[peteeng]

Hi Ard - Yes I'm reviewing the thread at the moment, there have been prior stiff configs, but many of these where based on E70 casting material which won't happen. The tool length and nose geometry is important. One difficulty with Milli is that its not for a specific purpose so its missing a rudder. A general machine tries to cover many things. One day I think the lifting gantry is great the next day I see how it may not be great. So I have to build one & use it to really figure it out. Milli is a project in conflict; the imagined E70 material has not worked out, the path to the "best" configuration has been unclear and the parts geometry has remained fuzzy. Its been a bumpy ride... If I had to give Milli a purpose it would be to finish machine gantries which means the tombstone type moving column (y axis in version you mention) would be too big as it would need to be 1.8m long. So the axis would have to change direction to a smaller moving column design. It also was intended as a benchtop machine but keeps growing to a largish machine! Perhaps I have to cut it back to 300x300x300mm and re-start there. To bring up something not mentioned before, the original spec of 500x500x500mm was supposed to be achieved fairly quickly then I was going to add a 2 axis head to Milli to make it a 5 axis machine. That's well off as well! So Milli may remain as an investigative Moon Shot project. Perhaps I build Frankenrouter as a lifting gantry test machine...

https://whatis.techtarget.com/definition/moonshot

I'll reread the thread and pick up the pieces. Peter

[ardenum]

Hi Ard - Yes I'm reviewing the thread at the moment, there have been prior stiff configs, but many of these where based on E70 casting material which won't happen. The tool length and nose geometry is important. One difficulty with Milli is that its not for a specific purpose so its missing a rudder. A general machine tries to cover many things. One day I think the lifting gantry is great the next day I see how it may not be great. So I have to build one & use it to really figure it out. Milli is a project in conflict; the imagined E70 material has not worked out, the path to the "best" configuration has been unclear and the parts geometry has remained fuzzy. Its been a bumpy ride... If I had to give Milli a purpose it would be to finish machine gantries which means the tombstone type moving column (y axis in version you mention) would be too big as it would need to be 1.8m long. So the axis would have to change direction to a smaller moving column design. It also was intended as a benchtop machine but keeps growing to a largish machine! Perhaps I have to cut it back to 300x300x300mm and re-start there. To bring up something not mentioned before, the original spec of 500x500x500mm was supposed to be achieved fairly quickly then I was going to add a 2 axis head to Milli to make it a 5 axis machine. That's well off as well! So Milli may remain as an investigative Moon Shot project. Perhaps I build Frankenrouter as a lifting gantry test machine...

https://whatis.techtarget.com/definition/moonshot

I'll reread the thread and pick up the pieces. Peter

I know what you mean, if you're building a normal sized machine, you can always add more weight to make sure it works, but with small machines optimizing everything might be just above our pay grade, mine definitely...

There are 2 routes for me, buying a small garage and building a normal sized machine or something smaller I can place in my flat. So many compromises to make...

2 axis head? aren't those prohibitively expensive? I remember one guy mentioned a 2 axis head from hsd being about +$80k

[peteeng]

Hi Ard -The smaller the machine the less geometry you have therefore the less stiffness you have. The 20N/um target is very high for a small machine. Maybe its not possible unless I use beryllium or high modulus CF. I may have to visit steel, it maybe the answer... When I used steel for the tables on Milli it made a big delta so that says something. One aim was to billet machine parts on a big 5 axis machine, so that machine does not care whether its silly putty or steel or titanium... so if steel I would have parts zinc plated so that solves the rust issue... I dislike rust and excess weight....

I don't think the head will be expensive. I have found the software to drive it. Again until I design it I don't know about cost. I do know that the stiffness devil will be very present and it maybe that I have to design the smallest 5 axis head I can then design the machine around that. Comes down to what accuracy is required for the machine. If you shot at 0.1mm then easy peasy if 0.01mm hmmmm if 0.001mm then your in $80k territory...

5 axis heads are big. Perhaps the 5 is another moonshot thread.... The Milli thread has been useful in finding out what does not work vs what works... Perhaps thats the way to look at it and draw a line under the thread...1100+ entries is a big read...Peter

[StrawberryBoi]

Zinc plating doesn't hold up well to abuse over time compared to Nickel. Depending on your plater, Nickel can often be a cheaper (and more durable) option because many platers have been moving away from Zinc and they have to setup a bath just for your parts. Worth asking.

[peteeng]

Hi Strawb - Yes nickel is good. But I have a zinc plater within an hours drive and its cheaper then nickel and I only want it to not rust. Unfortunately general platers don't plate nickel here properly. They do not copper flash then Ni plate. To do a proper Ni job I have to send it 2500kms away to a plater in Melbourne. I have had rails and things Ni plated for medical part purposes that worked well. If you just nickel plate steel it will fall off over time... One of my first casual jobs was loading and unloading plating tanks I stopped that in respect for my lungs... Peter

[noweshop]

Maybe Calvind Dude will suit you. Calvind Dude focuses on the research and development and manufacturing of electronic equipment and industrial manufacturing automation equipment.

[StrawberryBoi]

Hi Strawb - Yes nickel is good. But I have a zinc plater within an hours drive and its cheaper then nickel and I only want it to not rust. Unfortunately general platers don't plate nickel here properly. They do not copper flash then Ni plate. To do a proper Ni job I have to send it 2500kms away to a plater in Melbourne. I have had rails and things Ni plated for medical part purposes that worked well. If you just nickel plate steel it will fall off over time... One of my first casual jobs was loading and unloading plating tanks I stopped that in respect for my lungs... Peter

For sure, though I've had numerous 416 and 410 ferritic, and 101x series steels plated with low phos and high phos nickel directly, per AMS 24xxh whichever the specific spec number is. It dramatically outlasted zinc in testing, no copper layer in that one.

Different than the older nickel bath chemistries though, this is electroless nickel iirc. I also had the ptfe impregnated version made on 410 ferritic and 4240 through hardened for epoxy molds and tooling. Some A2 steel as well.

The MIL spec didn't hold up, but it was absorbed in the AMS spec which has a slightly different chemistry.

[StrawberryBoi]

AMS 2404h I think it was. It's been 2.5years so memory may not serve.

[peteeng]

Hi Strawb - Yes I've had electroless done as well. Lets forget the finish for a while and get the parts sorted. So with steel in hand I reduced the G-Penny 100mm spindle to a Dia85mm spindle and reduced all the part sizes to suit. While one of my FE projects was trying to solve I did this on the side. The bed is 100mm thick. Milli now weighs a neat 999kg, not quite a benchtop machine. I can lift the saddle and arm but not the rest! Anywho - X 29 Y 49 and Z 81N/um so all good for stiffness. Now its a matter of weight reduction... I'll get out the CAD Ryobi -- Peter

the combined bending and torsion of X has always been the main deflection, so will look at that closer as well...

[StrawberryBoi]

If you have access to it, a topological optimizer like in Inventor or Fusion360 makes great work of lightweighting and developing a case specific trussing for you.

Other than that I look forward to seeing what you develop! A truss is a good candidate with the stiffer steel now in play I think.

[peteeng]

Hi Strawb - earlier on in this thread I did a lot of isogeometric optimisation using Inventor. The issue with this is that 1) mostly the shapes can't be made unless you 3D print them which I'd love to do in Titanium and 2) as the mill moves around the resultant shapes change. But its good to get some other outlook on the problem like me now looking at steel. Maximising geometric inertia and attending to local shear stiffness is about all you can do... So I adjusted Milli-steel number 3 to 500x300x450 high envelope and made the column 30mm thick hollow (which I can't machine by the way and I'd like to stay away from fabbing) in X it is 22N/um so above target so Y&Z will be as well. It now weighs 800kg and the base is 520kg and 100mm thick.... So this seems to be a good path. So I'll have a little rethink, regroup and light a candle for all of those In Afghanistan that are about to be put in harms way. Peter

[peteeng]

Afternoon All - before I start on the new phase of this design I wanted to check if a circle or square section was stiffer in torsion. The square is stiffer. This makes sense as its inertia is bigger in the same envelope. So I modeled a 150mm square 30mm thick SHS and a Dia150mm x30mm thick column and applied a 100kg load to a torque arm. The square is 0.022/0.016= 38% stiffer. So max out the geometry at every opportunity. The proviso of this is if the square is thin then shear stiffness in the corners is low and the section will lozenge. Then the circle may win... So has to be thick for shear stiffness and for vibration minimisation. Off we go. Peter

[ardenum]

Afternoon All - before I start on the new phase of this design I wanted to check if a circle or square section was stiffer in torsion. The square is stiffer. This makes sense as its inertia is bigger in the same envelope. So I modeled a 150mm square 30mm thick SHS and a Dia150mm x30mm thick column and applied a 100kg load to a torque arm. The square is 0.022/0.016= 38% stiffer. So max out the geometry at every opportunity. The proviso of this is if the square is thin then shear stiffness in the corners is low and the section will lozenge. Then the circle may win... So has to be thick for shear stiffness and for vibration minimisation. Off we go. Peter

You're making a steel table for this right? So how about you separate the bed and alight the parts using these little buggers? With a steel table you don't really care about the sum of the parts as long as you can easily lift every piece.

[peteeng]

Hi Ard - I always care about the sum of the parts! "Easily" lifting the parts will be the issue. Its probably a motor hoist lift. But I'm cutting it down down down to benchtop... Since its small all parts will/maybe steel... need the material stiffness. Since the parts will be billet machined they will be aligned with dowels. So here's where I'm heading - billet machined parts that bolt together to make hollow parts. Like gearboxes and motor parts. First attempt get X 49N/um very encouraging. I will close out the top... each half weighs 41kg so its manageable until bolted together. Peter

[StrawberryBoi]

Hi Ard - I always care about the sum of the parts! "Easily" lifting the parts will be the issue. Its probably a motor hoist lift. But I'm cutting it down down down to benchtop... Since its small all parts will/maybe steel... need the material stiffness. Since the parts will be billet machined they will be aligned with dowels. So here's where I'm heading - billet machined parts that bolt together to make hollow parts. Like gearboxes and motor parts. First attempt get X 49N/um very encouraging. I will close out the top... each half weighs 41kg so its manageable until bolted together. Peter

Oh god, isn't lightweighting fun when it still performs! That's a nice looking column now. At this point though would it make sense to use square tube and machine flat the two working surfaces instead?

A moving I-beam column could be an interesting design, you're giving me ideas.

[StrawberryBoi]

Not suggesting you print this in PLA, but good support to using a gantry if you make it from steel and not PLA, where you specifically want to cut weight but can occupy a large volume.

https://3dprintingindustry.com/news/repraps-new-open-source-software-3d-prints-pla-beams-as-stiff-as-steel-194515/

[peteeng]

Hi Strawb - The claim that PLA plastic can produce the same stiffness beam as a steel beam is interesting . It maybe the same rigidity as a tiny piece of steel, in the articles case a 5mm section, PLA stiffness 3.5GPa steel 200GPa... but this does not scale well to large structures. Then there is the issue that plastic changes shape over time especially printed plastic.

Your idea about using an I beam for a column is poor. "I" sections or any open sections have extremely low torsional stiffness and this entire 1000 thread tome has been about chasing torsional stiffness. Do some FE on it to see. Ensure your FE system handles shear deformation correctly as some simpler linear systems do not model this accurately. If you have a non linear or large deflection solver use that to show up the issue better.

https://omnexus.specialchem.com/poly...ties/stiffness

Re- SHS or RHS tube - There are several threads here about the issues using cold formed tube. I'll summarise, firstly the aim is to produce a straight part. 1) rolled sections have high internal stress due to the cold forming process. This stress will release when you machine it and the section will change shape. If you get a small section and cut it longitudinally you will see it spring open 2) These sections are also welded together further increasing the internal stress and further adding to the stress/strain release issue 3) The sections are not geometrically accurate or relatively thick (the sides are cupped or crowned) so its usual to add plating to the side you want to dress. This is usually welded on, again warping the section and changing shape again when you machine it., Solution is to thermally or vibration stress relieve the structure prior to final machining. All of this is possible but my aim is to make parts that can be bolted together off the mill so I will start with a normalised billet.... or plate... my aim is make the most economical production machines possible that fulfil the 10-20N/um target stiffness... to achieve this requires minimisation of processes, freight costs (which now impact me due to closed borders to where I get stuff done - material, heat treat, painting, sheet metal work, plating etc I can't go there!!! I'll have to start up a courier biz to get permits) and maximisation of accuracy. Peter

PS - maybe a courier biz is better then a cnc biz at the mo??

[StrawberryBoi]

Hi Strawb - The claim that PLA plastic can produce the same stiffness beam as a steel beam is interesting . It maybe the same rigidity as a tiny piece of steel, in the articles case a 5mm section, PLA stiffness 3.5GPa steel 200GPa... but this does not scale well to large structures. Then there is the issue that plastic changes shape over time especially printed plastic.

Your idea about using an I beam for a column is poor. "I" sections or any open sections have extremely low torsional stiffness and this entire 1000 thread tome has been about chasing torsional stiffness. Do some FE on it to see. Ensure your FE system handles shear deformation correctly as some simpler linear systems do not model this accurately. If you have a non linear or large deflection solver use that to show up the issue better.

https://omnexus.specialchem.com/poly...ties/stiffness

Re- SHS or RHS tube - There are several threads here about the issues using cold formed tube. I'll summarise, firstly the aim is to produce a straight part. 1) rolled sections have high internal stress due to the cold forming process. This stress will release when you machine it and the section will change shape. If you get a small section and cut it longitudinally you will see it spring open 2) These sections are also welded together further increasing the internal stress and further adding to the stress/strain release issue 3) The sections are not geometrically accurate or relatively thick (the sides are cupped or crowned) so its usual to add plating to the side you want to dress. This is usually welded on, again warping the section and changing shape again when you machine it., Solution is to thermally or vibration stress relieve the structure prior to final machining. All of this is possible but my aim is to make parts that can be bolted together off the mill so I will start with a normalised billet.... or plate... my aim is make the most economical production machines possible that fulfil the 10-20N/um target stiffness... to achieve this requires minimisation of processes, freight costs (which now impact me due to closed borders to where I get stuff done - material, heat treat, painting, sheet metal work, plating etc I can't go there!!! I'll have to start up a courier biz to get permits) and maximisation of accuracy. Peter

PS - maybe a courier biz is better then a cnc biz at the mo??

LOL yes, no kidding. I wasn't thinking just a straight up I-beam, incomplete thought that I spat out there.

Definitely good to know about the formed tube! I work for a company that focuses on value engineering and hogging out billet is like value design enemy #1, at least 1 of our industry competitors took a $100k BOM and made it almost $750k BOM by going the billet approach, but the precision requirements are not the same.

[peteeng]

Morning All - The 49N/um seemed too high, so this morning I checked things through. I compared a bolted connection with a "bonded" connection (careful technical language alert) I used a 0.3 friction coefficient. I mucked with the preload a bit but not much diff. So the bolted connection is 15N/um and the bonded is 18N/um which is a 20% delta so going fwd I won't bolt until near the end. Plus the connection will use dowels and this will stop the slight slippage as well. Billeting these will create much swarf so plate construction maybe the go. This is sort of where I started some time ago then went to the casting idea... I also found an article on EG in which it was tested in tension. at 12% epoxy addition and using quartz it achieved a modulus of 6.0Gpa and a flexural modulus of around 13GPa. Both of these are quite low properties compared to others. They did not quote compressive modulus which would complete the picture.. Vibration mitigation will be the next layer of thought in a steel mill... Peter

[peteeng]

Evening all - Today I looked at a steel plate column. Will create much less swarf then the billet machined one. 20mm plate got X 23N/um so that's good enough need to work on the machine base. Six pieces weigh 91kg so that's Ok. Once the base gets detailed a bit then I'll try to spark up the pieces design a bit...Peter