Milli a new composite mill kit

[Buoyen]

Have you seen this:

Looks similar to the design you are heading for. Couldn't find any cutting with the machine, may not be finished.

[peteeng]

Thks Bouyen - interesting machine. So many bits to get aligned and stay aligned. He's filled RHS with concrete or EG then machined each piece as a jigsaw puzzle. Trying to avoid many parts I am... The vid is 3 months old so I expect he hasn't finished it yet. He's well set up with gear... Peter

[peteeng]

Morning All - Structural trivia time. In many designs fillets are used for transitions and for good looks. But they are not as stiff as the same size chamfer. In composites inside fillets create issues with laminates by running out of geometry for the laminate thickness.. I've found that chamfers are slightly stiffer and allow the laminate to stack better into corners.... Here's an image showing the chamfer to be slightly stiffer then a round. I suppose its straight which means its axially stiffer then a round in bending... Peter

[StrawberryBoi]

Fillets aren't selected over chamfers for stiffness, but for lower stress rise. The chamfer will definitely be stiffer for the same reason the isogrid is stiffer than a hexgrid. A fillet or chamfer is rarely so small even when "capped" as to be considered a core in a sandwich construction.

Likewise a circular undercut with a 15-45deg reverse angle exit into a return fillet is even better for stress rise than a fillet, but even less stiff. You will see this type of feature in high torque density shafts where torsion and not bending is the critical load. It reduces the bending stiffness but decreases stress concentrations so the torque capacity is actually higher.

[StrawberryBoi]

Here's an example profile. You typically see this with large diameter changes in short axial length, which is very common in geared or splined systems or space constrained motors.

EDIT: It's also very handy for sneaking in O-rings with no lost space, when you need them, while gaining your stress reducing feature.

[peteeng]

Thks Strawb - Yes undercut fillets are used a lot on shafts. I model them often for acceptable fatigue performance on wheel/axle shafts for mining equipment. An elliptical transition does better then a rad if the machinist can do it (or use two rads for a pseudo ellipse) . In CNC machines however stress is so low fatigue isn't on the radar but stiffness always is... Peter

Heres an axle I'm working through at the moment - polish is really important to stop the wheels falling off... sometimes they do...

[ardenum]

Hey Peter,

I'm trying to decide between Al 5083 and Al 6082 for under rail material. Have you worked with any of these? Is there anything that pops up in 'real world' that isn't necessarily on the material's spec sheet? Attached is how I'm thinking of using it.

[Zorbit]

Carroll Smiths's book, "Engineer to Win", has some useful information on improving fatigue life of parts by removing material. The undercut already written about can be used on bolt heads.

[peteeng]

Hi Ard- Both are good for structural purposes. 5083 is a rolled plate product. Typically here it would be 5083-H32 which means its quite hard. 6082 is an extruded product and is available here in T5 or T6. The issue for your application is will the part stay flat when you machine it? The 5083 being cold rolled will have internal stress that may spring the part when machined. The 6082 will have less internal stress as its been heat treated after extrusion if in T6. if its 6082-T5 this can be achieved by quenching and stretching at the press so could have some internal stress.

If it where me I'd go with 6061-T6 plate as this is flat and has been solution heat treated and aged after rolling so should have little internal stress. The cost of 5083, 6082 and 6061 is same here. If you are having it machined at a company then ask the machinist which product moves less when machined and use it. If you are going to epoxy the plate to the concrete then machine, these issues are less as the concrete will keep the plate flat. But if you are going to make the concrete block glue the finished machined part to it then the part needs to be true. Then there is the issue of the saddle bottom. If you make a finished machine part you will need true 90deg surfaces to use as a jig to glue the entire thing together,. Or you glue blanks to the block then finish machine. My plan with Milli was to cast the CSA parts - coat with epoxy and bond aluminium parts to these then finish machine... Peter

By the way concrete and aluminium do not like each other so don't "cast-in" aluminium parts unless they are anodised. If concrete and al are touching I'd expect the Al to corrode over time.... that's why I would epoxy it in after casting... steel and concrete are a happy couple... unless in a wet or seaside environment...

edit - I had a quick look and 6082 is available as plate overseas (but can't get it in Oz I think) . So maybe you can get 6082 plate where you are. if so get T6 and that's a good solution....

[peteeng]

Hey Zorbit - Glad your still looking in. Today I powered up the motors and spindle on YaG all good, no smoke let out. Had to flip the dirn of Y axis motors then all good. So next step is installing the couplings and getting it to move. Happy days, then I can start on making moulds for Frankenrouter or Milli... Peter

edit -turns out the bag marked 1kg of 12mm CF is actually 490g which means I can't make a very big part. Thinking about how to make something I can test for stiffness at the moment...

[Zorbit]

Your build threads are unique, I enjoy the way they wander through the options, exploring the + and - of each design, and throwing in a few engineering truths along the way. Sometimes they go full circle - I'm glad I'm not the only one to do that.

A retired friend is a chartered engineer, he says "if you can't put numbers on it you don't really understand it". Your posts are helping to put numbers on DIY CNC design for me, I'll be checking in regularly even if I rarely comment.

[peteeng]

Thks Zorbit - All my lifes in circles someone sang once... here's a wheel attached to an axle thats bigger then the last image stub axle that I worked on... They can be bigger then this as well but they all look the same on my computer screen....... Peter

[ardenum]

Hi Ard- Both are good for structural purposes. 5083 is a rolled plate product. Typically here it would be 5083-H32 which means its quite hard. 6082 is an extruded product and is available here in T5 or T6. The issue for your application is will the part stay flat when you machine it? The 5083 being cold rolled will have internal stress that may spring the part when machined. The 6082 will have less internal stress as its been heat treated after extrusion if in T6. if its 6082-T5 this can be achieved by quenching and stretching at the press so could have some internal stress.

If it where me I'd go with 6061-T6 plate as this is flat and has been solution heat treated and aged after rolling so should have little internal stress. The cost of 5083, 6082 and 6061 is same here. If you are having it machined at a company then ask the machinist which product moves less when machined and use it. If you are going to epoxy the plate to the concrete then machine, these issues are less as the concrete will keep the plate flat. But if you are going to make the concrete block glue the finished machined part to it then the part needs to be true. Then there is the issue of the saddle bottom. If you make a finished machine part you will need true 90deg surfaces to use as a jig to glue the entire thing together,. Or you glue blanks to the block then finish machine. My plan with Milli was to cast the CSA parts - coat with epoxy and bond aluminium parts to these then finish machine... Peter

By the way concrete and aluminium do not like each other so don't "cast-in" aluminium parts unless they are anodised. If concrete and al are touching I'd expect the Al to corrode over time.... that's why I would epoxy it in after casting... steel and concrete are a happy couple... unless in a wet or seaside environment...

edit - I had a quick look and 6082 is available as plate overseas (but can't get it in Oz I think) . So maybe you can get 6082 plate where you are. if so get T6 and that's a good solution....

Thanks Peter, that got me thinking a lot. Didn't think about gluing the plates to concrete honestly. That would save me a lot of $ were I to go this way. I was already considering building the mold from precision 3um granite slabs instead of plywood, my forms are simple after all. That would give me good surface and I could sell the granite parts afterwards as opposed to a plywood mold. I assume the epoxy layer would also help with different thermal expansion rates between AL and UHPC. I'm really liking this idea.

Is there a formula I can use to compute if the surface area of the UHPC-epoxy-AL bonding would be strong enough? If you've done this before any type of epoxy you'd recommend for this process?

Many thanks again!

[peteeng]

Hi Ard - If you have granite why not use it? Granite is twice as stiff as concrete. Most epoxys have a shear strength of 20-40MPa. You can model the bondline as 0.5mm thick object and run it in FE if your FE can deal with a thin layer. You need a thickened epoxy that won't run out of the joint. Have a look for what's available and then I can comment.

Adhesives - ATL Composites In Oz I'd use one of these. Follow the suppliers prep notes to the letter. If humidity is up use heat gun on surfaces prior to epoxy application to remove water vapour layer... Peter

[peteeng]

Hi Ard - I have used Techniglue to bond lashing points to concrete. In tests the concrete always breaks not the bondline. Epoxy suppliers in USA? will have similar product Peter

[peteeng]

Evening All - I think I'll convert Milli to one of these.

https://robotics.kawasaki.com/en1/pr...?language_id=4

1.5T capacity at 4m reach !! can drill and router and stir weld.... A desktop 7 axis router Hmmmm Peter

[joeavaerage]

Hi,
by crikey....I reckon I could program that to fry up bacon and eggs....I want one too!

Craig

[peteeng]

Hi All - Its been on my list for some time to design a robot like machine. Perhaps Milli-IRB should be it. The issue is writing the Jacobian for the co-ordinate kinematics. It may exist I'll have a look at the FreeeCAD robot package. Or if there's a bright electronics person out there wanting to collaborate that would be great. I want to use ballscrews to drive it vs axis motors. Will be much stiffer. That's how parts of the Kawasaki robot is controlled. Mechanicals no problem - kinematics and control new territory. Its been 30 years since I've done that sort of maths....Peter

[StrawberryBoi]

[peteeng]

Hi Strawb - I've watched some of his vids. Very good, but he does not discuss the control side of things. I may have to send him a note. Peter

edit - I downloaded the MG10HL cad model very nice.