then bolted (and shimmed as required) to a steel frame, made of plasma cut 32mm thick med tensile steel.

My reasoning goes that even if a change the frame, or machine it or whatever, that the axis beds will remain the same. I made them as best I possibly could (afford) and they will last me for twenty
years irrespective of what happens to the frame and other components.

Steel is good and versatile if you can weld. Once its welded you will need to stress relieve. Thermal Stress Relief (TSR) is the gold standard, but not cheap. If you can find someone who has Vibratory Stress Relief (VSR)
equipment that is much cheaper.

I priced TSR recently at $6.70NZD/kg ($4.00USD/kg). It is not cheap but the results are good, and contrary to popular belief amongst hobbyists is a very well known and industrially practiced process.
You just have to get in and play with the big boys!.

Craig

[joeavaerage]

Hi peteeng,

Cast Iron is not the "best" it is the most common for a few reasons.

Complete and utter rubbish. Why are all the leading CNC manufacturers using cast iron?. Simple......its the best overall for rigidity and damping at a cost.
If you can afford to make out of beryllium then you don't need to build a CNC, just go and buy a Kern and be done with it. Aside from anything else because beryllium
is a neutron reflector, you just cannot buy it, its use is closely monitored by atomic energy authorities around the world. You'll know when you've pissed them off when a company
of heavily armed commandos (in masks and radiological suits) come to help you unload that big chunk of beryllium off the back of the delivery truck!!

Craig

[peteeng]

Hi Craig - "Best" of anything is a combination of factors (& someone's opinion, which they are welcome to have) and CI is, as I said the most common for a few reasons. I appreciate your married to the stuff and I would consider it in the right circumstances.

I did discuss getting beryllium powder from a company in the USA a couple of years ago. They were quite happy to sell it to me if I filled out the forms. It was uber expensive but available. Recycled Be was quite a bit cheaper then virgin. Comes down to your cheque book width or credit card size. Peter

[joeavaerage]

Hi peteeng,
beryllium is essentially toxic to touch.

Beryllium is toxic as both a skin irritant and an inhaled substance and can result in dermatitis, acute pneumonitis, and chronic pulmonary disease. The first signs of serious or life-threatening acute beryllium exposure may involve difficulty breathing, chest pain, or shortness of breath.21 Aug 2023

Any sensible discussion (of materials for hobby CNC) would discount beryllium on the basis of cost, toxicity and availability.

Craig

[peteeng]

Hi Craig - $$$ were not in the discussion and I appreciate Be is toxic, so is concrete dust and hydrocarbons like coolant. All materials need to be used in a safe & responsible manner. Peter

[joeavaerage]

Hi,
cast iron is a sure sign from God that he wished humans to indulge in machine building......just as he covered 3/4 of the earths surface with water, he intends that we spend more time fishing
than working.

Craig

[harryn]

I am not sure if it is practical or not, but sometimes I wonder if a machine like that should be built to operate close to the floor vs at table height.

[joeavaerage]

Hi harryn,

I am not sure if it is practical or not, but sometimes I wonder if a machine like that should be built to operate close to the floor vs at table height.

Why?

The pic I linked of my machine (above) which is now fitted with a trunnion table and fifth axis is about chest height now, which is a little high for convenience but not a deal breaker either.
My first mini-mill was on a table that put the X axis at about top-of-thigh height. It meant when you were setting up material you were reasonably close to the action and looking
down at it. It was very convenient.

My machine now weighs something like 800kg, so making a new table is not trivial.

Craig

[Mwmx54]

Well this is all great info so far, and I’m not dissuaded yet. I know this will be a long process, I like the idea of steel with engineer grout, the epoxy I referenced for placing the linear rails would be something like Moglice, which has a compressive strength up to 23k psi, not sure if that’s enough, steel is much higher at 100k-200k, I would use this method to avoid having to machine a large frame that the rails would be mounted to.
As for stress relieving, it looks like TSR needs about 1100f id have to find someone that could do that. I think ideally a bed height of around 2-3’ would be awesome.
Any recommendations on an epoxy that could handle mounting linear rails to? If not, I’ll find someone that can machine it, I just know that’s going to be a large chunk of change to get done perfectly.

[joeavaerage]

Hi,

Any recommendations on an epoxy that could handle mounting linear rails to?

None. Epoxy is too squishy whether it's filled or not.

If not, I’ll find someone that can machine it, I just know that’s going to be a large chunk of change to get done perfectly.

Yes, fact of life.

which has a compressive strength up to 23k psi, not sure if that’s enough

Strength is not the issue. Even modestly strong materials are enough. There are no forces to which our machines are subjected to that come within 1% of breaking.....but its not about breaking, its about stiffness.

For example: ordinary mild steel has a tensile strength of about 60ksi. Hardened and tempered 4340 has a tensile strength of 200ksi. The Youngs modulus, ie the measure of stiffness is 205GPa for both.
So while hardened and tempered 4340 is three times stronger than mild steel, they are as stiff as each other. Whether you make your machine out of 10mm mild OR 10mm 4340 neither are going to be anything like
near breaking strain, but they will both flex....and here's the kicker, they will flex the same

As for stress relieving, it looks like TSR needs about 1100f id have to find someone that could do that.

TSR is about 650C or 1200F. This could be done in a pottery kiln easy. A full solution anneal (from austenite) requires a higher temp, about 912C for low carbon steel, still within the bounds of a pottery kiln.
A spheroidizing anneal for steels with 0.76% carbon requires 729C, still within the bounds of a pottery kiln.

Craig

[peteeng]

Hi MXM- its not about strength its about stiffness. Epoxy is 3MPa and steel is 200MPa. So the epoxy is rubber. You cannot side step machining if you want to do the proper job. The forum is full of machines with issues because of trying to take short cuts. Even machines that have "proper" machining have had troubles. Epoxy is fine if used to "set" a joint but to use it for levelling is fraught with issues.

Yes; to TSR you need the steel to be red hot and that starts at 1100F or 650C. You can side step TSR by bolting the structure together vs welding. A bolted structure is damp and can be dissembled if needed. A machine of this size is a big project requiring lots of $$$. Its actually cheaper to find a second hand machine and rebuild. It will take a couple of years to design/build this (looking at other threads that have done this) and in this time you could find a machine and get a second job to pay for it

So : 1) find a suitable machine shop
2) find a suitable heat treater if you intend to weld. Not needed if you are making a shell and grout filling
3) look at spindles before you do much more. If you want to machine steel the spindle is a big fly in the ointment and will take quite a bit of sorting and $$$. Its size and complexity flows back down the machine and if you start at the base I reckon you won't have enough real estate for the spindle when you get there... So start at the spindle and work towards the base.
4) then decide how accurate you need the machine to be as this impacts lots of choices & $$$

Chase the Grail Peter

[joeavaerage]

Hi,
this is the crazy part, you've witnessed peteeng and I having a stouch over beryllium. Beryllium has a tensile strength of 54ksi, about 10% less than mild steel, but is 36% stiffer than mild steel!!

What beryllium is good for is as a 'pit liner', the 'pit' being the spherical plutonium pit of a nuclear bomb. I always keep some beryllium on hand for that purpose.....I wouldn't waste it on a mere CNC!!!

Craig

[jaguar36]

Any recommendations on an epoxy that could handle mounting linear rails to? If not, I’ll find someone that can machine it, I just know that’s going to be a large chunk of change to get done perfectly.

Precision Epoxy's SC-15P is steel casting epoxy designed for linear rails. As long as you use a thin (1/4" or so) layer of epoxy to mount your rails to, the stiffness of the epoxy (or lack there of) is not going to have a significant negative impact.

[peteeng]

Hi All - All epoxies bond to clean steel very well. I'm unclear why people would place a 6mm piece of plastic under their rails if a better way was available. Plastics being viscoelastic relax and creep under sustained loading (linear rail bolt loads for instance) so I can't see how over time; how an epoxy can perform well in this application. Especially since a flowable epoxy is a light molecular weight epoxy that uses a very low activity hardener (so it cures very slowly, so the surface stays flat) which means it can't achieve full cure unless post cured which is also not done. If there's a polymer chemist out there that can explain how this works I'd appreciate it because its against the usual knowledge and recommendations (from epoxy chemists I have dealt with) that has been made to me about epoxy over decades of using the stuff. It would be a last resort for me... Peter

Precision do not publish a mechanical spec for their resins... it would be good to know their cured hardness.

Epoxies modulus varies from 2-4GPa the attached snip shows various plastic modulus. MDF is stiffer at 4.5GPa....

[peteeng]

Hi all - I modelled a 6mm epoxy bed under a 20mm rail and torqued the screws to 11kn in Fusion. Cap screws would usually be done to 60-70% YS torque which is 16kN but I did them up to what a 10.9 screw would be. I also ran the same model with the foundation in steel. The interface between the rail and foundation is a separating surface friction co-eff 0.3...

The epoxy bed has a "wave" of 7um the steel bed is 4um. So for a hobby level machine I suppose thats OK. The epoxy is 4GPa the steel is 200GPa. My concern is over time is that the epoxy will relax and creep and the wave will get worse... The spacing is 60mm and the bolt is M6... Peter