Parametric bracing frame vertex

After reading this thread [forums.reprap.org] and thinking about the issue, I feel the instability of the Prusa frame is due to it hingeing around the narrow part of the vertex, which does nothing to support the square areas of the frame. I've had a go at producing a new parametric frame vertex which aims to stabilise the square section of the Prusa frame, and also allows for the addition of bracing.

[www.thingiverse.com]

It does this by increasing the width of the vertex, strengthening the joint betweeb the triangular part to the square part, and making the depth of the vertex narrower around the area of the (as I see it) unnecessarily-strong constraint of the triangle, which is inherently strong. This maintains the volume of plastic needed to print the vertex. There is also provision in the vertex for cross-bracing each side of the frame with threaded rod, or using a sheet material (plywood or aluminium, for example - MDF may fatigue over time) to constrain the squares of the frame, with a stand off (essentially, a second foot) and an embedded nut trap. The scad file is parametric, so feel free to play around with width and depth of the vertices if you feel that the size I have come up with is compromised in another way.

It was inspired by the shape of the new emaker/RepRapPro Huxley, and can make an accurate version of that vertex, except the foot is a different style to accommodate the nut trap - there isn't an openscad version of that frame vertex. The compromise I reached for the Prusa frame is a vertex 24mm wide (as opposed to 20mm for Mendel, 13mm for Prusa) and a depth of 18mm (30mm for both Mendel and Prusa).

There are a few other advantages of this vertex compared to other vertices; the scad file is much more clearly written (the Prusa version uses some odd bezier-style forms), it scales to different width/depth well (the Prusa one doesn't, and the Mendel one is an extruded dxf that isn't parametric at all), the area for washers on the inside of the tight corner is correctly sized (the Prusa one isn't), and finally, because they are slimmer, you can fit more on a plate of parts.

The only disadvantage I can see at the moment is that it will use up a little X-axis clearance. It's okay if you have around 5mm spare threaded rod each side of your frame; if not, you will lose 5mm clearance each side. You can gain a little Y and Z axis depth/height. If you use the cross-bracing, this will effect the build area, but with the stand-off of the foot, hopefully not too much. If you use a flat sheet for bracing, you'll need a piece wider and taller than the square of the frame, and then cut out a central area for bed clearance.

I am printing out a set of parts, and intend to replace the vertices on my reprap. Currently, this has a threaded rod cable-tied diagonally accross one square of the frame, which helps a lot!

How did you manage to brace the frame rectangle without the brace cutting through the build volume?

I don't think all the flex is due to the vertexes. The M8 studding is is easy to bend over the length involved and we want it to move less than 0.1mm, which requires a lot of stiffness from a bar held only at one end.

Edited 1 time(s). Last edit at 02/07/2012 07:26AM by nophead.

---

[www.hydraraptor.blogspot.com]

Hi nophead. I wanted to ask you if, in your experience, your Mendel frame flexed less than the Prusa frame, and whether this was due to the extra width of the vertices. Any thoughts on that?

I agree with you about the flex in the M8 studding contributing to the bending of the frame, and adds to the 'springiness' to any return from a deformation, which can ultimately build up in frequency. The problem is inherent in the frame design, and beefing up the vertices in the right direction and/or triangulating the squares is the only way to limit it. I'm just offering an alternative to those who already have a Prusa frame, or can't easily/cheaply get aluminium extrusions.

My current brace (a single rod, cable tied) does cut through the build volume, and any cross bracing will. Using the feet as stand-offs will reduce this a bit, but not entirely. I can't be quite sure how much at the moment. I'm still playing around with the design of the vertex-to-brace-rod connector. But what I tend to find is that I print quite a lot of short parts, or singular taller parts can be positioned towards the back of the bed, so it isn't a problem. The braces would be easy to remove for really large parts, though obviously the print would be less constrained. Using a flat sheet (with a central area cut out of it, like the side of an ultimaker) as a brace would remove this problem.

My whole reason for doing this, rightly or wrongly, is the hope I can get the frame stiff enough to make it work for light wood/ply/MDF and PCB milling, where the bracing won't get in the way of the work. I suppose I would be better off with a second machine, but I don't have the space, and I'd need a second set of electronics.

Edited 1 time(s). Last edit at 02/07/2012 07:45AM by droftarts.

For good stiffness in the X-direction, check out Orca v0.2: [reprap.org]

The frame is braced by the stationary X smooth rods and one M8 rod above them. The frame itself is quite stiff in every axis, including X. (Sadly this doesn't help the Z-Y axis assembly, which is bit of a problem area...)

If I wanted to brace the Prusa frame in the X, I'd design the top vertex and Z motor mount into one piece with diagonal support structures running from near the end of the top rods to the triangle frame rods several centimeters down from the vertex.

I've printed out my new vertices and put together a frame. Without the cross-brace, it felt stiffer than a standard Prusa frame, though I think a square-section vertex would be the best compromise, like a scaled up Huxley vertex. I added one cross brace, and now it has zero movement, including with high frequency oscillations. Having Marlin firmware helps, as the acceleration does a good job of damping a lot of the oscillation, but I could probably increase the print speed without losing positional accuracy. I'm getting really smooth vertical parts, but this could have been helped by improving the z-motor-mount so that I can bolt down the motors; the z-leadscrews do not move eccentrically. I think this means that it's good enough to move on to the next phase, which means printing out a vertical x-axis, and then working on my milling mount.

The cross brace does cut into the build area, but because the brace is offset and there is only one of them, it's not too bad, and can be easily removed if need be. The bed and heated bed can move past it, but it is at the limit of movement of the bed. An alternative that I haven't explored is to use a sheet of MDF or ply bolted to four stand-offs, one at each corner, with a square hole in the middle to allow the bed and printed parts to pass through; I managed to put the vertices, with four attachment points, on the same side as the Y-motor, which would get in the way of the sheet. And I don't want to take my frame apart again!

Here's a couple of pictures, before I put the extruder and heated bed back on:

I like the way you mounted the z motors in the bottom. I expect that with this arrangement, you don't have to worry about the threaded rods sliding off the coupler. I am actually planning to do the same thing by printing two sets of z motor mounts and smooth rod clamps and mounting them in the top and in the bottom. This way, I can install the z motors on top or on the bottom. How is this arrangement working for you?

Hi brnrd, It's working really well. Like you say, I don't have to worry about the couplers falling off, no matter how much I push down on the X axis! Layer height and vertical walls are very consistent. I have the Z axis enabled (so it doesn't disable when idle). I changed the standard Prusa Z motor mounts to allow for more adjustablility, which allows me to get the Z leadscrew and Z smooth rod parallel, so the leadscrew doesn't run eccentrically. This allows the motors to be screwed up tight, so they do not move around, unlike most Prusa's I have seen, and the leadscrews don't stick. I also use the Z smooth rod to help stabilise the motors, as mine are long enough to reach the table, and help stop the motor and mount flexing up and down.

However, there are a couple of caveats. Firstly, I have to use short, slightly lower torque motors, though I haven't noticed any problem doing this. These are the ones I use: [www.slidesandballscrews.com]
Secondly, even then, you have to put the lower Z motor mount bar (goes across under bed) above the lower bar of the triangle. This can cause the Y axis belt to rub on the bar, so I put a 608 bearing as an idler on it.
Lastly, it does take extra vitamins; an extra bar, 2 clamps, and a bunch of nuts and washers. I used the standard Mendel clamp at the top, though with a smaller clamp on one side so the clamp can be slightly further out.

I actually think that the best setup would be motors at the top, with the leadscrew supported in a bearing at the bottom. This would take the weight of the X axis off the motor bearings, improving longevity. But the leadscrew and smooth rod must be parallel.

Edited 1 time(s). Last edit at 03/19/2012 01:08PM by droftarts.

I've just seen the developing RepRapPro Mendel variant wiki pages: [reprap.org]
They have elected to brace the frame too, but rather than brace the upper two squares, they have braced the base square. This is probably good for keeping the frame from twisting, but I can't see how it helps the wobble at the top of the frame.

I met same problem, and i figure that the top frame can swing left-right as much as the bottom of Z smooth rod can dip down. Because prusa has only 1 rod supporting it, its probably more prone to flexes like that, compared to solutions which use 2 bottom rods with a more rigid coupling. What i did to fix i left all horizontal rods way longer so i fixed 2 extra sets of vertices each side, totaling 4 sets of vertices triangles overall. I figure i can easily print another set of vertices and the price of some extra rods and nuts isnt all that high. Sort of brute dumb approach, but at the time i believed it was the best way to stabilize everything. Indeed it felt like that at least.
doubled_vertices

But indeed in a comparison i think the way the bottom constraints of vertical smooth rods are much stiffer in mendel than they are in prusa, therefore the top frame wobbling would be less in a mendel. Without adding supplemental structure, the smooth rods are sort of a structural part which provides stabilization against top wobbling.

Edited 2 time(s). Last edit at 04/03/2012 06:15AM by NoobMan.