[peteeng]

Hi Thomas - yes I remember that. Cement is interesting chemistry, its more complex then its solid ratio. It has ingredients that modify surface tension, particle dispersion and attraction and the binder is very very fine. Its hard to find for me, useful technical info on it as they (civil engs) are looking for other things then what we machine builders are looking for. The E90 material is silimax with alumina/alox/carborumdum etc added. Its time for me to move along to building vs research.... Peter

edit - 1) Thomas would you send me a cup of Silimax? if so PM me for my address.
2) Heres Plank2 in the honey warmer 30C and my electric blanket

[badhabit]

...Thomas would you send me a cup of Silimax?

I have pleanty, but I've looked at the shipping prices and it's around 150AUD to send a small package(4kg) down under:-/ It also seems kinda strange to ship "sand" around to world;-) There must be a similar alternative "local'ish" in Australia...

Anyways, i did experiment with Alox myself, before settling on the Silimix282 as my best option all things considered. I actually bought a "set" of 5 different sizes of high-grade/pure Corundum(the white crystal stuff) and made a Fuller-graded aggregate mix to get a feel for it and test it, but never meassured any actual properties on it...hence my interest in your tests...i guess i could do that if the issue ever araises again...but i doubt it since it's way more expensive than the Silimix, so it's not really an alternative...

One other thing: When i wetted out the Silimix282 with water in the test i mentioned above, i saw the volume shrink quite a bit, which puzzled me a bit(it actually still does), however it was rather significant. It could be something to wonder about, when you are so keen on the dry-stacking stuff, not sure if it affects your tests since you don't grade your aggregates, but i think the fluid helped the different sizes of aggregate to pack even better...

I might redo the water vs. silimix282 again and then use my concrete vibrator to release the trapped air and see what the volume shinkrage actual is...

I guess we'll never agree on neither the grading of aggregate nor the drystacking, but i'm looking forward to your plank-tests :-)

/Thomas

[peteeng]

Hi Thomas & all others - a cup would do not 4kg! I suppose I could go to the beach and get some for free. When you did the water volume test on Silimax did you dry the sand? using an oven? With a very close pack as I expect SM is, it will have air in it, this releases when wet and the bulk decreases.

I see this at the beach with our fine white sand if you place it in a bucket and add water it debulks...

I'll re-phrase my stance on grading: I don't see a leap forward in modulus from the grading. The technical docs don't support this, some of the tests in threads here don't support this, my tests don't support it and the more I look the less confidence I have about it. I want a number I can design with and I want a simple repeatable production process. The people that go deep into grading (diy machine builders) and do it don't test so don't know what they have. Many are grading to reduce cost not increase modulus. The civil engs are interested in compressive modulus so they can build taller buildings that don't buckle and only test in compression. Its a difficult area. So you have quartz which is 70GPa modulus even if its 100% efficient at 80% volume fraction its 56GPa which it isn't. So at 50% strain efficiency its 56/2=28GPa which it could/should be. I can build fibreglass (FG) laminates at 30-32GPa, tension, compression and flexure any day I want plus its 500MPa strength plus. Been testing laminates at this for years. If I build in FG I can machine it, if I build in concrete I can't unless I use inserts or diamond tooling. I want a material that gives and does everything! and the current answer is laminated aluminium.... E=70GPa damp, light, machinable, easy to make, can be bent, laser cut etc etc.... If I use local CSA grout I'll use 35GPa but I'll check this out in Plank3 maybe... tomorrow morn I release Plank2.... as I know the vol fraction I can check the maths. If attractive or close then I'll consider grading to improve the modulus but if it has poor modulus vs its volume fraction then there's no point, its the end of the road for that idea.... laminated aluminium is the clearest best defined direction at the moment. Thanks for participating appreciate all commentary appreciated... Peter

I'm reviewing the build costs on Brevis-HD/YaG /Scoot at the moment and I'm having some of the parts requoted in 3mm gal steel, 5mm aluminium and laminated versions of these vs thick stainless steel. Will be an interesting costing exercise...

[peteeng]

Morning All - There's good news and bad news.

Short story - Plank2 released well - being waxed its slippery and it slipped off the horse and broke - like Plank1!! Bugger. But still able to see deflection with short piece and its greater then the aluminium. Tested both ways, different clamps, even feels softer. So conclusion is can't achieve stiffness of aluminium. End of the road for this development.

Long Story - Should have used flash tape over screws. Heads filled with resin and some where hard to remove. Plank2 has a bit of porosity. Probably due to cold temp and thick resin. Calcs show that its 11% by weight of Alox (should be 18% if zero porosity). By volume its 61% Alox 25% resin and 14% air. If alox is 300GPa then 300x0.6x0.5 E=90GPa which it clearly is not. If alox is 200GPa then 200x0.6x0.5= 60GPa modulus which it is not by eyeballing... 6000 series aluminium is 69GPa.... As cast its density is 1.484/.000562=2640 which is less dense then aluminum at 2700kg/m3 so this should result in less deflection, again not true... deflection is more then twice that of the aluminium bar so ALox modulus maybe 30GPa. May as well use CSA out of the bag...

Conclusion - I have failed in finding a castable, machinable material that has a modulus of E70 unless I use carbon fibre (but CF too many $$$) . So aluminium is the current answer and to make it damp use a metal laminate. Cheers Peter

[peteeng]

Afterthoughts - thinking about the results and using ROM (rule of mixtures) ROM works well for long fibre composites and short fibre composites with high fibre volumes. Sand and granular materials however have a poor internal strain transfer efficiency in tension so my usual efficiency factor of 0.5 is too high. Its more like 0.2 or 0.25 This makes using roundish granules quite inefficient. So I think the two courses are: if I cast is to use a thick fibreglass skin (allowing post machining) with a sand infill if it needs to be solid or FG skin with inserts. This gives a modulus of 30-35GPa. But may as well use cast CSA with inserts much easier. If I use laminated aluminium I can get 70GPa and post machine on my router or give to any machinist. If I use a machinist I can use steel outers and al inners but this maybe too mixed with the CTE (coeff of thermal expansion). So review Milli once more and start again.... One objective of Milli is to have a mill I can mill composites on. I still haven't found a machinist that's happy milling FG or CF... Peter

[peteeng]

Evening all - Been thinking about Millis spindle. Anyone have experience with this sort of spindle?
https://www.aliexpress.com/item/1005...%2223416%22%7D

https://www.aliexpress.com/item/1005...%2223416%22%7D

Peter

[joeavaerage]

Hi Peter,
both of those look to be good value.

As I've posted elsewhere I made a spindle from a 1.8 kW 3500 rpm Allen Bradley servo direct coupled to a cylindrical ER25 toolholder
running in P4 angular contact bearings. Despite being homemade I am quite proud of it. I have not used it a lot to date because the torque/power
of the spindle showed up the modest rigidity of my (then) machine, a mini-mill.

I am in the process of exploring the envelope of my new build mill and being that much more rigid am finding my Allen Bradley based spindle is a good
match and getting plenty of use.

I have come to understand that certain of my design goals back when I built it , 3-4 years ago, are actually not that important. To whit I thought I needed a
really power/torque dense motor and that in turn lead me to an AC servo. They are hugely power dense....but does that matter as much as I thought?
My servo weighs about 8kg but a regular induction motor of the same power might weigh 12-16kg. The extra weight is still a small fraction of the Z axis capacity
of my new build....so why was I so fixated on power density.? Second was that I wanted genuine position control capability so I could do coordinated threading
and/or indexing.....only to find that I use those features so infrequently that all that extra performance and cost is wasted.

One design area that I now consider even more important than the motor is the spindle/toolholder/bearing/housing combination. Absolutely the best quality
and accuracy of these parts is essential for a good spindle.....and the motor comes second. What's more the quality and accuracy of these parts is the single largest
determinant of cost of the overall spindle.

Milli is about a high performance mill at affordable price. Given that is the case I would suggest researching the mechanical spindle components first THEN look
for an affordable motor, and even a humble induction motor can be surprisingly good.

Craig

[peteeng]

Hi Craig - I looked at induction motors and they are quite large. I've been looking for a direct drive solution to sidestep pulleys or gears. I agree the spindle is part of the heart of the machine and have spent quite a bit of time on this in the background. No clear answer yet and maybe there won't be one. If I stated that the mill was only for aluminium there are clear solutions, but if steel & al is required its a challenge. Peter

[peteeng]

Hi All - The Milli thread has had over 41,000 views and over 1,000 replies so it seems its a thread with good interest. Just to ask you all, is this thread interesting because of the material investigation, the Mill design investigation, the engineering or the poor humour? Peter

[koenbro]

The design process, the engineering and the erm, humor. In this order.


Sent from my iPhone using Tapatalk Pro

[RubeG]

Hi Peter, my interest is primarily design investigation and the chance of a well designed mill is for sale/ constructable option available followed by material investigation, great decorum and good humour. Cheers

[peteeng]

Morning all - I'm about to do a border run to pick up stuff that I can't get local. Wish me luck as the rules keep changing... I have been helping one of the members in his early design stage of a new router. I imagine his Q's are similar to many about this sort of thing so I thought I'd put down a bullet list of how I do it.

1) work with 3D CAD - its worth it
2) write an objective for the machine and how you want to go about it. This clarifies some issues like heat treatment processes (welding bolting etc) and its target accuracy. This will ensure you buy or at least budget for the correct tolerance bits like class of ballscrew
3) as the design develops refer to the objectives to check your on track or have veered off somewhere.
4) work with scaled hand sketches early on, this is quicker to develop ideas then in CAD until you become quite good in CAD
5) Once you get onto a serious pathway your first aim is to create a good general arrangement of the machine. The GA will consist of logical sub assemblies. If this turns out to be the "one" then you need to bring that GA to a manufacturing level asap
6) Create sub assm drawings with BOMs on paper. Paper is important for keeping track of notes and part details. Once dwgs are established these are the top level documents not the models
7) Now you have dwgs of assemblies, use the BOM to create part dwgs. Tick each part dwg off in the Assm BOM as you go. Once you have a part dwg for every part your very close to manufacture

8) Add manufacture notes to all part and assm dwgs
9) Review and check all holes line up
10) your good to go. But put it aside and let some time pass then review again. You can be too close to a project and miss stuff or assume stuff.

Good luck with your builds!! Peter

[ardenum]

Hi Craig - I looked at induction motors and they are quite large. I've been looking for a direct drive solution to sidestep pulleys or gears. I agree the spindle is part of the heart of the machine and have spent quite a bit of time on this in the background. No clear answer yet and maybe there won't be one. If I stated that the mill was only for aluminium there are clear solutions, but if steel & al is required its a challenge. Peter

Hey Peter, Mechatron makes these adapters that fit a regular spindle, I suspect you could run these with a servomotor instead, the catch is price, another catch is you get atc, I think that balances each other out quite well. (https://www.spindel-shop.de/en/stc-tool-change-adapters). Another option, you could buy one and reverse engineer it.

On a side note, in your study of CSA/UHPH did you come across the maximum permissible load per cm3 or cubic inch?

[peteeng]

Hi Ard - Thks for the info.

Re - Grout, here's the data sheet for the CSA concrete I would use to make machine parts. It has Modulus, compressive, tensile and flexural strength properties quoted. The aggregate lifts the modulus about 5GPa from the plain HE80 material.

By the way stress is related to cross sectional area not volume. Peter

[peteeng]

Hi All - Starting to think about Milli again. Heres a big Okuma lifting gantry "mother machine" Peter

https://www.ctemag.com/news/articles...other-machines

I was at the machinist yesterday picking up parts and discussed joining various parts together. We decided that bolting and dowelling would work for what I had planned so I'll start reorienting my thoughts around that, plus the laminated aluminium....

[StrawberryBoi]

Incredibly fascinating build pete, if you need it I have some experience in designing and also working on/with servo motors and rotating assemblies. I can sound off on any questions or ideas in that regard.

[peteeng]

Hi Strawb - Thanks for the offer. Milli was going to be my first servo machine so stay tuned. Peter

[peteeng]

Morning All - Since we rely on bolts so much I thought I'd look at a bolted connection as a separate model. I made a T model of aluminium 16mm thick and 200mm long. I connected it with 2 M12 bolts and preloaded them to 32kN the nominal preload for a C8.8 bolt. Class 12 can be done up tighter... I loaded the T in tension, compression and bending. I made the bolts as best analytically as possible and used a friction coefficient of 0.3 for the joint and made it a separating connection. So its as close to practical as this software can make it. The tension case is 95% efficient so If I used 3 bolts it would be better. The compression case is 90% efficient and the bending 42%. The bolted connection is compared to a 100% bonded joint by the way.

So this sort of construction depends on the global geometry to keep things intact if the joint is in bending... I may update the model and glue (loctite) it together as well to see if it makes a delta. Peter

[StrawberryBoi]

Morning All - Since we rely on bolts so much I thought I'd look at a bolted connection as a separate model. I made a T model of aluminium 16mm thick and 200mm long. I connected it with 2 M12 bolts and preloaded them to 32kN the nominal preload for a C8.8 bolt. Class 12 can be done up tighter... I loaded the T in tension, compression and bending. I made the bolts as best analytically as possible and used a friction coefficient of 0.3 for the joint and made it a separating connection. So its as close to practical as this software can make it. The tension case is 95% efficient so If I used 3 bolts it would be better. The compression case is 90% efficient and the bending 42%. The bolted connection is compared to a 100% bonded joint by the way.

So this sort of construction depends on the global geometry to keep things intact if the joint is in bending... I may update the model and glue (loctite) it together as well to see if it makes a delta. Peter

Awesome to see the sim results, the one thing I would add as a caveat on the bonded constraint is that when using adhesive bonding there will be a small gap/groove required for bond line thickness, and the adhesive will act as a stiff spring. Adhesives that are super rigid are brittle and not suitable for vibration.

One I've used extensively in high temp applications is 3M DP-320, as well as Henkel HT-40. Both are very good structural epoxies for bonding metals, both have great strength to high temperatures and withstand vibration. Under high loading both behave as a spring and not as a rigid connection as assumed by "bonded" constraints. In these cases you will want to create your bond line and add the volume of the adhesive and use the adhesive mechanical properties if available. It is a plastic part in the end.

If the properties are not available you can go with the bonded constraint, but watch the stresses, you may be approaching the limits of the adhesive in the bending operation.

Most adhesives, like most plastics and epoxies, are incredibly stiff in compression. Where they gonna go? No air in here to compress, buddy! But in tension they form hourglass shapes, with the center that's not bonded on the sides pulling inwards.

If you make a fully captive bond that would solve that issue, but introduces another (no where to off-gas when curing). So you'll need to use a gas emission free adhesive if you fully enclosed the epoxy. IIRC both the above mentioned also fit that bill. We used DP-320 for full encapsulation in a sealed mold that was pulled to vacuum prior to injection.

[peteeng]

Morning Strawb and others - Please pester the site admin to include mp4 videos so I can upload those. So much better then images. To explain about "bonded" connections to people. In simulation or FE work there is a thing called a bonded connection. It means the parts are connected together with a perfect elastic connection. It would be like a perfect brazed connection with no "reinforcement or fillet" if the parts where metal. Then there is an adhesive connection. In this you can specify the adhesive thickness and stiffness. Same as a weld, you can model the weld geometry or you can use analytical welds that include the fillet. You can specify the fillet size on each side of the connection. I use simsolid for this sort of thing.... The other aspect of the T model is that I have used very high loadings to create moderate stresses in the parts. About half way to yield. In a machine part, as it is designed for stiffness the stresses are extremely low. The purpose if the adhesive (loctite say) in this case is to distribute the joint contact better not to hold the connection together that's the bolts job. If you bolt up a joint its very unlikely it has 100% contact unless you lap the faying surfaces together which we don't do.... Peter