[Stevecjx]

Hi,G-spot.We run a store for all kinds of machine components,linear bearings,mini screw jacks,couplings,handles,bushes,screw & nuts.....It will be our pleasure if there is something we can help you. https://www.msmstore.com/

[G-Spot]

if you send me the step file I can run it through the fusion optimiser, get rid of the rod in the cars and simplify a bit, add a spindle and your close...---- Peter

Thanks, I've attached it I added the rails and bearings for the Z axis now.

[G-Spot]

re weight and speed - you have to decide what the max cutting speed you need. Its big structures with small stepovers so you require very high speeds unless you want to take a year to get the job done. so research cutting speeds. I expect to cut very fast you will require a 40,000rpm spindle. Then work back from that using a suitable chip per tooth spec to figure out feed speeds. Then once you know target feed speeds you can figure out air times for a typical job. This will tell you what sort of accel you need and then weight is your enemy. If you accel at typical commercial mill accels say 0.2g then your 500kg gantry weighs 500*0.2*9.81=981kg to move... ie big motors; inertia loads can get very high... Peter

consider some commercial mills are at 1g thats scary big stuff...

I am not going to use this to do large 3D aluminum printing, I am going to use it make my aluminum printer.
First thing I would use it for is to produce my own linear motors and bearings so I can afford to print at 15m scales.

I hate the whine of those high rpm motors, was taught metalwork at school, everything had to be done by hand, drilling, cutting, no electricity allowed. Only when you had mastered cutting and drilling by hand were you allowed to move onto powered machines. Cutting or drilling at too high a speed would have got you clipped around the ear with your work-piece.

Commercial is all about meeting a specification at the the highest speed with the lowest cost, I don't want to do that. I would like a machine that can mill any material to very high accuracy with a beautiful finish.

And from looking at the FEA and research I have done so far, it would seem that you are not going to reach across more than 500mm of empty space and applying a force of 1000N without at least 150kg of steel behind you.

Need to span 1.5m then that is going to be 3 x 150kg and then another 500mm for your Z-axis will be another 150kg on top of that, so in my case a minimum of 600Kg.....lol.
Doesn't seem to be any chance of making a machine with the same size work area as I want with less than 400kg. So is there that much difference between shifting 400kg compared to 600kg? There isn't much difference in other machinery that moves those type of weights. You don't have a half ton truck, then a 600kg truck, and a 700kg truck. They just make a half ton, then a full ton etc.

And don't the best commercial milling machines of this size come in at between 1500Kg and 2000Kg ?

[G-Spot]

This shows you the power of triangles over squares.

Replaced the box shape with triangle of the exact same outer dimensions and the same wall thickness 30mm

The weight dropped from 239Kg to 169Kg but the deflection only increased 1.43 micrometers

[peteeng]

Hi G - To work through some of your statements:
1) Producing linear motors is a 20 year project - good luck Hiwin stock them just buy it
2) I have looked at and built linear bearings and again good luck on producing something that you can buy for <$50USD that is accurate and reliable. To produce a 15m machine you will need to use as many commercial parts as possible
3) High rpm spindles water cooled are nearly silent. I have to be careful near mine as I don't realise it's on sometimes. You say you want to produce a mill that can machine any material. Unfortunately that is a conundrum. Aluminium and steel require very different spindles. Plus the size of machine you are talking about requires very high feed speeds otherwise a large job will take forever to complete. These days "high speed" machining techniques are used and high speed is getting even faster material removal rates. Everything is getting faster... and sorry hand skills are not taught at school anymore. Schools have lasers, routers 3D printers. Digital manufacture is taught and hand skills are old school.
4) Its easy to shift 1000kg or 2000kg I can push my 2T truck across a flat carpark...What is the problem is accelerating and decelerating masses. This requires powerful systems, then accuracy follows on
5) You have moved the goal posts a couple of times in this thread. You need to document your intent and development path. Otherwise you will chase your tail quite a bit in the development of whatever you are trying to achieve
6) The "best" is subjective but a mill this size will be several tonnes way way past 2T and 10T
7) This size machine and what you want to do with it will probably need to be a 5 axis machine. A huge 3 axis machine will be limited in what it can machine apart from flat stuff. Any sort of height in the part will mean the tool will need to cant or tilt to reach the surface... So you need to look at what you want to do and clarify the machine further

https://www.haascnc.com/machines/ver...hdc-3-5ax.html heres a HAAS about your size and its $1M USD and weighs well over 10 tonne. Even a smaller $0.5M machine weighs over 10T. If they could make lighter machines they would (clarify that statement - machine builders are at the cusp of building light machines but until the user industry demands various things that won't happen. Its like electric cars, was possible for decades but took various players and circumstances to line up before they did it)

So onward and upward (imagine holding fist in air at angle and trying to launch like superman) the grail is out there. Peter

[peteeng]

Hi G - I looked at your model and you will need to simplify the cars. They have tiny screws in them, seals and engraving which the mesher will create lots of elements and connections with. This will take a long time to solve and not provide any useful information. Create simplified cars and rails before you go forward. Will save a huge heap of time. Peter

car 1 - stripped down Hiwin car. The rods are not needed as there are other contact surfaces in the model. But it has thin & thick areas. Simplify! plus the top of the car is in contact with the rail, this does not contact in practice make it a gap
car 2 - hiwin model has seals and tiny screws and oilers, strip them out!
car 3 - engraving get rid of it, mesh nightmare

make the cars and rails as simple as possible that captures the geometry and the main contact areas... Peter

edit - you also have cars back to back across the saddle. Agree that's ideal structurally but how are you going to assemble it in practice? Time to look at assembly logic as well

[peteeng]

Hi G - I meshed your model. It had a few issues but was good enough to run, then the solver failed so I expect the mesh was poor in some spots. Simplify the cars and I can rerun it. Peter

[peteeng]

Hi G - I looked at the model closer and a couple of the cars still had balls. That would be the failure. Anywho I replaced the cars and rails with some dummy geometry and ran the optimiser. The optimiser bands the elements into low stressed to high stressed vs % weight of the part. You can only run one body at a time. So the images are showing what the structure would be like at 50% weight. Sort of a truss structure. The issue is the gantry moves so this structure needs to move with it, which it can't. But gives me ideas to how it could look. The triangle is quite good. Maybe I run it in generative to see how it goes. Peter

[peteeng]

Hi - I ran it thru the generative but it attached the Z axis to the bed. I did say make it as stiff as possible and I suppose that's the correct answer. I'll have to add some preventative material between them. Peter

[peteeng]

Hi all - The generative design always wants to connect the Z axis to the base... So I tried the automated connection feature. Got this funky shape....Plus others Peter

edit added funkier 2

edit 2 side view - what this tells me is triangular is a good shape - as the side view is triangular. Now as the z axis moves across the gantry this network won't work. But the triangular section will just need to optimise the thicknesses and place a few holes maybe.

[peteeng]

To describe what's happening with Funkier 2 -

In general the AI has connected all the bearings (with all the other bearings) with ligaments that form triangles.
1) The saddle area is solid and the top two saddle bearings go to the back wall bearings as braces. They also reach out to the fwd wall bearings to form a triangle. This means the front plate will be solid as this bit moves.
2) The two fwd wall bearings are connected with a large circular tube, this is the main torsion and bending beam. Not sure how to analogue this in the tri beam. The bottom plate will have to do. It will be the front of the bottom plate
3) The rear wall bearings are connected with a smaller round beam. This will be the rear of the bottom plate
4) The bottom saddle bearings have braces back to the rear wall bearings. This is the bottom plate with some holes....

The ligaments do not go directly bearing to bearing. They are shortened along the main beams as shear is dominate at the ends. Its a 3D triangular truss in 3 planes. Once understood its quite simple. Join all the bearings with lines then simplify to make the ends less complex. Triangles rule.

Peter

[G-Spot]

OK I went and did a thousand simulations, literally, first spent half a day experimenting with my FEA settings until I could get it to pump out the deformation maps in under 15 seconds.

Managed to get the weight down from 750Kg to 500Kg, that includes the weight of the Z - axis post, which I am using a standard piece of 300x300x10mm tube section which is 900mm long, almost the length of the main beam.

Also managed to do all the bolting fixtures, which helped me get the weight down, I saw that I could use them like an I beam works, create thick mass far from the center, I am going to call it my Tri-Beam, killed two birds with one stone, made a super stiff bolting plan, helped me reduce the wall thickness by 20mm.

So the face plate minimum/average thickness is now 20mm and the bottom plate too, the back plate is only 15mm.

I don't really want to get any more detailed in my drawings until I get exact costs for supply of the steel and cutting/shaping services.

If the engineering is going to cost too much I will change my design.

[G-Spot]

Hi G - To work through some of your statements:

It's a DIY project for me where the fun is in the often rocky journey and not always the end result. So I get to do all the crazy stuff that would normally get you fired. Thanks to your honest critique I am moving the goal posts to avoid losing the game...lol

I am looking at making sure what ever I build now can be upgraded as I go, also spending months ordering insane amounts of expensive equipment in the hope that I can throw it all together and get a good result at the end, is too risky for me. So I am going to break it down into stages and after each stage is complete only then do I order the next lot of equipment needed.

First stage will be construction of the beam and supporting base, going to make the base out of a steel frame linking the rails directly with steel but buried in 500mm thick cement. Going to build it outside and use the surplus from a cement truck job I am doing. Going to...eventually....make the roof of the building part of the frame for unimaginable stiffness.

The thing is the commercial machine makers have to be able to transport what they make to the customer. I don't, and I know from experience with concrete that it is very difficult to break up when lying in the ground first you have to try and get the earth from around it because how well it absorbs blows. Also hit an excavator with a concrete vibrator and it will ring and vibrate like hell, put the vibrator down on top of a slab of buried concrete and it's a totally different story.

Old age and treachery will always beat youth and exuberance...lol

[G-Spot]

you also have cars back to back across the saddle. Agree that's ideal structurally but how are you going to assemble it in practice? Time to look at assembly logic as well

My FEA software allows me to set "Physics = off" for objects, I have multiple parts loaded and set them to invisible and physics off instead of deleting them, that way I can quickly do an optimisation using an old part to compare against the new one. I will remember to remove them before I export as a step file next time.

I just spent a lot of time tuning and validating my FEA data and meshing. The deformation maps had started taking like 5 to 10 minutes each until I did something about it.

I am glad I didn't spend hours working out a bolt plan for the saddle as I have decided to use the 35mm bearings all round instead, did the finances last night, was a much of a muchness, and the difference between the 25mm is like chalk and cheese, 3.6 tons max static load compared to 6.9 for the 35s. And their geometry alone makes them so much stiffer never mind the quality of fit. And I can see there is no chance of reducing my machine weight below 400Kg.

[G-Spot]

Got this funky shape....Plus others Peter
e.

It's amazing how nature has a few million years head start from us in the optimisation game. These optimisations always look like living creatures and not machines.

[G-Spot]

Hi all - The generative design always wants to connect the Z axis to the base...

Check the green creature out, I see this shape time and time again in all the optimisations, the L- shape, like a piece of angle iron. Your face and bottom plate need to be thickest but the back plates are always left off.

[G-Spot]

I believe I can Fly!!!

My FEA software is better than your FEA software...at least mine look like something you could build and not something you would throw to your dog...lol

[peteeng]

Hi G - Topological "optimisation" algorithms are generally based on removing low stressed material. This approach often does not develop high stiffness solutions. I can realise all of the weird solutions if I wanted to by making 3D prints (or machining timber patterns), then casting the weirdness is quite simple. But I use them for visualisation and alternative thinking.... the AI may do something left field and good. It has given me some ideas to persue so all good. Peter

[G-Spot]

Also look at how the FEA loves those Fillets I added so I could bolt together well. Has not touched them at all, obviously the front one is how it is because the simulation does not know that it moves.

On all the simulations since I started this, it likes an L shape with a huge fillet between the 2 faces.

[peteeng]

Hi G - Thats a local shear transfer tactic. Shear transfer and distribution is something I have had to learn about in this sort of machine design. Machine parts like to be thick. Peter