[peteeng]

Morning All - I have scaled down Mill No3 to Milli No4 and lost some stiffness. I went from 200mm gantry to a 150x175mm and this is too small even though I put bulkheads inside. Surprisingly Z stiffness improved. I did move the cars so they have bolt access from the side that allows the biggest bolt. So my thoughts are:
1) move to 25mm rails and cars
2) 200x200 gantry minimum
3) break the current mono parts down into plates
4) think about a low speed spindle. Currently the motor would have to live way out front, maybe Ok.... Make a bespoke spindle mount integrated wit the Z axis
5) Break the base into base and walls
6) current envelope is 500x400x300mm Z . I think 400x400x400mm maybe a good size its still big for a bench machine. Bigger then a Tormach 770

Milli-No4 X 20 Y 10 and Z a cracking 139N/um weighs 394kg got to work on Y and weight... Peter

[StrawberryBoi]

Can you space the cars further apart on the y-axis?

[peteeng]

Hi Strawb - By spacing I think you mean the transverse spacing on the saddle? IE the along the gantry dirn? All bearings can be spaced further apart. But that then chews into the footprint unless its the vertical separation u are meaning? . My plan is to make the saddle thicker as that has shown up as warping a bit. Saddle 1 was 50mm thick saddle 2 is 15mm thick and it warps a little. This is because the rail/car is mounted on its side and the major loads are trying to peel the rail vs push straight in and out if face mounted... Ideally they should be mounted in the same plane to remove eccentric moments but then its difficult having so many bolts in the same place.

By going from 500mm wide to 400mm wide the gantry will be much stiffer. I like the current saddle and gantry rails relationship as it allows the cars to have easy assess to its bolts both ways. 300x600mm or 500x250mm, maybe a better bed size, I'd better decide fast. Peter

[peteeng]

Hi All - I looked up the stiffness of the bearings and 20 vs 25mm not much delta. I thought it was larger then charted by memory. But using 25mm will make it higher so will check the ballscrew vs rail heights. Maybe advantage in that. Peter

[peteeng]

Hi All - I have been reviewing the saddle design and have modelled HR-5-0 and HR5-1. I realised that the gantry loads are mainly horizontal and thats the direction I made it smaller. So I made it 250mm long and 150mm high for No5. The stiffness rose a bit. I looked at Mori designs and they use type 2 saddles. This is a good design in that the cars can have excellent access. But the Z axis has to "reach" down further from the bottom bearing. But the gantry can be mounted lower as the spindle then has room to move up to the bottom of the gantry... If I can get the Y axis up from 15 to over 20 I'm a happy camper. Peter

[peteeng]

Evening All - I decide to compare the Top Saddle to the U saddle. The U saddle gets 20/15/132 N/um and using the same parts and a solid steel top saddle it gets 11/10/93 so the top saddle is out. Even with the walls reduced in height due to the top saddle allowing that to occur its no-where near as stiff. So the U works well, need to tune it up a bit... to get to 20N/um all round...

[peteeng]

Evening All - Been wondering how to decide on which bit makes the biggest difference. Usually I restrain bearings at each structural loop and figure out the loop compliance then this tells me the least compliant bit so I address that. But its a rigmorale to do. So today I I made the main bits "rigid" and looked at what made the biggest difference. So the FE system has a "rigid" material spec which means the material is infinitely stiff. So the part geometry becomes insignificant. Making the base rigid changed the Y stiffness from 15 to 17 so base does not matter. It did change the Z dirn from 133 to 330! But we are trying to get the Y axis over 20... Changing the gantry to rigid got the Y to 25 very good. Changing the saddle took it from 15 to 19 so don't change the saddle and changing the column to rigid changed it from 15 to 24N/um so it seems the gantry and Z column are the candidates to look at... The gantry is 250mm deep x 150mm high x12mm thick. So I can make it higher back to 200mm and 16mm thick. The column is 16mm thick and 150x150mm so take it up to 175x175x16mm . The y axis is in combined bending and torsion so has always been a tough cookie to crack. The saddle has always been a difficult to design part. Trying to get cars close but access to bolts etc is tough. Maybe try to sort that a little better then its time for a rebuild... Peter

[StrawberryBoi]

Looking forward to see if your size adjustments bring you over the 20 line!

[peteeng]

Evening all and you there with greasy maker hands pay attention - Seems I'm there. Milli-HR No7 achieves 20/24/160 N/um. Had to make the gantry a bit thicker and from prior work added a diagonal web to triangulate the section. Not sure how to realise that detail yet. Next step is to redesign for manufacture including drive system and bits and bobs. I will also design the base and walls in Granite and get a quote for these to be machined in China. That will be twice as stiff as concrete. I also re-costed materials today. Plate aluminium is running at $26,000/m3 AUD I thought I'd make an acetal sandwich base but acetal is $30,800/m3 AUD. Makes the CSA look remarkable at $2700/m3 AUD & $1.29AUD /kg. I suppose thats why they use concrete vs aluminium for buildings...Straight epoxy is $18000AUD /m3... Can't get 25mm, 32mm or 50mm thick aluminium plate at the moment. So the plan is laser cut stainless steel, fabricate, stress relieve, electropolish, finish machine. I found a great data sheet on SS today... If plain steel then its all same except zinc plated at end... I want to downsize more so need to look at the low speed spindle solutions again. ISO20 is the go there... So into refinement cycle. Peter

[ardenum]

Evening all and you there with greasy maker hands pay attention - Seems I'm there. Milli-HR No7 achieves 20/24/160 N/um. Had to make the gantry a bit thicker and from prior work added a diagonal web to triangulate the section. Not sure how to realise that detail yet. Next step is to redesign for manufacture including drive system and bits and bobs. I will also design the base and walls in Granite and get a quote for these to be machined in China. That will be twice as stiff as concrete. I also re-costed materials today. Plate aluminium is running at $26,000/m3 AUD I thought I'd make an acetal sandwich base but acetal is $30,800/m3 AUD. Makes the CSA look remarkable at $2700/m3 AUD & $1.29AUD /kg. I suppose thats why they use concrete vs aluminium for buildings...Straight epoxy is $18000AUD /m3... Can't get 25mm, 32mm or 50mm thick aluminium plate at the moment. So the plan is laser cut stainless steel, fabricate, stress relieve, electropolish, finish machine. I found a great data sheet on SS today... If plain steel then its all same except zinc plated at end... I want to downsize more so need to look at the low speed spindle solutions again. ISO20 is the go there... So into refinement cycle. Peter

instead of the diagonal bit, use perpendicular plates that you can bolt from the outside, there was a paper that described this exact method but i don't remember its name. Essentially think if you cast it in iron like the big boys do, it would be the exact same structure. Space the plates so that you get square structure inside.

[peteeng]

Hi Ard - The issue with using bulkheads is that if the Z axis is in between the bulkheads it performs differently to when its at the BH's. BH's are a global solution (restrains the square tube from lozenging) and easy to do but its not a local and global solution. I can do the diagonal brace and will address it better next round. Version 6 had bulkheads... When I did the optimisation runs it always put a diagonal behind the bearing. The structure needs a shear path at the hard points to minimise shear local deflection. The brace gives it a local and global solution... There are some machines out there that use a triangular gantry for this reason. I have played with triangular gantries, some were good some not, depends on the bearing arrangement. Regards Peter

[peteeng]

Afternoon All - Been busy doing various jobs but got to do some refinement this afternoon. The main news is that I'm back to a "conventional" saddle bearing arrangement. History is that looking at the component deflections I noticed the lower rail is being distorted as its being "peeled" from the bottom of the gantry. If I made it rigid the stiffness jumped from 20 to 35N/um. No9-01 stiffness is 20/21/119N/um.

So I decide to compare to a conventional arrangement. The column is 94mm from the gantry with the U saddle and 120mm with the flat saddle. The flat saddle scored No9-03 19.5/25/74N/um which I think is a better compromise. Plus now the front face of the gantry can be machined in one OP. So That's probably why this is a common arrangement. Took a while to figure that one!! I did make a model where the saddle bearings where on top of each other and it did better but that is too difficult to assemble and maintain... Also I made the walls and base separate and in granite although still looking at casting. Will get a quote for granite asap. So now to add the motion and manufacturing layer to the design.

main features:
footprint 1000x700 envelope X550 Y350 Z 350mm 1.5kW high speed metal spindle , looking for iso20 spindle for slow speed (steel) Construction stainless steel and aluminium. Looking at laminating some parts for damping, saddle is ideal for this... Still want to downsize a bit.. current model mass 500kg... anything else you want to know please ask.. I also increased rails to 25mm vs 20mm Peter

[peteeng]

Hi All - If I downsize I run the risk of not needing the gantry config. This has happened earlier on. Since this project was a "moonshot" project into a composite mill design and that has not worked out how I thought it would and since the configuration has a very good stiffness baseline I think I'll extend the moonshot into a 5 axis mill.... I've been wanting to investigate 5 axis for a while and this seems to be a good foundation for one. The attached chart shows that Milli is in good stiffness territory so lets go the big 5A.. 5 axis will also help in my mould work so its a worthwhile trajectory for the machine Milli5A.... Peter

[pippin88]

Afternoon All - Been busy doing various jobs but got to do some refinement this afternoon. The main news is that I'm back to a "conventional" saddle bearing arrangement. History is that looking at the component deflections I noticed the lower rail is being distorted as its being "peeled" from the bottom of the gantry. If I made it rigid the stiffness jumped from 20 to 35N/um. No9-01 stiffness is 20/21/119N/um.

So I decide to compare to a conventional arrangement. The column is 94mm from the gantry with the U saddle and 120mm with the flat saddle. The flat saddle scored No9-03 19.5/25/74N/um which I think is a better compromise. Plus now the front face of the gantry can be machined in one OP. So That's probably why this is a common arrangement. Took a while to figure that one!! I did make a model where the saddle bearings where on top of each other and it did better but that is too difficult to assemble and maintain... Also I made the walls and base separate and in granite although still looking at casting. Will get a quote for granite asap. So now to add the motion and manufacturing layer to the design.

main features:
footprint 1000x700 envelope X550 Y350 Z 350mm 1.5kW high speed metal spindle , looking for iso20 spindle for slow speed (steel) Construction stainless steel and aluminium. Looking at laminating some parts for damping, saddle is ideal for this... Still want to downsize a bit.. current model mass 500kg... anything else you want to know please ask.. I also increased rails to 25mm vs 20mm Peter

Instead of Z bearings above and below gantry bearings, have the gantry bearings spread wider apart, either side of Z (instead of hidden behind)

Because gantry goes over the column walls there is room to do this

[peteeng]

Hi Pippin - not clear on what your saying. Sketch please? If the Z bearings are inside the gantry bearings then they are difficult to do up. have to use hex head screws and ring spanners vs keys? I'm happy with arrangement has to be easy to work on... Peter

Hi again Pippin - I reread your post slowly and I got it that time around. I'll have a look at that...it does have bolt access issues but maybe its good... a bit of Art Deco for your evening.

[ardenum]

Hi Ard - The issue with using bulkheads is that if the Z axis is in between the bulkheads it performs differently to when its at the BH's. BH's are a global solution (restrains the square tube from lozenging) and easy to do but its not a local and global solution. I can do the diagonal brace and will address it better next round. Version 6 had bulkheads... When I did the optimisation runs it always put a diagonal behind the bearing. The structure needs a shear path at the hard points to minimise shear local deflection. The brace gives it a local and global solution... There are some machines out there that use a triangular gantry for this reason. I have played with triangular gantries, some were good some not, depends on the bearing arrangement. Regards Peter

all cast iron machine structures have a grid inside them, might not be that big of an issue.

Pippin meant to space the saddle carriages horizontally. you wouldn't have issue of bolting carriages on top of each other and you get a slightly more stiff connection.

[peteeng]

Hi Ard - Always more ways to skin the cat... Agreed with the horizontal cars but then you can't get access to the inside car bolts easily. Generally happy for now with its config and geometry, do some refining and move onto the A & B axis that's my area of interest now. Peter

[StrawberryBoi]

What kind of speed and power (or torque) are you looking to actually be running the spindle with?

Oh, just as a side note, at work we have a Jia-ken spindle that goes 24krpm and is 2.2kW on a not terribly stiff router. We machine steel and stainless steel on it, 1/4" DOC, using a dough hook (single flute) 1/8" end mill. It's not accurate to much better than 0.003" on small parts and 0.010" on large sheet parts when not clamped near the cutting site, but it cuts the steel cool running at 2000W comfortably. Thicke pieces of stock it's not even noisy, the sheet metal itself vibrates due to large area thin sections.

Would it be faster with 6kW? Sure. The end mills might cost less $/part too due to less wear. But just to point out that low speed high torque is an optimization mostly for wear and DOC (due to end mill size) power is a volume removal factor.

What are your targets for end mill size in steel?

[ardenum]

Hi Ard - Always more ways to skin the cat... Agreed with the horizontal cars but then you can't get access to the inside car bolts easily. Generally happy for now with its config and geometry, do some refining and move onto the A & B axis that's my area of interest now. Peter

let me save you some trouble, for the C axis(the axis you rotate the spindle head about/perpendicular to the table) I came up with a slewing ring/harmonic reducer combo. The stiffest slewing ring at small sizes is cross roller type, INA makes these quite affordable(XU product group). For my application I'm planning using xu120179 INA slewing ring and CSG-32 harmonic reducer. they fit quite nice. Total package costs about $600 give or take.

[peteeng]

Hi Strawb - What are "my" targets for end mill size in steel? Answer - don't know - I started this thread in Oct 2020. The machine definition was benchtop mill, 500x500x500mm with a target stiffness of 10N/um made from cold cast composites. The 20N/um comes from using a 50% efficiency factor in the "bonded" model. ie the model will be >20N/um but in reality it will be >10N/um taking all the bolts, friction, clearances etc into account. In reality I expect it to be 70% plus efficient but may as well aim high.

By mill I meant that it could mill steel and aluminium easily. What was in my head was something better then a Taig, Syil, skyfire etc. Hobby to semi-commercial machines... Don't want to upset anyone out there about their machine, broad statements...1/2" 12mm tool biggest I imagined.

Nearly a year down the wobbly linear rail this is what transpired. 1) The 500x500x500 envelope is too big to create a very stiff compact hobby machine Mill 2) My aim of creating an E=70GPa casting material failed 30-40GPa is possible. As the machine gets smaller the material has to get stiffer to compensate so we have ended up with steel 3) Milli is not like a Taig!

The spindle strategy is an interesting space. There is no reason why a high speed spindle can't work on steel, comes down to machine stiffness, spindle power and chip load and tool strength an finish requirements. As you say there are "routers" out there doing it.

Every time I tried to shrink Milli it grew back. So I have decided to stay with the current size and add the A-B as I have a project that will need an A-B in the near future. This will start getting some solutions out there.

What is the feedrate, chipload or step over for the 1/8" endmill? and actual rpm? you are using. I'll compare to some toolmakers data and try to find out what's happening. Many years ago when I ran machines I preferred to run fast and light vs slow and heavy. Seemed to get the same MRR and was easier on the machines. I have not done a lot of machining just enough to be dangerous. I worked part time in many machine shops doing all sorts of things while I studied... turned a lot of handles. CNC machines were rare...Peter