[henrychan]

So, I'm building a CNC mill and I hope to cut steel with reasonable speed and quality. Word area is about 1000mm(x)*500mm(y)*500mm(z)

The machine will be mainly constructed using 3" .25" thick steel square tubes with .25" steel plates as reinforcements.
I am still at a very early stage of designing the machine, but some rough calculations and FEA of the frame assembly gives me about 1 thou deflection along x and y axis with 150lbs applied force.
Is that enough rigidity? I really can't find any useful information about how much load needs to be designed to withstand. Some sources say 200 safety factor while others say 1 million lbf per inch of deflection.

I also did some calculations to see how much force each member is subjected to if I were to make Z work area 750mm. It doesn't look that good, looking at about 4 thou deflection under 150lbs load on entire assembly.

Thanks.

[arizonavideo]

You might give us a little bit more to look at.

Steel is quite hard to cut for a light mill. Large bits will generate much larger forces so will flex the mill a lot. Smaller bit will cut better but break easy if the mill flexes, which it will.

So what kind of steel do you plan on cutting? Is that the main use? If so the mill needs to be 3 or 4 times stronger.

If you only want to do 1/8" steel plate rarely then a 1/4" cutter might be OK and 5" box section frame sand or cement filled might be OK.

Your Z it too tall right now unless you have to have it that way.

Dave

[henrychan]

I would say the machine will mainly be used to cutting aluminium and cuts steel occasionally. I want it to be really rigid so I can make deeper cut in aluminium too.

Right now, the y gantry is 18”x12” 1/4” steel box section with 3” square tube at each corner. After some though, I think I’m going to shorten it a bit.
I hope to leave the Z as tall as possible as I plan on adding a 4th axis later on. However, I do see why that is quite a big challenge as every bit of z height I add requires exponentially more reinforcement.

Do you have any specific deflection/rigidity numbers that I should be aiming for, or calculators to help me get them? It’s easy to figure out how much force is needed to cut the steel, but really hard to find out how strong the vibration is going to be.

[ChrisW]

Here is a dumb idea.

Fill the tubes with hi density concrete.

Or why not use solid steel..

Sent from my MHA-L29 using Tapatalk

[peteeng]

Hi Henry - welcome to the forum. Machine stiffness is discussed here in a few spots and there seems to be no good source of absolute deflection specs around. So the answer is as stiff as you can make it within the envelope you have. Figure it out then make it bigger and thicker Look for a similar machine to see what its dimensions look like. FEA is helpful, what program are you using? By the way your better off using thicker tube then filling things with concrete. Concrete is mentioned a lot here. Its half as stiff as aluminium 30GPa steel 200Gpa, will crack up inside things as it shrinks and loss continuity. So a thicker steel solution is better. Since you can do FEA you will sort it faster than most. Lots of cross bracing. By the way if your using linear bearings you will need heavy preloaded versions not the std light preload. You are also better off fabricating large shapes rather than using lots of tubes. Your gantry is shown as two tubes better to be one. Use as much geometry that you have for each loadpath. A 25% increase in size doubles your stiffness which your gonna need lots of.... Peter

[henrychan]

The reason why I don't use solid metal is because of cost and logistics. I cannot spend more than $1000 on the frame and I also don't have a garage nor a car (college student rip) so that kind of limit my choices (everything has to be delivered to my apartment and fit inside an elevator). It is also nearly impossible for me me to carry 300 lbs of steel to the machine shop at school so I prefer to keep individual components relative small. I found out that using 1/4" steel plates to create a box achieves similar results as cross beams while being cheaper and easier to manufacture. All I am looking for is some sort of numbers that would even remotely reflect real world performance, like is 1 thou deflection at 150lbs load good, bad, or terrible? It is impossible to me to design a machine without knowing what loads it will take. I'v also redesigned the gantry a little and now it uses angle iron which are much cheaper, deflection still sits at a little over 1 thou. (https://drive.google.com/file/d/1dQ9...ew?usp=sharing) The gantry is about 10"x10". Does it look like it will work?

[peteeng]

Henry - Given your situation why are you trying to build a machine? You can contract your parts much easier than building something that will be underpar in performance and have no resell value? If you have seen a 1000x500x500 mill it certainly is not something that can fit in your apartment. If you make decisions based on angle iron is cheaper then you have lost the game. You are using a linear solver in your FEA work. This will not predict the shear deflection of an angle iron and in reality angles twist under load becoming inefficient. Sounds like your a student so look up shear centres of sections. This can only be predicted via a non linear solver approach. You need to design whatever you are building and contract the build, you will do much better.

If you want to cut aluminium a typical chip load is 0.1 to 0.2mm (0.1mm is 0.0039") if the cutter deflects more than that it will not cut so look up some cutter force charts and see what a cutter force is and the chip load and then you will have some numbers to digest. Its scary!! I've seen medium routers spec'ed at 150lbf cutting and that's for timber perhaps aluminium. So there is your 4thou at 150lbf so you have failed as the tool will slide past the cut and just rub rub rub. Typically if you want to achieve 0.1mm then you have to aim at 1/10th of that so you want 0.0004" at 150lbf for instance. I haven't measured mills for direct loads but I have measured drill presses and they have over 150kgf (300lbf which you are going to need for a plunge) Steel is a new ball game. One poster in another thread said his mill produced 1 tonne of tool load. Good Luck Peter

[pioneer-tech]

Steel is bad for cnc frames because it "sings". It amplifies vibration and has bad dampening characteristics. What You really want but I don't know how to get is grey cast iron. It is used on most metal cutting machines because it's molecular structure actively dampens vibrations ( don't ask me how ). This may only apply to really big machines, I don't know but I know it's the best frame you can get. My Haas VF3 is grey cast iron. They don't do this because it's cheaper than steel. They do it for it's vibration dampening properties. This may not help but maybe it will.

Just an afterthought My cnc weighs 12,500 pounds so you can see what it takes to get the required rigidity.

[henrychan]

Thanks peter, that totally makes sens. I am trying to build a machine because
1. it's much more convenient to have my own machine
2. I can do stupid things on it without being yelled at
3. I will convert it into a 3D printer and laser cutter with some inexpensive modifications.
4. It's dope and I will learn a lot in the process
I redesigned it to use mostly solid 1" steel plates (they aren't toooo expensive if I keep it small) and significantly reduced the work volume down to about 18"(y)x 12"(z). Using my FEA solver, I get about 2/10th deflection under 150 lbs or a little over 3 thou at 1 tonne on the gantry in both x and y directions(add about 2 thou at 1 tonne if I include the entire frame assembly). The linear solver is all I have, and I believe it does a decently good job at solving systems with small deflection (correct me if I'm wrong, I am absolutely no expert in FEA). Also do you have a link of the article talking about 1 tonne tool load?
https://drive.google.com/file/d/1O1N...ew?usp=sharing
Does it look like it will work and do you spot any glaring design errors? Thanks a lot.

[henrychan]

Cast iron also seems like a good idea. Should I also fill the rectangle tubes with sand to dampen the vibrations?

[MillNut]

This PDF might help you answer many questions you have and some you haven't asked yet. There is a wealth of information on "stiffness" and "damping" in welded steel construction precision machines.

http://www.mech.utah.edu/~bamberg/re...e%20Design.pdf


Sent from my iPad using Tapatalk

[henrychan]

That is actually super useful. THANK YOU. Guess I have something to do during my liberal studies lecture now

[peteeng]

Thats a great article -Thanks Millnut.

Henry why don't you buy a small mill as a starter? You then can make parts, learn about machining and get on your way. In US should be about $1000 USD? You can CNC convert it in time and do good things? Your heading down a path that is very difficult for someone in your situation and state of knowledge.

Hysteresis - Materials have internal friction called hysteresis. If you get a paperclip and bend it backwards and forwards a few times then feel the middle you will notice its hot. This is the heat due to friction of the steels grains rubbing together just like you do to keep your hands warm. Steel has small grains very well packed together and this slippage is quite small resulting in a low hysteresis material. Also the grains all have the same elastic properties so they move together in harmony sort of. Cast iron has large grains and also has graphite particles within its grain boundaries. This allows the grains to slide past each other quite a bit more than steel so they have more internal movement and friction then steel has. The iron and the graphite being different modulii also means they move differently to each other further contributing to internal disharmony and increased damping. Epoxy concrete and fibre composites rely on this effect to be damp as well. Their different materials move differently internally so as to create internal resistances to motion and therefore have high hysteresis.

One strategy used by aircraft against vibration (particularly flutter look it up) is to use metal laminates. Airbus make their fuselages out of "GLARE" an aluminium laminate. So Henry, consider creating a bolted metal laminate mill. a) the laminates being dismantlable are lighter for you to lug around b) they are damp. So your 25mm steel could be 4x1/4" steel or Al - in-plane they are the same stiffness as 1" solid, in bending if you bolt it correctly it will be the same in flexure stiffness as well. Being 1/4" thick you can get a local sheet metal place to bend it for you to create bolting flanges vs trying to bend 1" etc etc. You can also tailor the thickness in different areas easily vs having to mill them thinner. Look up a laser cutter/bender near you its a cheap process if you get the right company. More like 3D printing approach in layers even use 1/8" ??Peter

Also do not think that a bolted up structure is not as good as a welded structure. Welds and bolts/rivets are just different concepts to get the job done. There are heaps of old bridges still standing that are riveted and bolted together. Aircraft are bolted together. If you welded them they would fall out of the sky due to fatigue. So think it though you maybe able to build your dream machine after all.

[peteeng]

Hi Henry - I've just read the thesis linked prior. Excellent case study and summary of how a machine should be designed. The machine in the study uses viscous damping via layered tubes to dampen vibration. I mentioned that you could use laminated metal to build your machine. If you place a thin rubber sheet between your layers you have an excellent damp structure. Perhaps the start of a new technology for machine building and your just the person to get it off the ground. If you are not breaking down the structure for transport then epoxying them together or contact cement is the way to go. I was going to look at this for Maximus. 2mm zinc anneal is the cheapest material to cut via laser. So a ZnA laminate is the process I've been looking at. Peter

[peteeng]

https://drive.google.com/file/d/1O1N...ew?usp=sharing
Does it look like it will work and do you spot any glaring design errors? Thanks a lot.[/QUOTE]

OK why do you have two beams for the gantry? One beam the size of the two beams will be much stiffer. Every component must take up the maximum space made available for it to maximise stiffness. And the thick top and bottom of these for the rail connection, how do you intend to achieve this? Peter

[peteeng]

Hi Henri - I mucked about with laminates and learnt a bit. The images attached are of a 20mm thick solid and 20mm thick laminate. The laminate has similar vib frequencies and shapes. But I expect that the viscoelastic nature of the interlaced material will dampen vibs quicker compared to a solid. I'll try to find the damping info for the materials and run some dynamic solvers to see if they are damper. I created them the same thickness as for a flat plate like this the modes are all bending so the thickness needs to be the same to keep it apples for apples. This model has the 1mm interlayers as epoxy. I played with aluminium, plastic and other stuff as the interlayers. There's lots of variation so lots of learning in this area if you get into it. Peter

[hanermo]

The pdf posted is excellent - read it some years ago.

One takeaway - for milling steel.
Milling machines need 20-30 Nm/um deflection // rigidity iirc.
And better machines have 70 Nm/um and up.

Look at a small 2 HP Bridgeport.
If you push the edge of the table, 2 fingers, == 10-20 kgf = 150 Nm (middle), it will twist 0.04 mm, plus or minus something.
So 200 Nm twists 40 microns, at the edge, or 5 Nm/um at edge ==> around 5x less at spindle, == 25 Nm/um rigidity.

Likewise, a modern 40 HP HAAS VMC running, slap the spindle, the workpiece will have a visible divot (many microns, === 10-20 um maybe depending on the tool).
==> maybe around 40-50-70 Nm/um rigidity /my guesstimate.
A japanese Mori Seiki or Okuma would typically have 50% more mass, and about double the rigidity of a HAAS, and double the price, per work envelope.

The bigger machines have a big advantage in that the heavy tables and machine frames absorb vibrations.
So they are more rigid and also more forgiving/efficient.

Lathes since 80 years ago and modern machine tools (VMC etc) are all pretty similar.
The rated load is == 50x the breaking load.
Machine tools are only loaded at max 2% of their "strength" as in breaking load.
The 1200 kgf max push force == 40 HP power equates to about 2400 kg mass on the table // VF3 / similar.
Using 40-50 mm linear guides == 70-90 metric tons theoretical load.

[peteeng]

Hello Hanermo - So if we use 25Nm/um as a design target are you saying that if I place a torque wrench on my machine tool end and exert 25Nm it should deflect 25um? 25Nm is 2.55kgf/m so over say 10mm its 255kgf. ? ie if I push the tool with 250kgf it deflects 0.001mm roughly Peter

[hanermo]

If you push the spindle nose with 25 Nm, it should deflect 1um, for 25 Nm/um rigidity.
25 N force is 2.5 kgf, or 2.5 kg of push force.

Force at the tool end no, at the spindle.
The toolholder will flex a bit, so will the collet a bit more and the tool a lot.

A toolholder with collet, gage pin inserted, torqued properly, is ideal.
Push-pull toolholder, not the gage pin.
Or push-pull spindle if possible.

A dti clamped to table, indicate spindle nose, will show spindle nose deflection.
It is likely to be 10x-50x more than You ever expected on (best) hobby stuff.

There are youtube videos showing x3 mills, and iirc skyfire cnc machines, dti readings under load with hand forces.
About 0.1 mm deflection or more under 20 kgf load (200N).
Terrible, 5x worse than a bridgeport.

Comparison.
My 12x chicom lathe deflects about 0.015-0.02 mm, 10-20 um, under 40-50 kgf load by hand, 12 cm out from spindle nose.
12 cm out is a big deal.
Rigidity is length power 3.
As I pull the 10" chuck really hard.
It deflects about 10 um fairly easily, maybe 5 kgf, and then gets exponentially more stiff.

The bearings have low preload, imo, and this is why it deflects easily at first.
Probably 70 mm inner D bearings, == 2500 kgf rated load.
Chester craftsman lathe, new from chester uk circa 2005.
Heavy, light-industrial type lathe.

Lathe (and vmc spindle) bearings are a very delicate topic.
So, since it works extremely well, and I need the lathe, I have not had the courage + spare $$ on-hand, + potential downtime, to adjust the spindle bearing preload.
The primary locknut needs about 0.05 mm of angular movement.

Hello Hanermo - So if we use 25Nm/um as a design target are you saying that if I place a torque wrench on my machine tool end and exert 25Nm it should deflect 25um? 25Nm is 2.55kgf/m so over say 10mm its 255kgf. ? ie if I push the tool with 250kgf it deflects 0.001mm roughly Peter

[hanermo]

I used Nm incorrectly, when I meant N as in force.
10 N == 1 kgf, or 1 kg of push or pull.
The f indicates force.

My examples are meant to illustrate, qualify, add credence, and act as facts others may test and comment on.
Thus where I make errors, others can easily see and point them out.

Likewise, I try to add real-world values, so all here can see if these posts pass the smell test and are in line with generally accepted data / videos /etc.
Likewise, I try to add some industrial data and stuff from what I learned and saw, within limits.

Precise customer data and industrial data and pics must not be posted, unless it was posted/permitted by the customer/manufacturer to my personal knowledge.
So I don´t post pics, sorry.