I had a lot of fun, it was really interesting to a lot of other peoples projects. My wife said she'd help me setup a booth if I wanted to do one next year. I got all of my replacement part prints done this weekend on my big delta, I'm hoping that stiffens it up enough, if not I'll probably miter cut my extrusions and drill, countersink, and tap everything and add plates for extra stiffness.

 

Ah, ok. I was thinking it would be a little looser like the Maker Fair Primes over on the West Coast. Still, you got to show off your progress and direction as well as OpenBuilds, so it was still reasonably successful. Next year should go much more smoothly!

That would make sense to me. Apparently after the first month or so the shrinkage becomes on the micron level anyway. That said, epoxy concrete only takes a week to fully cure, with no shrink...

Vibration is apparently the original intent of epoxy-based composite structures- epoxy-granite/quartz is an excellent vibration dampener and steel machine structures can be filled with it to increase accuracy and tool life. You don't need to worry about its structural integrity when you're just filling a steel tube with it, either. No vibration or vacuum is really necessary. Might be a good direction for you for a production line, since all you really need for the steel is a drill press for mounting bolts/brackets- welding isn't strictly necessary (I don't have a drill press, hence my fascination with epoxy). It would be quick and easy.

Not sure I'd go with rubber, since it would probably swing the stiffness:damping ratio too far in favour of damping.

No, it's not based on math, more on how many concrete-based machines I've been looking at lately and their general gantry support height/width ratio. They're typically very fat, and I'm not sure you have enough mass on those uprights to support the kind of forces you're looking to generate. Some kind of calculations would be handy, though, you're right. It probably ultimately depends on the mass of your milling head assembly and arms, and your projected feed rates. Some kind of computer engineering model would be handy there.

Castors are available with very high load allowances- up to thousands of pounds each. I saw some on Grainger that could do 750lb for $50 apiece- with brakes. I'm just not sure about the accuracy of a machine tool on castors. Maybe if you could lower the machine onto a fixed base, then raise it up to move it? Seems like vibration would concentrate at the castor hubs.

I think that's the best idea. I'm up to page 80 of the infamous Epoxy Granite thread on CNCZone, and if you're looking at short lead times on production units, that's the way to go. Just make sure that your steel structure is fully capable of 100% of the load bearing.

Experimentation is good! I'm hoping to get some pool filter sand, quartz pea gravel and maybe some ceramic microspheres, hit up US Composites for 600cpm epoxy and make me some bricks to smash soon. I'm planning on making more simple structures before attempting a big (well, a couple feet cubed) fixed-gantry mill, like a bench sander, flywheels, maybe a simple wood lathe bed... A tape dispenser, even!

I'm not entirely sure about reinforcement yet. I may go with some chopped 1/32" glass fibres, to see if they affect impact resistance or bending/breaking forces, but it seems likely that I'll end up having to compromise on the vibration front somewhere and add some threaded rod reinforcement for tensile strength.

Hehe, yeah. I doubt it'll matter unless you're planning on some serious stock removal though.

 

Rick's on the right path.

I'm now almost finished the CNCZone EG thread, on page 382, and I can now pretty categorically state that 1) Concrete is next to useless for the type and style of machines we make because it has no tensile strength and low damping, and 2) there's absolutely no way you could even use mineral casting (epoxy quartz) in the dimensions you have, even with internal reinforcement, due to both tensile strength of epoxy and shear modulus (-> stiffness) of steel issues over that length of beam.

Polymer additions to concrete don't appear to be very helpful either- vinyl ester, acrylic and polyester, the typical resins used for polymer concrete, are all too soft, too cure shrinking, too much coefficient of thermal expansion, and/or too chemically non-resistant to be useful for machine tool construction. Epoxy is the only viable resin for hardness (particularly "tabletop epoxy", which also self-levels to very high accuracy), tensile strength, low shrink (as long as it's VOC-free, ie. non-accelerated), thermal stability and chemical resistance (mainly coolants and lubricants, in our case).

For the base, it'll have to be epoxy-based mineral casting if you're going to get out remotely cheap and avoid expensive months-long metal casting processes or weldments. It should probably be about 6-8" deep, where the concrete round in the photo looks more like 3" deep. Tossing some threaded rod into the form probably wouldn't hurt, and you could use it to attach other parts.

For the uprights, there are basically three options: 1) thick wall square steel tube, probably about 2.5-3", filled with epoxy quartz for damping. 2) ~6" square epoxy quartz columns with threaded rod and/or carbon fibre tube for stiffness (in the tensile region and neutral zone resepctively, I think). 3) 10"+ epoxy-quartz-only columns, with probably a slight reduction in overall height in order to reduce the torque moment about the anchors. Your EQ columns will have to be significantly thicker at the bottom than the top, too.

Those are obviously math-free approximations, you'd need to crunch some aggregate numbers and do some kind of finite element analysis on the results to get absolute answers, but yeah, this is an engineering challenge, for sure. Definitely not just a case of pouring some concrete and having at it.

Put it this way: if you have a big heavy spindle trying to push through metal, the downward force of a column is imparting a force proportional to the compressive force on the head lever (itself proportional to feed rate and workpiece hardness, I guess) and the angle of the lever to the upright. That's not counting sudden forces from the spindle encountering resistance or digging in. The magnification of the stepper and spindle forces through that lever are placing quite significant cross-loads on your uprights.

I think it's time to make a decision on whether a delta bot is really worth the headache, or scale back your "industrial machine" aspirations. Cartesian machines are more popular for high power applications for a reason, I'm now realising!

 

I finished the thread last week, still have the same view, though marginally relaxed on the production tolerance front after the last few pages. I haven't yet started digging into the associated literature, but Principles of Rapid Machine Design and Concrete Mixture Proportioning (Bamberg and de Larrard) are definitely on my to-read list!

Reinforcement with regular concrete may end up having to be so substantial that you essentially end up with a steel machine. Not that the idea of a non-welded machine from slotted steel with added external damping hadn't occurred to me- rapid bolt-together power CNC would be quite an achievement! I think we're still aligned on the overall concept of simplicity, power and non-expertise (in terms of end users building and maintaining). I think my concern is that trial and error may prove to be a longer, more expensive process than you may be anticipating. Hopefully not, though.

Interesting, I'll have to look further into the damping of reinforced concrete... Given how readily concrete "tinks" rather than "thuds" on impact, I suspected that the damping was quite low, but didn't have a source to pull from when I made the comment. I imagine that it is reasonably on par with cast iron, but hit the nearest anvil and you'll see how damping that is! Haha. The numbers thrown around as well as the subjective testing in that thread indicate to me that EG is a significant improvement over either of them in damping- as well as all those other properties I mentioned. I should have the opportunity to test that soon empirically, though.

I'm not sure how much you're paying for concrete, I hear EG comes out to around 60-80c a pound. Concrete's probably cheaper. My issue is trying to get aggregate without having to pay shipping on 50lb bags and drums cross-country. That said, hardness, fracture toughness, etc all being reasonably equal, aggregate sizing seems to be much more important than actual material. I may look into getting some cheap sieves and see if I can't come up with a "Home Depot Mix". Whether that will be sufficient for my final purposes, I don't know. I was, er, "financially motivated" to switch recently towards an initial lower power (maybe 400W?) 150x150x70mm benchtop manual mill/precision drill design using V-Slot and then proceed to the 450x450x200mm CNC desktop gantry-style later. That should provide me with a testbed for a cheaper mix in the shorter term.

Really, the thing that excites me the most about epoxy is its self-levelling. It's possible to both create a gorgeous, glassy machine as well as almost measurement-free surface production to 0.03%. The only bit that matters is ensuring mating surfaces will be exact right angles, though even that could be designed around so that the actual parts don't have to mate perfectly, but can be later squared and trammed to tolerance after construction. That would be preferable, I think.

If we stuck to just milling wood, we wouldn't have these problems.

 

Of course we will!!
This thing will be awesome, and the journey is nearly as good as the destination.
Good luck, thanks, and best wishes.

 

Why cannot steel be used, with temperature compensation for expansion incorporated into the computations for the movements needed from each stepper motor?

 

The TCoE I refer to there is specifically the differential between the steel and the concrete- the structure delaminates itself. That may or may not be significant, depending on the specific circumstances. Under stable conditions this obviously isn't a concern though. I believe aluminium has a much more similar TCoE to mineral casting, but it's also far better at transmitting vibrations.

 

How's it all going? My big delta has been a workhorse for a while. I've worked on making some custom (printed) parts for it and everything has worked out well.

 

Interesting, and why is that?
If it's the computing power then a Arduino Due with a ramps-FD would be equally capable i guess,
or is it a software/firmware issue which causes better accel and jerk handling?
Which firmware are you running on the beaglebone?

 

Hi Brian,

Good to see you back making progress again - I am really looking forward to reading series II.
Although some of the things / processes you have tried so-far have not been 100% successful I have learned a great deal from your research. Keep up the good work my friend.

My regards to Boo and Lord Zulu.

Tweakie.

 

Brian, where did you get your metal tie rod ends? I've had good luck with my big delta over the last year, but my tie rod ends are starting to wear down now and I've compensated by preloading with springs, which works fine until they wear down so much that it restricts movement.