[yezard]

I’ve been reading around these forums for a while now, and would like to venture into the world of designing and building my own 3 axis CNC machine.

I’m planning on using the machine mainly for aluminium milling but would like it to have the capability of machining steel. Now I know I’m pushing the boat out here for a DIY machine and that a machine capable of milling steel has to be very well designed and stiff, so I was hoping you may be able to help me.

IV sketched up a design on soildworks, and as it stands

it has a X travel of 1000mm

Y travel of 460mm

Z travel of 110mm

The machine will mainly be comprised of aluminium and steel with some HDPE parts, its design and dimensions are built around the already machined ball screws available at ZAPP automation.

The X axis is comprised of a unsupported linear rail with angle irons on which V grove bearing will run across. I didn’t go for the traditional THK rails or supported rails due to their price, but feels as if the combination of both the unsupported linear rail and the angle iron will be adequate.

The Y axis runs on a set of 20mm unsupported rail, I know there has been hype on here that unsupported round rails are next to useless so I am weary that design will actually work. But the supported round rails seem so much more expensive. So my question would this work, if not would increasing the diameter of the rails to say 30mm help. Or is unsupported a complete No No.

The Z axis runs on a set of unsupported round rails of which are bolted to 5mm aluminium plate

The gantry is manly comprised of 5mm thick aluminium with an aluminium box section (2" by 1" and 1/8" thick) joining both sides (should i go which steel?). I’ve gone for 5mm to keep the costs down and think it should be adequate

Also does anyone know of any good and cheap suppliers of cnc stock such as suported rails and V groove bearings, im based in the UK.

Any help is much appreciated and thanks for looking!!

[handlewanker]

Just my 2 cents worth, so don't take a panic attack if'n I get highly critical.....it won't mill steel!
I doubt it will "mill" alluminium even, it might mill plastics and MDF stuff.

One error you have is the side supports to the gantry....thay have no means of resisting side loads, and milling is all about 360 degree side loading.....the cutter will deflect to the side of least resistance.

You would be better off with a vertical mill design, C frame type, fixed motor on top etc, either a quill in a fixed head or a motorised head as the Z axis.

Bench mounted not knee type as it's too difficult to get enough metal in the body without going to extensive castings.

The design you show is probably more suitable for a router on large wood panels. (scale not apparent)

It also depends on how big a piece of work you would "like" to machine.

Define the length breadth and thickness of your maximum work piece in steel and design the table XYZ travel round it, allowing for cutter/chuck stick out.

Look at the Mill/drill design with the round column....this is so crap, when you try to mill anything significent the whole head vibrates and deflects from the thrust of the cutter.

All the best.
Ian.

[jsheerin]

I agree with Ian. Even for a wood router, your design would be marginal. You would want to stiffen the sides of the gantry and replace all the unsupported rods with fully supported bearings. You would also need to add some more bracing in the bed.

If you want a metal cutting machine, I'd recommend a fixed gantry carrying the Z axis with an X-Y table mounted on the bed as opposed to a C frame. You'll also need to make everything much, much stiffer to get decent performance.

[yezard]

Thanks for the responses, the reason I didnt go with a fixed gantry was because most of the work im planing on doing is with aluminium plate, the fixed gantary design would probally restrict me.

I can see the problems with this design but would really like to work around them. I will probally scrap the unsuported linear rails on the y axis for suported round rails and secure them to a steel box section (rather than the aluminum).
The bed is actually going to have beams of steel running across to stiffen it, i just havent added them into the drawing.

As for the X axis linear rails i would like to change them but the price for the unsuported rails are sooo much higher. Is there any other cheaper alternatives?

I've come to relise this machine will not machine steel but I really want a machine capable of machining aluminium. Do you think with the improvements it would be capable of doing so?

[wildwestpat]

Take a quick look at the size of vertical mill needed to machine a flat plate 500mm x 1000mm and look at its weight. Now weight does not directly equate to mass of metal but with an all up weight of over 300Kg you will get the feel for what is required. Yes aluminium is weight for weight stiffer than cast iron.

Work out the tolerances of the parts to be made both in absolute accuracy and in terms of repeatability for the biggest parts you want to make. These tolerances are not the same. Repeatability is mainly due to lost motion and the absolute accuracy additionally relates to the ability to place features and have them measured independently to be in the correct position.

In the case of repeatability this tolerance band needs to accommodate the sum of all the bending, sagging, crabbing, tooling and back-lash as well as resisting any tendency for the cutting action to induce vibration. This applies to all axis. The absolute accuracy is determined by the drive mechanism. The use of linear glass scales or laser measuring systems as feedback elements to control the position of the axis are expensive. (Glass scale servo motor and amplifier for each axis is going to pitch up in excess of $1000 and more.) A simple open loop control system based on stepper motors will be limited in speed of traverse and will have the absolute accuracy obtainable fixed by the lead screw or belt drive but a lot cheaper.

Those conceptual single skin side plates need to have depth and a lot of cross bracing to withstand the forces that would be imposed. Cutting MDF is a different ball game to metal and you need to do the arithmetic.

Hope this helps and does not put you off having a go. Regards - Pat

PS - Yes I think you could make this work. I have cut 30 mm aluminium using hand power tools primarily intended for wood and given enough lubricant this works. (Shaping with circular saw and routing.) However high speed cutting will probably induce too much flexing to be viable. Vibration and the resulting chatter / tool dig in situations are going to stress your over head gantry. Think in terms of damping as well as cross bracing. If you can laminate the beams as a metal plywood metal sandwich you should get better stiffness but you will need loads of cross bracing and gussets no matter whaht the beams are made of.

[jsheerin]

It will cut aluminum as is - it just won't do it very well or very quickly. I cut aluminum with my wood router which is very likely stiffer than your design as shown above, and it does okay for hobby use but leaves a bad surface finish on side milling and goes slowly.

The fixed gantry would require that your long axis have longer linear rails, and then depending on you did the Y axis, everything else could be the same.

Ebay is one option for cheaper linear rails. I've bought a lot of THK stuff there. Another could be to build your own bearings. Depending on the speeds you want to run you could build sliding bearings, rolling bearings using skate bearings (which done right can be stiff and have low friction), or hydrostatic bearings (although those would be vastly overkill for your frame and the support equipment would probably be too expensive). You could potentially look into V-groove bearings as well. Unsupported rails are about the last option I would look at for a machine like this.

[wildwestpat]

As you are in the UK take a look at Zapp Automation they some times have surplus trucks and track.

Regards - Pat

[yezard]

Thanks for the information its really helpful. I agree with you jsheerin that the unsupported X axis is going to lead to problems but is why I added the sliding bearing unit that runs across the angle irons to prevent unwanted movment. In the drawing I havent added the V-groove bearing in the sliding unit partly because I couldnt find any in the solidworks tool library.

As for the single skin side plates, how about adding a layer of HDPE and then adding a additional 3mm alumninium plate to sandwhich it altogether. would this work, would HPDE do anything to stiffness the design and would it absorb some of the vibrations?

[wildwestpat]

You would need to find a glue that would bond to HDPE and the metal. Other feature of HDPE compared to a nice bit of solid birch ply wood is the fact that it will give a bit more under stress. With a composite the aim is to get the two outer-skins held the same distance apart this would mean a load of spacers of the same thickness as the HDPE and pop rivets or bolts to hold the layers together.

Take a look at one of those cheapo interior flush doors. Two layers thin outer layers held apart by a paper honeycomb. This form of construction is used in the aircraft industry using alloy sheets and honeycomb with the glue being applied in sheet form. The five layer sandwich is then put under a vacuum blanket and baked until the glue sheet (Redux ??) melts and sets to bond the parts together. Balsa wood is also sometimes substituted for the honeycomb. These sheets make very good light weight beams with low internal resonances. Pity we do not have the DIY aircraft hobby fraternity like they do in America as they might have off cuts.

Regards - Pat

PS like the idea of dampening the vibrations with the three layer sheet. Some of the old Hi-Fi gramaphone pickups used foam plastic to dampen out the arm resonances. - Those were the days prior to MP3! Might be worth and experiment with aerosol foam as this sticks well and has good structural strength.

[yezard]

Thanks guys, the information has been great.
I am in the process of drawing up some of the changes in solidworks, so will post the updated drawings within the next few days.
I'm hoping to start work on the machine this week so should be uploading some photos soon.
In the mean time does anyone know of any good sources of V-groove bearings?

[yezard]

I have updated assembly to incorporate the supported rails.
Im planning on buying some of the HDPE from direct plastics and the aluminium from clickmetal has anyone had any experiences in using them?

[digits]

I don't mean to sound harsh, but I really think you need to do some sums to look at the forces this machine is going to be under. 5mm thick aluminium sheet is pretty weak - sandwiching it with HDPE sounds a bit like using cheese-spread to stiffen a pair of crackers to me...

You aren't trying to build a nice strong but flexible aircraft wing here - you are trying to build something that won't deflect measurably even if you stand on it...

Also, if you primarily plan to cut aluminium on this, I would seriously consider designing your machine around how you are going to get a lot of coolant in and a lot of chips out. If you don't need much Z-axis, you might do well with an overhead gantry design, with a base that has no electronics or moving parts in it - then flood cooling will be trivial...

I've not tried your metal suppliers, but I would suggest you shop around - try and work out how much they are charging per kilo. If you try an industrial place like metalfast, you might have to email for a quote but you might get a better choice of materials - mill finished toolplate is great for machine construction as it is very,very flat.

Hope that helps...

[yezard]

Thanks for your input digits. Could you then point me in the right direction for the calculations required for a machine like this, and possibly advice me on the thickness of aluminium i would need for the gantary sides to be stiff enough.

I have currently have a furnace, if i go about casting a set of gantary side, let say 15-20mm thick would that be adequate?

[digits]

I am no expert, but I think I've got the hang of the basic principles here:

You have 2 sets of forces to worry about here - cutting forces and acceleration forces.

Cutting forces are obviously due to trying to remove metal, and acceleration ones are due to trying to move the various parts of your machine.

You can calculate the acceleration forces based on the torque of your steppers/servos and the pitch of your screws - there's lots of info on here about that.

Cutting forces are trickier - but really depend on the size of cutters you plan to use, the material you plan to cut and the spindle power and speed (rpm) you have avaliable to you.

I stumbled across this today, and it looks handy:

aluMATTER*|*Aluminium*|*Milling*|*Power

Remember you can use less force and more speed to remove the same amount of material in a given time (power) - you can dig a trench with a spade or a teaspoon for example!

Once you have the forces, you can decide how much deflection you can live with in your machine to achieve the accuracies you want, and then you can work out how much stiffness you need.

If you have a furnace and can make castings, I'd imagine you can have lots of firey fun with shapes too complex for most DIY'ers. Stiffness isn't just about huge thick slabs of material - think about how little metal is in a coke-can and how strong it is. Or, more usefully for a machine builder, how much stiffer 1kg of metal is as a box-section than as a thin rod...

I hope that helps a bit!

Cheers.

[handlewanker]

Exactly precisely as Digits said.....accceleration and force.

The force may be with you, but the bogey in the woodpile is cantilevering.

They are the easiest to calculate....empiracaly of course.

The furthest you get away from a fixed object the less ability you have to resist deflection.......try holding a cup of tea at arms length and then walk around.....you won't have much tea left.

In the milling mode, be it for flat bed router type machines or verticle table top mills, the forces are all against the column, for C type table top verticle mills, and the side frames for router type gantry mills.

Move the table or move the gantry.....the cantilever forces are the ones furthest from the fixed point.

The RF50 mill drill, so popular with home workshops, gained the reputation during the 70's as the "Taiwanese Terror", mainly due to shoddy manufacture, but mostly because all the reactive forces were directed against the round support column, and the cutter, being at least 350mm away from the centre of the column, exerted tremendous side force when only .25mm depth of cut on a 19mm diam cutter were applied, resulting in the relatively massive head casting vibrating like a lollipop on a stick.

The rule for boring bar stick out in the lathe is 4 times the boring bar diam, any further and you risk deflection in the cut, but you can take this to 6 times the diam for a finish cut and get away with it.

So deflection and cantilevering are two empirical forces that the eye can appreciate immediately from a scale drawing.

Designing the machine means you make it to just cover the maximum size of the job it must handle......a machine that can handle 300mm X 300mm X 100mm high is a mighty piece of metal and no compromise can be tolerated.

Machining a piece of plate alluminium 300mm X 300mm X 25mm thick and using a depth of cut at .25mm gets you there eventually using a cutter of 19mm diam, but if'n you go to a 100mm diam slabbing milling cutter with 6 Carbide tips, positive rake and depth of cut .5mm, or a fly cutter sweeping 100mmm diam and DOC of .5mm, the machine may be able to swing the cutter but the deflection will probably just lift the cutter off of the job.

CNC mode means you can take many shallow cuts and get there, but in the end the cutter surface contact area will determine what the job turns out like if it's dangling far from a support point.
Ian.

[jsheerin]

Some general guidelines for stiffness, from my research for my own mill, are that for cutting steel you want around 50,000 to >200k lbf/in, for aluminum you'd want 20k to >50k, and for wood around 2.8k to 12k. The stiffness necessary for decent performance scales with the modulus of elasticity of the material. As has been said, you can get away with less but it limits what you can do with the machine. My wood router has a stiffness of between 1000 and 2000 lbf/in in various directions, and as I mentioned it does not do a very good job at cutting aluminum, but it will cut it. I've also cut unhardened A2 tool steel and 304 stainless steel, once again very slowly and without a very good surface finish.

Now the real trick is calculating what the actual stiffness of the machine is before you build it. I use finite element analysis to do this, but even that is not 100% accurate. The trickiest bits are the screws (or other drive systems) and the bearings. Both the screws and the bearings have a stiffness. So even if the various parts of your frame are infinitely stiff, if you use floppy (relatively speaking) bearings and screws, your entire machine will not be very stiff. The stiffness of bearings and ball screws is based on the hertzian contact between the balls and the screws or rails. Hertzian contact is when a ball or roller contacts a flat surface. In theory there is only point or line contact, thus zero area, and thus infinite stress in the material (stress=force/area, so if area is zero, stress=infinity). Obviously stress can't be infinity in the real world - the ball and rail deform. That deformation under load is what defines the stiffness. This is a long way of saying that it's tough (although definitely not impossible) to calculate so you need these values from the manufacturer. Then you need to combine these values with the stiffness of the various frame members. While it is possible to do this stuff by hand, it's not something I would ever try. That's not to say that you can't do basic calculations to look at the stiffness of individual beams, but as far as calculating the stiffness of the entire machine, it's tough.

Where this leaves you, imo (unless you can do fea), is that you can look at what other people and companies have done for various machines that have various known levels of performance and decide what you want to do for yours given your goals. This is what everyone here is trying to help you do.

[handlewanker]

Hi, I doubt many DIY mill or lathe builders go down the path of calculating ANY forces that might be present, because those forces vary with the type and newness of the cutter.

Most of the forces are hypothetical, and in the real world the order of the day is to make it as strong as you can with whatever material you can get.....nobody "designs" a machine to meet the requirements that exist on paper.....the forces at play are too obscure, and cutter geometry can make any calculations a tiresome process.

For instance, how much torque would you require to drive a 20mm diam end mill (4 tooth) at 5mm depth of cut and using whatever the recommended speed and speed for maximum roughing material removal in mild steel.

To most of us....quite a bit....so don't have too much overhang or tool stick out when roughing, otherwise if'n it resists the cut when the machine is built, just reduce the feed rate...problem solved.....we only intend to make one machine, not a prototype for an envisaged production run for the market, and certainly not from previous experience of lathe and mill building, so everthing that makes it strong, and to hell with the cost, will apply, you just need a good eye for form.

Having made the lathe or mill and found it wanting, the next step is to "beef it up" a bit by adding material in various places or re-doing certain components with the exact science of hindsight to guide us.

It is said that anyone can build a bridge that won't ever fall down, but it takes a very experienced bridge building engineer to make a bridge that will just carry the designed load with minimum material input and not fall down under the load.

Cars were made by the rule of thumb process, in the beginning, but got better as more experience revealed the need to improve the strength of all the bits that fell off and reduce the bits that stayed but just added weight.

In the end, when all the calcs are done, a safety factor of X amount is applied that doubles or trebles the structure to make sure on the day it doesn't disintergrate....LOL.....so most of us can fall somewhere in that picture when it comes to "designing" a lathe or mill etc.

Applied mechanics is a golden rule....you can pull a 100kg load with a 10mm diam screw thread a metre or more long, but the same thread would buckle when the same load was pushed a distance of approx 10 times it's diam.

I can see all the professional engineers reaching for their pens with quivering hands to refute the logic of DIY engineering using empirical formulae, but the eye is usually the most accurate design tool in the box.....calculations and formulae are derived from experience and empirical design.
Ian.

[wildwestpat]

There is a lot more to rigidity than the resistance to bending. Resonance of the various parts also has a considerable effect. Selective cross bracing and tailoring of the sections mass help. FEA analysis certainly helps as it will pinpoint the high stress areas.

If you can borrow a high intensity strobe it is instructive to look at a friends gantry machine under strobe lighting. You might loose a friend when you see how much the various parts pant and how the resonances suddenly start up. If you are still friends you could experiment by adding cross bracing and fine tuning by adding mass with plasticine. Often adding a few cross braces the stiffness will increase but as importantly the resonances will move higher in frequency and reduce in amplitude.

If you can not find a strobe light and you don't have a friend gently dusting the support structure with talcum powder with the power off and then try a savage short cut. The powder will fall off the points where the resonance peaks. These points can also be identified by running a finger round the main structure and across all joints and bearings BUT DO NOT take risks with personal safety. Having identified the points of high flexing add cross braces or reduce bearing clearances if possible.

Regards - Pat

PS calculation and computer simulation is good and so is practical experimental engineering. Combine the two is a recipe for perfection but at a cost!

[digits]

I do hope we haven't killied off this build by sucking all the fun out of it!

My reason for wading in initially was because it seemed obvious to me that the machine wouldn't be sturdy as designed.

The OP also mentioned that certain components were being chosen to cut cost - and IMHO not building it twice because the first one wasn't what you expected is the best way to save money...

All I was suggesting was that a little maths might save a lot of heartache. Simple things like how much will the weight of the spindle deflect the Y-axis etc.

Another great way to ease the design process is to learn from the mistakes and ofcourse sucesses of others. This doesn't look like a radical new design to me so there must be a lot of similar builds on here that would provide some insight as to what really works and what needed a lot of modding to get right.

This is a fun hobby and sometimes just building something to see what works and what doesn't is the best approach but you do need to be able to afford to bin (or really recylce and reuse) your less sucessful prototypes.

Cheers - I'm off to knock-up some motor mounts of my own

[cyclestart]

This is a fun hobby and sometimes just building something to see what works and what doesn't is the best approach but you do need to be able to afford to bin (or really recylce and reuse) your less sucessful prototypes.

I like that response !

Some cnczone members have deadlines and demanding customers so this is all deadly serious business. Others are dedicated hobbyists who want to squeeze every bit of performance possible out of their designs. Some of us simply enjoy playing with mechanical concepts, experiment, and things are learned from the occasional spectacular failure. Whatever the approach, for the hobbyist it's for love of the game.

My router won't win any performance or beauty contests but it does cut stuff (possibly aluminum, haven't tried) and it's a great deal of fun. It evolves as time and funds permit. Some of the structure is currently doubled up 3/4" plywood and MDF. Eventually it will be all metal construction and by that time I will have a full idea of what the final design will look like. Changing your mind in metal is seldom as cheap. The dimensions were a firm decision so the ballscrews, rails, etc are keepers.

My philosophical 2 cents