[G-Spot]

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

Thank you so much for looking at my drawings and taking your time out to analyze them for me. Really appreciate that mate.

[G-Spot]

I am having a hard time locating the deck clearance you need above ball screws.
Some of the tables have obvious errors in them.

I worked it out indirectly to 54.5mm if you are using BK20 supports and 72.5mm if you are using BK25s. Can anyone confirm this?

[peteeng]

Hi G - I suggest you download one of the main manufactures catalogues and use their numbers. They have detailed dimensional data on their products. Hiwin skf INA etc they all build to international std dims on most of their lines. Peter

[peteeng]

Hi G - Here's a link to a BK dimension table. Plus FBT have all the other component dims as well. Peter

https://www.fbtmotion.com/html/12.html

[G-Spot]

Hi G - Here's a link to a BK dimension table. Plus FBT have all the other component dims as well. Peter

https://www.fbtmotion.com/html/12.html

Thanks for that but I see this one has obvious errors in too. The counterbore depth for the BK25 is 17.5mm and not 175mm as tabled.

Also how are ball screws chosen? I presume after how much load they need move it would be acceleration ? They must want to buckle if you stop a heavy load quickly and whip if you accelerate it too fast.

They would be the first thing I want to upgrade or do away with and use linear motors instead, if I did not need to use them then I could just use flat plates for my machine, I am sure the milling that I am going to need done to use them will cost at least $1000. If it does I think I will just bolt riser plates for my rails onto my flat plates until I can make a better solution.

Do you think 20mm will be sufficient to move my Y and Z axis which will only be moving 150Kg > 200Kg?

My plan was to use 25mms for Y and Z and then to use dual 32s for the X axis which will be moving the 500kgs. But now I see that they need 80mm clearance for the saddle to pass over I really want to use only 20mm.

They are like Vinyl records of the music industry....essentially following a scratch in lump of material and trying to suggest you can have high fidelity.

[peteeng]

Hi G - X&Y Ballscrews are sized in a 3 step process. Because they do not lift mass (X&Y) they only need to move the inertia and the friction. Neglecting friction you need to pick a suitable acceleration value you need (0.1 to 0.2g is a common number for non-commercial mills. Some commercial mills are over 1g ). Plus first you also need to check that at max motor torque (hence screw force) they do not buckle at max length. Hiwin publish a design manual for the these things... So once you decide what diameter is needed to stop it buckling you then check it does not have a critical vibration (whirl or whip) once there your good to go. Then review the screw inertia to check the motor has enough grunt to spin it at the design accel.

The Z motor has to do the same but is generally short enough to sidestep buckling and whirling. Plus the motor has to lift the spindle and z axis mass. So the motor has to lift the mass, spin the screw inertia and move the linear inertia.
I've made a spreadsheet I'll find it... Peter

The force is proportional to pitch nothing to do with diameter... Heres a typical hand calculation for a mill.

edit - found the spreadsheet

[G-Spot]

Upgraded to 35mm bearings and saw a huge increase in stiffness.

Like the same increase in stiffness that increasing the plate thickness on the beam by 10mm.

Made a new bolting plan for them on the saddle I see they must have been designed to be placed like this as their bolt holes align to be all off each other.

I dont think it will make much difference as they have 40mm between them and the bolts are only M10 and you only need 5 thread turns to hold the load, no matter how much more thread you have it makes no difference, 20mm should be more than enough thread depth. Had a pressure vessel blow its lid off at 800 Bar, a gas pressure vessel too, luckily for me while doing a hydrostatic test and it was because they engineer who designed the lid thought the more thread engaged the stronger it will hold, it doesn't.

[G-Spot]

Here are may latest figures, the same as the first design using 30mm thick plate all round.

Going to strip the design right back to basic geometry and then recheck it. Also did some research into corrugations in steel and discovered that a 70 degree angle of corrugation is stiffest and not the 45 degree I have used in my latest design.

So going to try and verify that in FEA and also explore my tribeam design more. What is the optimal size for the fillets ?

[G-Spot]

Hi G - X&Y Ballscrews are sized in a 3 step process.

Thanks for the excellent explanation and calculations explaining it all.

[G-Spot]

Was going to make my base in the corner of 2 sheer walls at my house. To be super sneaky.
But then I thought that I could use the power of concrete and steel working together without any tricks.

Considering this base has only 1230Kg of steel in it and is supporting a 20mm thick base plate which is 413Kg of those 1230Kg.....And the work area is 2000mm x 1500mm, I think I proved a point.

X axis stiffness 61 N/um
Y axis stiffness 62 N/um
Z axis stiffness 108 N/um

Did the research into increasing wall thickness inwards. It is totally not worth it, if you have steel to spare you need to put it on the outside.
Filling steel pipes with concrete gives excellent increases in stiffness considering the price of concrete compared to steel.

If the choice is between increasing wall thickness inwards or adding cement. Then its adding cement every time.

[G-Spot]

FEA results

[peteeng]

Hi G - Be careful with "adding cement" the FE model has the cement bonded to the steel with 100% efficiency. In reality most "cements" shrink and over time will dis-bond from the steel. Then the structure will only be the steel structure not the combined structure. There are grouts designed to expand so they mechanically lock into the void they are poured into. I suggest you investigate those. Plus looking at your part geometry and carriages currently it's impossible to assemble. What modulus are you using for the "cement"? Peter

[G-Spot]

Only 16 N/um without the cement.

Too boost that up backup to 61N/um with steel, you would need at least another 1000Kg.

[G-Spot]

Hi G - Be careful with "adding cement" the FE model has the cement bonded to the steel with 100% efficiency. In reality most "cements" shrink and over time will dis-bond from the steel. Then the structure will only be the steel structure not the combined structure. There are grouts designed to expand so they mechanically lock into the void they are poured into. I suggest you investigate those. Plus looking at your part geometry and carriages currently it's impossible to assemble. What modulus are you using for the "cement"? Peter

I've never tried to stick things to cement...lol.... the steel is wrapped around 2 massive longitudinal cement beams, it's all purely a mechanical setup, check the cross section. Everyone reading this will be dead before that steel decides to leave the party.

[peteeng]

Actual no, the bond strength between steel and concrete is very low especially if its not prepared correctly. Which in the case of hollows. So when it disbonds you will have a 16N/um machine vs a 61N/um machine... What modulus are you assuming the concrete is? Peter

To do a test get a large hollow section say 150x150x6 say 100mm long. Fill it with the proposed concrete and come back in 60 days and you will be able to knock the concrete out if its normal concrete due to its shrinkage... if your going to use the steel as a mould and skin then forget about all those ribs. Not needed.

[G-Spot]

What modulus are you using for the "cement"? Peter

I am going to use the cement I have been working with the last 10 years, casting with it, I mix it up with just the right water in it, it's thick and will not move anywhere unless you physically move it, its 40 MPa mortar cement . All that shrinkage story is because builders can't work with it when it has the correct amount of water in it, so they add more water or plasticisers or other junk to make it easy to work with. If there were any scientific issues with cement not bonding to steel well, we would live in a very different world, steel parting ways with cement is either because the cement was poor quality or the steel rotted out from water getting through the cement, see poor quality.
And I must have stuck hundreds of metal fence posts into concrete, never ever in my life saw a metal post that was loose in good concrete and all the years I have broken up concrete with rebar and steel inside, never saw anything other than the cement hanging on with it's last dying breath.

It's like who am I going to believe, my lying eyes or this research?....lol

Plus I did a check with the FEA where I made the beams removable, in case I wanted to move the machine, so they were still FEA "bonded" with the metal but no mechanical join, the FEA showed a lack of stiffness between the cement and metal and the results were pitiful.

[peteeng]

Hi G- so your using 40MPa as the modulus in your modelling? 40Mpa is its strength not the material stiffness. Metal posts with concrete around it means the concrete has shrunk onto the post... Peter

[G-Spot]

What modulus are you assuming the concrete is? Peter

To do a test get a large hollow section say 150x150x6 say 100mm long. Fill it with the proposed concrete and come back in 60 days and you will be able to knock the concrete out if its normal concrete due to its shrinkage... if your going to use the steel as a mould and skin then forget about all those ribs. Not needed.

I just used the lowest spec concrete that I could find in the FEA the 20MPa stuff.

I am not going to be doing any pretests because I have been casting this concrete for over 10 years, works great, never had surface or any other type of cracking. I regularly use it for metal posts, I am a wizard at working it with a vibrator, I know all the tricks for getting the air out of it, and how to lay it so you don't trap any in it in the first place.

Also had to break it up when there have been mistakes with formwork, 4 days with a hammer chisel just to remove 100mm from a few square meters.

I could not think of a more risky move than to use a cement you have never used before for an important job. And if you never casted with it before you are never going to know how to vibrate it properly so you don't send the aggregate to the bottom and the water to the top. I can well imagine anyone who has not worked with it before making a mess of filling up a machine base with lots of nooks and crannies.

But I thank you for your concern because if I knew someone who had never worked with concrete before was going to do a machine base, I would advise them to find a friend who has experience with casting concrete. Vibrating the steel from the outside and all the usual mistakes, whatever you vibrate the water below is going to come to that area so you end up with water rich cement near the surface.....and the most shrinkage close to the surface.....but it looks real nice, really smooth, you will think you did a good job and never know.

[peteeng]

Hi G - Again you have not answered my question. If you pick a modulus of 70GPa then it will be exceptional stiff. Most concretes are around 30GPa.... Using concrete is up to you and your experience but Portland cement absolutely does shrink and if your modelling and making decisions about concrete you do need to understand what modulus you are using for the concrete... I too have spent a lot of time in my youth laying and breaking up concrete as my father was a builder and I was cheap labour. By concrete comes in 100's of flavours most are not suitable for a machine. I tend to like peppermint, but Italian cactus ice cream is pretty good. Peter

[G-Spot]

Lol.....I saw there was an option for high performance concrete and like a hundred different others, I knew you would ask about the concrete so I chose the lowest grade they had.

OK Peter I will follow your advice on which concrete to use, but it better not end up being so stupidly expensive as to make just increasing the steel the best option. There is no way I could afford to build this machine at its current dimensions unless I use less than 1000kg of steel for the base, so its cement and steel base or I have to build smaller. I am counting on a cubic meter of cement costing less than $100.

Only just discovered how well the FEA models concrete and steel, watched a lot of simulations, if it's not mechanical then it shows really poor stiffness. The best part about the FEA I am using is the replays of the force ramping up, it uses massively exaggerated movement to show you clearly the manner in which it is deforming. This helps more than any final fancy graphic. When the concrete is only bonded at the surface with the steel then it behaves exactly as you would expect concrete beams just surrounded by steel would behave.