Own design - help needed

[boltzzz]

Hi,

I’m interested in designing my own CNC machine. Ideally, I would like the machine to be able mill aluminium in 3D shapes with an accuracy of +/-0.001" which I know is optimistic, so I’m looking for a rigid design. At the moment I’m planning on using ball bearing linear rails and ball screws on all axis as well as using 2 ball screws for x-axis.

I would like a cutting area of 3’ x 2’ x 8” but I’m happy to sacrifice cutting area for accuracy.

I have a couple of questions.
1. Aluminium vs steel for build material? Pros and cons? From memory steel’s stiffness is 3 times stronger than aluminium which is better but I want to use aluminium as it’s easier to work with and doesn’t rust. I also question using aluminium structure with steel linear rails and ball screws; I’m effectively building a bi-metallic strip!

Now onto my main question. Look at design 1 and 2 below. Yes there are multiple differences, but consider the designs to be exactly the same with the following exceptions:
Design 1 has no gantry sides as such (sides are there for torsion support only, not to raise the y-axis). Gantry beam is directly mounted to x-axis linear blocks. Also the cutter bit of the router is always at or below the x-axis rails.
Design 2 has gantry sides. The cutter bit of the router is always above the x-axis rails.

What are the pros and cons of each design? I thought that design 1 would be better as the moment arm on the gantry sides from the cutting force is minimal.
Where’s the best place to position the ball screws in both designs?
One last question is it better to mount the x-axis linear rails on the top of the beam like design 1 or on the side like design 2?

Thanks!

Similar Threads:

• Design opinions needed please
• Design check needed
• Design Advice Needed
• Help needed design
• Comments needed for this design...thx!

[ger21]

Ideally, I would like the machine to be able mill aluminium in 3D shapes with an accuracy of +/-0.001" which I know is optimistic

That's more than optimistic.

Steel is far superior. Consider that a similarly sized industrial machine designed for the same intent would weight thousands of pounds and probably cost a few hundred thousand dollars.

Design 1 is usually stronger and more rigid. Design 2 gives you better access to your work.
I'd probably put the ballscrews on the outside.

Rails on top is the preferred method. Especially with a lowered bed, where they are not in the way.
They are typically only mounted on the side with a flat table router, where they'd get in the way if they were on top.

[awerby]

A fixed-bridge moving-table design would be more rigid. The advantage is that you aren't constrained by the weight of the bridge; more mass is better. But they do take up more space.

Steel is stronger and stiffer than aluminum, and certainly a lot cheaper. But it doesn't come very flat - you generally need to machine or grind it flat before mounting rails to it.

[diepchess]

Hi,

I’m interested in designing my own CNC machine. Ideally, I would like the machine to be able mill aluminium in 3D shapes with an accuracy of +/-0.001" which I know is optimistic, so I’m looking for a rigid design. At the moment I’m planning on using ball bearing linear rails and ball screws on all axis as well as using 2 ball screws for x-axis.

Oh milling it at 0.01 mm accuracy is no problem of course if you construct machine well yet odds you manage that without spending first 50k euro and years of experience is of course complicated.
Also important question is where yo uwant to measure the error.

Size of workpiece or how straight you can mill a workpiece?

that 0.025mm is pretty good if we speak about straightness of a workpiece you mil at that 3 feet. That's really complicated.

I would like a cutting area of 3’ x 2’ x 8” but I’m happy to sacrifice cutting area for accuracy.

Well that is a problem of course. The only problem you have is rigidity of the machine like previous posters already mentionned.
Fully agree with them.

I'm sure you realize that cutting area and size of the gantry are 2 different things. You might be interested in a different sort of design, at least explore it as you want.

Just define clearly WHAT SORT OF OBJECTS from steel do you want to cut that are that huge?
Do you have melting oven there that produces such huge steel objects?

The cheap steel here has bars at a maximum size of 3 x 3 CM x 6 meters and there is 200mm x 15mm x very long.

I'm sure it exists yet a solid steel object of roughly 90 cm x 60 cm x 20 centimeter that will be pretty heavy.
9 x 6 x 2 x 7.8 kg/liter = 842 kilo.

That's gonna be a part of an ocean liner engine or something?

The heaviest steel object intend working on is 200mm x 15mm x a few feet. Or a mold which is gonna be way lighter and smaller.
So if we add the milling cutter to that, then we end up for the Y - axis of 200 mm + 2x cutter diameter + 2mm extra = 200 + 24 + 2 = 226 mm.

With in mind an ATC later on getting added i then take roughly 300 mm of movement for the Y axis.

I have a couple of questions.
1. Aluminium vs steel for build material? Pros and cons?

Well very simple - what is your budget?

Your machine will weigh quickly 2500 kilo if you buy in Europe aluminium as a guy who doesn't proces tons of aluminium a day,
that's gonna be easily 4 - 10 euro a kilo so that machine would cost you in construction then tens of thousands of euro's.

If you on other hand construct your thing from a few 2nd hand beams then for a 2500 kilo heavy construction you'd pay something like 60 eurocent a kilo
and for new steel you pay under 1 euro a kilo when buying that much. They cut it for free then (not very accurate though).

The real problem of aluminum is that it expands in a different manner than steel. Actually 2x worse than steel is not the problem. The problem is that you
combine at the machine steel with aluminum. So something is gonna bend there and end up not straight.

Suppose you have a bridge of aluminum that you constructed when it was very hot in Down Under.
Some of my nephews live there in different parts in Australia at temperatures i can't imagine they can still function as a human being!
Then in winter it's 20C colder. The aluminum shrinks 2x harder than the expensive steel linear rails. What does that do to your machine?
With aluminum you can be happy with 0.1 mm errors, and we aren't even talking then about the 0.3 mm or so straightness error (excuse my poor English).

So there is 2 obvious ways to construct your machine with steel.

1) just use a bunch of beams with bolds and some welding (or no welding in which case you will need to put in hundreds of bolts - closer to 1000 bolts than 100).
2) steel with concrete.

If you mix the concrete in the right manner with a little bit percent of epoxy, that also happens to have the same expansion a centigrade like steel.
So that's cool and nice and dirt cheap.

A combination of a few tubes filled up with some concrete and your thing actually will be able to handle easily 840 kilo loads to mill.

Of course with concrete as filler you'll end up with a way heavier machine. Maybe 3000 kilo or more - yet that's very great of course.
I assume you DO have a floor that can handle such weight?

If your house is from wood - it will go right through it of course.

That's way cheaper than building with beams.

Myself i'm a simple guy - i prefer building with steel only yet with such huge machine you should consider using concrete as additional filler and strengthener.
There is a reason why all skyscrapers get built from steel with concrete!

It might also dampen the noise better when using concrete.

There is some great explanations on different websites how to build a CNC gantry using a mixture of epoxy with concrete.

The actual % given of 10% epoxy is not needed. Over here a guy managed, by sieving accurately the right size of stones to be mixed into the concrete,
he used like just a few percent of epoxy - very little in short - into the concrete.

The real problem of using massive concrete structure is scraping the thing flat.

So price of concrete with some steel is unbeatable.

Construction out of beams with some strengthening plates is gonna be more expensive. I'm guessing around a 1500 euro or so if you'd build it here from new steel.

You need to choose this yourself.

I do not know prices of steel in Australia - some stuff there is pretty expensive out there.

The important question is: how much time do you have to build this one and what is your budget and what are your skills?
I
f you can weld - i know an obvious solution for you
And if you have time yet not much cash - go for steel with concrete.

My new plan here: a machine that has X,Y,Z :760 mm, 300mm, 300mm (or less in Z yet with adjustable bridge height for larger workpieces)
as the CAD didn't get made yet, only calculations on paper.

Yet the CAD will follow and i already asked funding for it from a microcredit fund. Hope they respond positive as i intend producing some great machines to sell (3d printers and robots and drone parts).

I only started this design AFTER i realized what sort of objects i need to produce.
So the dimensions it will be able to mill are a result of a need - not because i "like" to have such huge gantry.

Now this isn't a pure gantry and you should consider this design as well. It has a fixed bridge. So i have a milling bed that moves over X axis and the bridge is fixed.
the spindle therefore only moves over Y axis and Z axis.

Allows more accurate milling than a gantry - not by definition - yet practical.
This is only interesting for you if you limit the travel over Y axis a little.
2 feet would make it a big pain to construct the crosstable/milling table.

If you really want 2 feet rather than 1 feet at Y-axis, a gantry where bridge moves over X-axis it is.

I did do a drawing for a guy building a 5 axis machine using steel beams with bolts and a little bit of welding.
He has 2 meters x 2 meters.

Getting it stiff is not a problem. Finding the time to build it correctly is.

Whatever method of buiding you are gonna do - you need to manual , so BY HAND , make the surface very flat, so that the rails that runs over it, is very flat on the machine.
You do that by scraping.

In the end 99 out of 100 guys building their CNC machines end up too lazy to do that. So building a rigid stiff machine from steel and or concrete/epoxy
is a waste of time if you do not flatten the thing afterwards for the linear guideways.

In the end most are all too lazy you know to do that. Such handscraping if you do it first time could take you a week (hah months in my case as i wanted to do the work first 'by machine' to make the scraping easier - big failure).
Whatever stiff cheap construction you build it will be not straight. So you need to hand correct it to get surfaces straight.

Let's assume now you have scraped the X-axis at 1 side veyr well. Then the next problem is doing the EFFORT to measure the other side of the machine. Measuring at a few feet away is pretty big problem.

You want both sides to be perfectly level and straight. Preferably within 0.02 or even less millimeters at this huge distance.

I"m gonna do here an effort to get the steel beam scraped to < 0.01 millimeter so 10 microns, over length 1.9 meters from which 1.7 meters nearly is the rail length at X axis.
This for just 760 mm movement. So i'll be scraping most accurately that 1.7 meters for the X-rails.

In the end that just gives me about 760 mm of movement. So i scrape nearly 6 feet to get 2.5 feet movement

Yet please calculate the weight of my milling bed. First estimate here is that it is under 100 kilo's. That's not much huh?

You'll be busy half a year just with the construction of the base of the machine to get it stiff and scraped level and it is worth it.
If you want to do that effort of manual scraping the steel - you can end up with a very accurate machine for not much cash.

Total machine cost will be around 6000 euro anyway. Bit more with more accurate rails than H-quality hiwin.

On the spindel. Realize that runout is a big issue if you want to achieve < 0.02mm accuracy.
You probably want to buy spindle straight from factory in china. Half the price and P4 bearings. Around 600 US dollar. Add some 150 dollar shipment.

Will have not measurable runout (a micrometer or so at most).

After you have a very great structure for the X axis - the next problem is designing a good Z-axis. It is the Z axis where most designs suck most even though some simple measures can limit
the problem and/or fully fix it.

Then the positioning of the rails you asked in case of a gantry (which is not my advice to build except when yo ureally want that Y reach of 2 feet that you never need for steel except
when you are in ship construction). Well that's a dead easy answer. Realize that the construction that is on top of the rails will weigh hundreds of kilo's.
So you want the rails in a normal manner, not 90 degrees.

As otherwise a few small bolts of the rails is all that holds all that weight and that rails is just what is it 20-25 mm thick or so with a small bolt each 7 centimeter or so?

Never challenge gravity -as it really is there!

On other hand if the rails is on a flat rigid structure then all that weight pushes onto your structure and the rails gets massive support from the structure.
Realize the rails is not that perfect straight when it arrives. You bolt it onto flat surface. That bolting will get the rails flat and within the specs of the manufacturer.

Now choice of rails for such huge gantry. You soon look at the P type hiwin rails. That's the P from precise. Yet never order those rails prior to having handscraped your structure and measured how straight it is.
If you do not manage to get it straight ordering expensive rails is a nonsense.

If it is 0.1 mm or more off in straightness much easier is to order some woodworking cheap SBR25 rails and go for that.
Much cheaper.

That P from precise wouldn't do for a professional 5 axis machine. They have little option but to go for the highest quality. Yeah well such machine is a million euro anyway

So you look at a big price already for the rails in itself. What will it be, like 2x more expensive than H quality rails i reckon. Maybe 2500 dollar or so in rails alone.

Rails + carriage blocks will be most expensive thing.

From memory steel’s stiffness is 3 times stronger than aluminium which is better but I want to use aluminium as it’s easier to work with and doesn’t rust. I also question using aluminium structure with steel linear rails and ball screws; I’m effectively building a bi-metallic strip!

don't worry about rust. if you mill happily you need to cool while milling anyway, especially with steel. That means that the machine has a layer of water+oil on top of it. And as a good machinist you clean afterwards your machine and oil it. That'll keep away the rust!

Also the parts of thick steel that aren't protected anyhow. Don't worry about it. Will take a 100 years to rust away if it is in a dry barn

Now onto my main question. Look at design 1 and 2 below. Yes there are multiple differences, but consider the designs to be exactly the same with the following exceptions:

design 2 is for cheapskate chinese trader who wants to sell you a kit he buys parts in for a few bucks and make hundreds of dollars of profit.

Just calculate how much the deflection is with those hundreds of kilo's of weight you put onto it. Like 1 millimeter or so? Maybe 3 here and there?

You screw the rails for sure when doing this. For metal milling it really isn't an option at all to use design 2.
that is for lightweight small gantries that you mill some wood with. And only if that Y-axis length is under 1 foot or so.

Those popular gantry design with a single plate for the bridge part of the gantry that is as flexible like a rubber glove

Design 1 has no gantry sides as such (sides are there for torsion support only, not to raise the y-axis). Gantry beam is directly mounted to x-axis linear blocks. Also the cutter bit of the router is always at or below the x-axis rails.
Design 2 has gantry sides. The cutter bit of the router is always above the x-axis rails.

What are the pros and cons of each design? I thought that design 1 would be better as the moment arm on the gantry sides from the cutting force is minimal.
Where’s the best place to position the ball screws in both designs?
One last question is it better to mount the x-axis linear rails on the top of the beam like design 1 or on the side like design 2?

Thanks!

Design 1. Design 2 has so many problems - where to start?

[diepchess]

oh by the way - i avoided the usual ball screw discussion.

If you manage to build an accurate steel (with or without epoxy concrete) structure that is very flat.
You might consider putting high quality ball screws in it.

You can sell your thing for 50k dollar easily then.

I wanted to avoid this discussion yet because with accurate ball screws they charge prices that are as realistically as putting a human on Mars (a specific administration forgot we nowadays have robots i guess - probably doesn't fit in the communistic mindset to consider replacing workers by robots even though that makes perfect sense not only on earth yet even more at an unlivable planet where your blood boils instantly as planet is too small).

In this city basically everything is there if you have major cash. I asked THK factory in this city (Veenendaal - the netherlands) what price is of 2 ball screws. Ok i asked at not a perfect time - namely in summer. took a month for the guy to get back on me.
Then it was: 5000 euro excluding VAT and 3 months delivery time.

Now that 3 months wasn't the problem.

This was for small diameter C3 quality ball screws. Whereas the thing i build now cannot use such thin ball screws.
I bought on ebay some old ball screws that are new (never buy used ones). They are C1 quality yet rejected and when measured 20 years ago they were C0 quality and C3 quality over 1 feet distance and C3 quality over the entire 760 mm.

Only downside: yet that right now is a positive thing: they are 40 mm diameter.

Right now i'm overhappy with that with the new CNC design i'm busy with on crabble paper

In general a part in machining that doesn't have specifically mentionned how accurate it is - it isn't accurate. If you go use c7 ball screws you can forget 25 micrometer error of course.
First of all the chinese stretch the definition. If definition from C7 to C9 says 60 um (micrometer) error and C9 is say 120 um error then they act as if something under 120 um error is c7 quality.
So realize definition online gets stretched already by 1 quality grade there in quite systematic manner.

Obviously milling at 25 micrometer error is impossible just from theoretic viewpoint seen if the ballscrews you use are junk.

Now the additional problem is synchronisation of 2 ball screws over X axis.

Which is of course why i only use 1 ball screw over X axis and a milling bed that can be moved over X axis rather than the bridge carrying the Y axis and Z axis with spindel on it.
In short i also save out the cost then of a correcting measurement system. No measurement system at all here in the CAD plans. Just simply one accurate ball screw each axis and that's it.

If you build your machine very well and precise you can easily sell it for 50k australian dollar.

[boltzzz]

Thanks for the help guys. I definitely think that a steel frame is the way to go! I'm thinking of a construction technique similar to the MadVac machine - bolted, tapered pins and liquid steel for shimming between parts.

As my cutting area is going to be 3' x 2', it seems 2 ballscrews on x axis is they way to go to eliminate racking. I've been going from 1 ballscrew under the table to 2 ballscrews on the outside of the table through my head over and over. (see photos below)
I'm leaning more towards 2 ballscrews with one stepper motor and timing belts. A few reasons for this:
- The frame is a lot simpler.
- The table is better supported.
- Overall gantry height is less.
- Gantry weight is less.
- Can easily machine a part that's tall in z axis if needed. (remove the cross members and but it in between)

Yes 2 ballscrews is going to be more expensive (not too much as I'm leaning towards linerbearings2008 ballscrews), but I would hate to build a single ballscrew machine, have racking issues and then have to purchase a second one and install it under the table.
I really don't like the complexity on the single ballscrew under table design so I would hate to use it with two ball screws.

From reading all the 2 ballscrew x axis threads, it seems I should make the x axis 2' and the y axis (gantry) 3'. Why is this? It seems it's because the two ballscrews will be shorter and that any error between the ballscrews will produce a smaller angle because of the greater distance between them.
The disadvantage that I see it that this will increase the gantry weight and associated loads on the ballscrews/rails.