[peteeng]

Evening all you night Ewoks - This evening I detailed the Frame model. Took 4 goes to get it to target stiffness. I scaled down the Tee model a bit then had to creep up again in some dims. It's stats with an 50mm thick Al table:

X 23 Y 19 and Z 36N/um. If I run a 50mm steel table it becomes X 26 Y 22 and Z 37. As usual a steel table makes a big delta. I'll put in the 3rd table rail to see how that goes as well.

My main concern is that the table rails travel across two parts. All parts are less then 54kg so a little more tuning and holes should get the parts to <50kg but at the mo I'm happy. So another attempt at shrinking it a bit and then to the motion parts. Will try to use 60mm servos but 80mm maybe to go...

The base is 120mm thick and the Tee base was 130mm which would get the Y over the line with the alloy table. Its still 400x400x400 travel. Wednesday I can rearrange the workshop as the floor resin will be hard then I can do some casting tests at last... Peter

addition - starlog 3567853-001.234
Completing the discussion on motor sizing and ballscrew whirling and buckling.. The top speed of the servo is 3000rpm (can go faster but V happy there) Calculation via Hiwin formula says the 16mm shaft whirls at 3764rpm <3000rpm so all good there. The max force of a 2.39Nm motor and 5mm pitch screw is 275kgf. The buckling of a 800mm 16mm screw is 880kgf so all good there as well. So Dia16mm screws are OK Jose.. Peter

[pippin88]

Peter, how are you expecting the end user to align things?

Aligning rails across two mating parts may be a major problem.

Even aligning X and Y perpendicular may be difficult for most users

You may find you don't have much of a market if cost becomes mill price plus thousands in master squares, straightedges and indicators.

[peteeng]

Hi Pippin - Yes it is a quandary... I don't like the idea also so have to think it through. I can do the alignment and use epoxy to create precision dowel fits prior to supply. Not ideal... I intended to have aluminium faying surfaces machined to fits/registers so would solve that...

The other plan is to bolt the columns on the sides vs the top of the base but that puts bolts in shear and friction again not ideal for a machine...

More ZZZ's needed on it. I think I have to scale Milli down a bit more maybe 350x350 table but the bearings drive the top end part size so may have to go to 15mm bearings.... Peter

I also have to review the stiffness and the way the model is restrained it maybe a bit too high a target for the intended market. Its massively stiff compared to other small mills visually. I'll look at the skyfire/mikini model again for some insight. Peter

[peteeng]

Afternoon All & Sundry - I have been looking through various documents looking at how researchers and analysts hold their models and machines. Some hold the base some hold the table. if you hold the table you get a smaller stiffness then if you hold the base. The photo shows a machine test on a VMC, it got over 150N/um. If your comparing designs I suppose this does not matter as its apples for apples. From an absolute method we want to load it close to practice so thats table to tool. I reloaded the Mikini and put my std load post on it. With the post it scored X10 Y14 & Z49. Holding the base it scored X18 Y26 and Z48... So I suppose I stay with what I have to be consistent.. Peter

[peteeng]

Evening all - Its late and my brain needs a diversion. Saw this mob on TV they are a choir nearby us. Bit of fun

https://www.facebook.com/DustyeskyChoir/

Now I have, through this machine development played with sandwiches for the base but they have not worked out. Generally because the foam centre has been plastic. PVC and 400kg/m3 PU have been looked at. Tonight I revisited concrete foam and seems to work. I wrote a little spreadsheet to design the same stiffness base yet reduce weight by half. I'm assuming I can make 1300kg/m3 foam and it will be half the stiffness of std CSA. ie half the density half the stiffness. I've seen articles where concrete foam has been made down to <500kg/m3 so the densities not an issue I believe. I have not foamed concrete before so that's a trick to learn... I have a concrete guru friend with a foamer so will get some samples from him...

Didn't like the two piece base, sounded a bit risky. My intent now is to get a top 25mm alum plate and bottom plate machined with all the holes and registrations and epoxy these top and bottom of a moulded concrete foam core...

So the figures are X 29 Y26 and Z 37 N/um these are very encouraging. Still would like to shrink Milli a bit so will look at how to do that next. Another day closer to a solution.

Another way to halve the weight of the core is to make voids in it... this sidesteps the foam development. I'll run this tomorrow

Keep Making - Peter

[peteeng]

Evening all - - - Been thinking about an exo frame for the mill. So I mocked up a frame using 100x100x6 steel tube. It scored X9 Y9 and Z23 N/um. So I then used 150x150x6 and it scored X10 Y10 Z 26. Befuddled I looked at it closely and the rails where lifting the middle of the SHS sections so it was flexing locally. So don't have air under rails. This arrangement looks like it maybe a pain to load. Did the final levelling on YaG today... and started sorting the workshop. Peter

[catahoula]

Why not just take all the extra material from the exoskeleton and make the base and columns twice as large?

[peteeng]

Hi Cat - Been thru several config exercises and so far no joy in getting the base to 100kg at 20N/um . The exoskeloton can be broken into separate bits but its very heavy and has the issues of local deflection, heat treatment, finish machining. I think the foam concrete sandwich is the go. Will get back to this soonest. Next step I think is to make some CSA and prove I can get 70-80GPa out of the mix... Peter

The other option (there are always options) is to connect the columns to the side of the base, Then the base will be <100kg. The columns can be L shaped if needed. Will do that. If the wedge shape is the nominal shape and nominal area of the base then a steel base has to be 25.5mm thick to be 100kg an aluminium base would be 74mm thick and both of these are clearly not stiff enough. the base has to be at least 130mm thick to be stiff enough in E70 material. If the base is 800x300x130 it weighs 84kg if its aluminium so that seems to be the next try... The column weight goes up but then that can be managed....

[peteeng]

Evening All - I built the block base and it looks good. The base is 81kg the columns are 63kg so manageable... Its stats are X21 Y20 and Z 35 N/um so were there.... I've have to do a bolt analysis as the bolts are in a different plane to conventional.... The weights are based on aluminium density. I'll tidy this a bit and smooth it out as well.. Peter

[peteeng]

Morning All - I'm boiling down Milli to her essence. I changed the table size to 400x300 and adjusted a few dims. Now its X21 Y22 and Z 37 N/um... The base is 69kg and the columns are 61kg. I included a 123 block (25.450.876.2 block?) on the table for scale. The footprint is 800x675mm which is good for a benchtop. It will have steel table this makes quite a big delta without having 3 rails and 9 cars. I'll look at removing the 3rd car from the rails as well. A small ray of happiness is starting to shine... Peter

[peteeng]

Morning Tea all - Up to Block-4. I have whittled some weight out, added flanges to the columns and it scores well. X 22 Y 22 and Z 36N/um so now I refine the structure then move to adding the motion bits.. Which is her best side?/// Peter

addition - I reran Milli-Block-4 with CSA E40 material. The deflection as expected was less but when you multiply the stiffness ratios you don't get the expected value. That's because rigidity is a function of geometry and material stiffness. So the X for instance in E40 gets 16N/um and in E70 it gets 24N/um. So 16/40x70= 28N/um if its purely a material ratio. But it only got 24N/um in the E70 model. So the structure could be a little bit more efficient I think... Peter

[peteeng]

Evening all and sundry types. - Been smoothing out Millis lines and adjusting part weights. won't do too much aesthetic work because I know it will change once I start adding motors and screws etc... Peter

[JuanGg]

Looks great!

[ardenum]

Whoa look at those smooth lines, mental.

[klaas123]

Very nice design Peter.

[peteeng]

Morning all and thanks for the good thoughts - 4 months to get here but I suppose I lost a month at Xmas... But I think the topology of Milli is decided and fairly well formed. I did a little bit more on the aesthetics bringing a few features into line and consistency. She weighs 255kg in this form happy with that. The base (69kg) and columns (59kg) can lose a little midrift (like me) once we get through the next bit. So need to cast some materials and figure that out. Been sorting the workshop so now have some space for that. Last night I did some structural loop stuff to see where the stiffness is. Found a perplexing property of Milli. I "fix" the base of the bearings from the carriage back and see which loop is most compliant. Being a sort of cantilever I'd expect as we move away from the tool the deflection would get bigger and bigger until we get to the table. So here's the figures for X Axis. These are um at the tool:
carriage bearings 25
gantry bearings 43
column base 21
table cars 31
load rod (std restraint) 42

The column base support gets stiffer! can't figure that out. Because this is torsion it may have a preferred movement direction. Will have to look at the vectors to see which dirn is dominant... I'm off to get some water lines for the 80W laser so I can switch it on... Peter

Millis stats are X 23 Y 25 and Z 38 with the steel table. X is just under 18.5 with aluminium table... this is also with 2 table rails not 3. I do need some room under there for a ballscrew. The table is 45mm thick so is a serious chunk of metal

[peteeng]

Hi All - I want one!! Peter

[peteeng]

Hi All you cnc gurus out there - starting to plan the casting tests. So I drew up a table of materials possible. Anything to the left of aluminium is good. The 60%alox/ep scores the best stiffness, while sand epoxy is the worse. Now I have to figure costs to see which one is <<$$$ Each I feel is doable. The next trick is how to test for the modulus without spending $$$ at a lab... and I'll have to get some alox which is a 2.5hr drive north but the border is now open so that makes it easier....Peter

[peteeng]

Afternoon all and sundry - Looked through my files and organised the $$$ of the materials. The epoxy in small qtys is really expensive. So the CSA/ALOX seems to be the best value... So I'll try to make an equivalent of the European E80 material... so I need to get some ALOX!! cheers Peter

[peteeng]

Evening All - I added 100% CSA to the table. If the part can be done at 34-40GPa its the cheapest solution. I've come across some SiC and will cost that monday... SiC is 400Gpa so stiffer then alumina comes down to $$$. Peter

https://www.ceramicsrefractories.sai...trol-equipment