Looking great Brian!! Although Bo is right, need to organize a bit!

 

Are you using the metal wheels? I think they stopped making the Type III anodized rails. Those cuts look sharp. I'm in the Orlando area and am working on an openbuilds parts delta build too, although nothing quite as beefy. I'm pretty sure I'll make it to the local maker faire to check everything out.

Some of the stuff I've ordered, I ordered more, but haven't updated the list yet.
https://docs.google.com/spreadsheets/d/1PT268f6hHXvjb6wftFS08loYAGB0X351ih2cORe8GhQ/edit?usp=sharing

 

I went ahead and bought tickets. I've got the majority of my parts in now. Waiting on my belt, gears, and carbon rods. I'll probably get a piece of glass cut at ace and run without heat while I dial it all in. I tested out the 20x20 v-slot with a gantry plate while I wait on them to get the wider stock sent, it was even smoother than I imagined it would be.



I didn't start a build on here yet, because I've been winging it with my 'design' I was thinking about bending some metal brackets (hydraulic press and drill press at work) or getting melamine laser cut and making some printed parts to attach it to the rail to keep everything squared up nicely.

 

Hi Justin - nice looking stuff! I especially am impressed with your spreadsheet - NICE!!!

Sounds like you have access to some nice tools!

Not sure how big of a Delta you're building, but if it's not too late, you live in the Orlando area and don't mind changing your plans a little you're welcome to use our forms! (attached jpg)

Hope to see you at the Faire!

 

Looking good Brian

Tweakie.

 

OM goodness - that thing's a monster! Take care & keep up the great work!!

 

Ok, thoughts:

1) Big thumbs up for WWII-style concrete machinery. That idea is in my future too, at some point, most likely on a metal lathe though I may try it on the mill that I have planned next (which itself is to help build a laser cutter next year!). I'd like to find a more environmentally-friendly alternative at some point too, but it works for the time being.

2) Are there prepackaged delta-bot control systems, or are you going to have to come up with a novel method of control? I see a whole lot of simultaneous trigonometry involved over the more simple Cartesian systems, given that it'll still have to run from Cartesian vector->step commands. It seems like an X-Y gantry would still have saved the space over an X-Y table, but allowed a far more simplified control system and ease of reproduction/maintenance. It's a bit late to be thinking about this now, obviously, but unless there are solid software/firmware solutions already in place, the next stage will be quite difficult. But I assume you've done your research as far as that goes.

3) Plasma cutter. Win. Can't wait to see that.

4) Are you planning on having the head self-selectable? I saw you were trying to fit the print head to the bottom of what looked like a spindle enclosure, so I'm guessing you're going somewhere in that direction. A GUI to select between functions isn't much use if the machine itself can't automate its own function selection, after all! I'm envisaging some kind of 5-axis type mill bit selector, with heads that sit in place near the top until needed, with power cables individually going out each third of the support structure. Some kind of mechanism brings them into the centre, drops them into the main head holder and locks them in place. If you have to switch out heads and print platforms and clamps and stuff between runs anyway, it seems like separate control circuits that you manually plug and unplug would be no less convenient.

Those are my initial thoughts, anyway. You may have already covered them and I missed it, though I did read all 5 pages! It's looking very impressive.

What was the fiber-reinforced concrete that you used? I don't think I saw a label for that; I foresee a Home Depot trip in my near future...

 

No worries, it's an interesting experiment with hopefully very practical results. And it looks awesome, more like an industrial robot than most OpenBuilds projects.

Well that's handy! That should be almost all of the way there, just needs a little adapting to milling- which I just had a thought about right now. The actual depth of a cut will be increasingly limited as you move towards the outside of the platform, due to the arms interfering with the workpiece, unless your milling head has its own separate Z actuator. The form factor is fine for printing and cutting, but I think you might hit some unfortunate limitations on milling. Looking at the image below, you'd have to perform some kind of check prior to hitting start, like integrate the arms' total under-volume over the entire toolpath and see if the model "pokes out of" that volume when they're superimposed in 3D space.

I don't think you need to waste memory storing arrays, because the vector calculations for inverse kinematics on this are relatively simple- there are only two possible positions of each belt for any given point in space of the head, and only one that's higher in z (positive theta) than the head itself, so each arm can be calculated independently. If you assign the centres of the three hinges connected to the belts exact 2D coordinates from the bed, you're only solving for Z. For the other end of the arm, I'd probably use a constant to adjust from the centre of the head to the centres of the arm pivots for calculation purposes. If you're keeping it all positive for the 2D coords (ie. arbitrary "top left" of the bed is 0,0- or an assigned column hinge nearby) which is probably advisable with Arduino's limitations on variables, the calculations should be really quite straightforward:

(i,j,k)+(i',j',k') = (x,y,z)

|(a,b) - (x,y)| = d

θ = cos^-1(d/l)

h = l sinθ

z+h=c

where (i,j,k) is tool center location, the primed version is the constant transform between the tool centre and the pivot, (x,y,z) is head pivot location and (a,b,c) is column gantry pivot location. d is flat distance from pivot to pivot, or from the side of the head to the side of the gantry- the bottom of the right angle triangle. l is length of the arm, and h is the height from where d ends to the gantry pivot- the riser of the right angle triangle.



...Which seems a lot when it's for all three arms on every step, but I don't think it's an issue for even Arduino-level computing power. I'm assuming that the Arduino files you have do all of that though, right? There's still the issue with milling, though, particularly at the extremities.

Nice, once you have that, a whole custom GUI to go with it will be perfect. Maybe an entire self-contained package that reads STLs and AIs and stuff too. But I'm getting ahead of myself.

Awesome. Fiber bonding and rebar both, I think, for me. Rebar should protect the overall integrity as the fibers protect the local integrity. It would also reduce the size of the rebar that I'd need and bond more tightly to the rebar with the glass fibres wrapping around it.

 

That sounds similar to my general goals too! Yay!

That's what I'm saying- because the arms aren't fixed like a gantry, but vary in their height depending on the head position, there could be problems with depth on certain models that appear to be compatible with the geometry of the cutter but actually aren't once it starts moving around:



It works ok as long as deep cuts are only made in the centre. Of course, the size of the spindle+mount+Z-rail is also something other people have to bear in mind on their Cartesian projects, so I'm not sure it's so much a limitation as a difference. Just something I thought of, anyway, I don't think it'll be relevant in day-to-day operations.

I'm not sure that you'll see a significant difference in speed between calculations and memory access operations, though it's probably worth experimenting with given that the trigonometry might be relatively heavy for it (I don't know, for example, how many terms it calculates the Taylor series to for the sin operator). Reflashing directly from a control PC (or miniPC) might be the best idea given Arduino's sketch size limitations.

Interesting. I just thought earlier that I may want the unit to be carryable from desk to workbench or something, so I may have to consider a different concrete density. I don't want a 1.5kW spindle hitting 6061 Alu to make it vibrate around all over the place either, though. I may even do some kind of layering of steel meshes and tubing and different types of cement. It needs to be simple and easily reproducible by non-experts, but sturdy.

 

Brian, I'm almost done with my delta now too. I have a Machinekit (linuxcnc) setup that's been tested with deltas and has much better accel / jerk handling than the arduino based boards.

Have you seen the 4th axis delta? That might be a good fit for your scenario. I actually fried two Beaglebone blacks with what I think was a bad PSU, if I can get that solved I wouldn't mind letting you test it out with your setup. I'm running Drv8825s on that board which can only run at 2amps but they can go up in voltage.

 

Good luck Brian - perhaps get Boo doing a bit of overtime and take care of you back (and toes) when lifting the completed frame into the truck.

Tweakie.

 

^Ha! It really is a transporter chamber.

Looking good! Looking forward to discussing it post-Faire.

 

You can put in an acrylic additive to help with that. You could probably fix what you have by wrapping a mold around, taping it up, it and pouring in some resin. We'll be heading that way in a bit.

 

From the research I've been doing recently, for the type of size and capability you want this machine to have, you're gonna need support beams 2-3x as thick, made from polymer/mineral concrete, with carbon fibre or similar (not steel, apparently- some coefficient of expansion thing) tensile reinforcement. This is something I'm currently actively researching and hopefully doing some experimentation with as well. Industrial-capability machines, though, tend to weigh several hundred kg even at our sizes.

Also, Portland cement doesn't stop moving whilst curing for around 2 years, as the water not used in the hydration process slowly leaves the structure. The larger the object, the more it shrinks- so your upright beams are/will be shrinking in the Z axis significantly. Bolting together your rails whilst the beams are still curing means that the anchor points are being held in place as the concrete shrinks around them, putting the concrete under significant tensile stress. Since it's designed to compress (millions of PSI) rather than stretch (a couple thousand PSI?), this will inevitably cause significant stress weakness in the structure and inability to transport.

Also, the steel reinforcement is very effective at transmission of vibration- so even if it had worked, it may not have survived a particularly fast or tricky metal milling job.

I wish I could have contributed this a few weeks ago when you could have reacted in time, but I've literally just learned it in any kind of detail over the last couple of days. I still have a long way to go, too.

Acrylic additive is a good call though, if you can knock together a melamine jig to hold it in time, though the curing may not be done until after the weekend at this point.