Welcome to the SeeMeCNC Community

Hello, all. This probably isn't a very good first post, but oh well. I've been lurking here for about two years and I'm pricing out parts for a RMv2 of my own right now.

But, something struck me as a thought. Delta math is difficult, but we have the hardware power in a cheap desktop to render incredibly dense poly meshes, and that tech is easily minaturized- in fact in things like the Beagleboard, we already have it so that's a non-issue.

So, I was thinking. What if we built a hexagonal 3D printer? Looking at the way a delta works, it's fairly easy to surmise that precision of the hotend will be affected by the points of the delta's triangles being closely sync'd to the center of the hotend nozzle. Terribly messy AutoCAD drawing of the concept of a delta's central point here; a corner of the polygon's outside is a tower; the intersection at the center is the ideal location of the hotend. EDIT: Yes, I know the "delta" screenshot is cockeyed. I accidentally scrollwheel-clicked and spun the thing as I was taking the screenshot. It's an equilateral triangle, it's just viewed from a weird perspective. Sorry.

[img]https://i.gyazo.com/a1642030d6335ebe117 ... 10751a.png[/img]

The angles are fairly wide between the towers; there is significant room for play. If we move up to a hexagonal formation, as shown below, we should get a significantly stiffer printing platform, and with that, theoretically, a far faster print speed. I'm actually going to look into this.

[img]https://i.gyazo.com/57ee4d28b1c7bbed15b ... f25d32.png[/img]

It is an intriguing concept, but I do want to get my delta printer up and running. There are a LOT of technical challenges for this; for one, I don't even know how to code a delta's basic operating system so it CAN do the math.

In any case, I put this here because this seems like where it belongs. If I'm wrong the mods are free to move this where it belongs.

An intriguing concept indeed. However, I believe we have more play in the rod ends, carriage drivetrains, ETC, that would not be greatly reduced by going to a hexagonal design. Not to mention that we have a pseudo-hexagonal setup already where each carriage carries two arms which link to the effector, giving us a small offset in angle for each rod from the perfect location. If we carry this over to the new design, then we will need to accomodate no less than 12 rod ends or other universal joints at the effector.

I believe that you could approximate it by driving two delta's worth of Kinematics simultaneously with the angles offset, however there are problems with this. Firstly you would need to synchronize these calculations fairly well, which means that path planning must be well implemented, and also you would need at least twice the horsepower to drive this, making it a BBB exclusive in almost any conceivable case, which, when paired with the as yet basic support for delta's on the Replicape, which is usually considered the go-to for BBB 3D printing, would not end well. You also would run rapidly into limitations on IO pins on the BBB, as it would have to drive 3 additional steppers, which would eat one off an expansion board (And to use it, you would have to use a Model A replicape, which is no longer available and the Reach, which was only ever made in limited quantities) without even getting into the extruders.

It would likely take a whole pile of home-brewed systems to make it work. If you decide to go for it though, best of luck. (I personally would potentially take a stacked carriage approach with pre-loading springs between them to try to reduce play in the system.

And welcome to the forums. It's far from the worst first post you could make (And I like the avatar/name combination. I need to get around to watch/playing/reading that someday)

If we switched to a magball system like we have for some of the delta arms, we'd be golden. And I figured 12 arms was implied. Yeah, the hardware and software would be hard. I could do the hardware but I don't know how to code that sort of thing. However, most of our mechanisms as they stand are perfectly strong for all practical purposes. If we use linear rails instead of cheapskates then we could get around the bit of wiggle the RMv2 comes with (which is slight at worst, but zero > slight.)

And, honestly, I'd rather just print a larger effector platform out of solid PC for the 12 arms. It'd be cramped around the edges, but it'd be extremely durable and hard to knock loose probably. The extruder platform would probably be just an E3Dv6 as its initial test hotend, though I'm actually looking into installing an E3D Chimera on mine when it arrives.

This was just a post of a theory I had while I was getting my morning coffee and figuring out how to explain it.

If you stick with 6 arms, you can get 6DOF motion, and it's been done, I've seen at least one Linux CNC based Mill designed like this, and there are various high end 6DOF Mills that use similar arrangements.
The problem with going to 12 arms (or 6 pairs of arms) is that your motion is massively over constrained and you'll run into binding issues with any marginal manufacturing variance or inadequately accurate software.

Anything is possible.


https://groups.google.com/forum/#!topic ... ibYgiTxU5U
[img]https://lh3.googleusercontent.com/-B-oX ... G_0340.JPG[/img]

Most rigid delta, EVER.

g.

Reminds me of a 30 ton block of concrete, rigid but does not print worth a damn.