UMO Zero-backlash Zero-belts High-rigidity Project

[andywalter 7]

Ok, some of you may have seen my other appends in this thread https://community.ultimaker.com/topic/1114-throw-away-your-short-belts-direct-drive/?page=9&_fromLogin=1 , but it's time I had my own thread & stopped hijacking that one!

I want the best quality I can get from my UMO, by which I mean round things really should be round, and there should be no ringing/overshoot at sharp corners etc. This requires 3 things; zero backlash in the X & Y drives, the highest possible rigidity, and the lowest friction.

 

Backlash means you get flat-spots in round objects where the step motor reverses direction but the print head fails to move until the backlash distance has been moved.

 

Ringing, or overshoot, is a symptom of the drive system being flexible, e.g. press lightly on the side of the print-head with your finger, and watch the ( hopefully cold! Don't want Elfin Safety complaining...) nozzle move. So if you decelerate hard at a corner, the printhead's momentum can carry it beyond the desired stopping distance. Then it bounces back towards where it should be, etc.

 

Any friction there might be in linear bearings in the printhead can make things worse, as it may help the slowdown, but it hinders the accelerate-away at the next corner causing increased forces on the drive system so less positional accuracy. Friction, a.k.a. stiction, creates a kind of backlash-like effect, where the final resting place can be offset from the desired one. The more rigid the drives are, the less positional offset error we'll get due to some small unavoidable friction that we're bound to get.

 

I've already eliminated the rubber belt drive, replacing this with a helical rolled steel shaft which drives a pair of plastic nuts. These nuts are tightened very slightly against each other to eliminate backlash; I will be altering this slightly so the tightening will be done by a small tension spring as the nuts move, rather than being clamped as now. Details to follow when I get round to that.

 

Others have noted that the 6mm dia rods going through the printhead are rather flexible. There's a suggestion to replace with 8mm rods which would be 3.16 times stiffer, but would weigh 1.78 times as much. So to improve that, hollow 8mm shafts can be used which keeps the mass down, but presumably cost a bit more. I looked at getting Misumi hollow rods in UK, but they don't sell to you unless you're a company, and I'm not going down that path of complication just for 2 rods. I couldn't find anyone else in UK making these at anything under 16mm diameter.

 

So I've made my own system, using 4 cheap-but-accurately-made steel rulers, and 24 flanged ball-bearings. One drive axis uses 2 rulers; one ruler is laid flat and controls the horizontal positioning, the other is laid on edge and holds the printhead at the correct height at all times. Each ruler is supported by 2 flanged bearings along one top edge, spaced about 30mm apart with the flange on the far side of the edge, and two more similarly placed with flange on the near edge. In the middle of the opposite edge are 2 more flanged bearings. So the flanges stop the ruler flexing sideways & falling out. The two bearings on their own are mounted on a sprung slider, this eliminates backlash in this part of the drive.

 

The greatest forces will be on the 2 flat rulers, as these do the X & Y acceleration, and this acceleration must not be enough to overcome the tension springs.  I've measured my X & Y tension springs at about 250 Grams weight, call it 2.4 Newtons. The rulers on-edge simply take the weight of the printhead, so these won't really need backlash removal, and these tension springs are 150 gr weight so about 1.4 N. The force needed to push the flat ruler through the bearing-clamps is tiny, about 10 gr weight, or 0.1N . The force needed for the on-edge rulers is about 1/2 this, so 0.05N.

 

The mass of the print head so far is 63 gr, this is the plastic shown in pics below plus all 24 bearings. The 4 rulers weigh 70 gr, they're original length of 330 mm, and will be cut down to about 270mm so should reduce to nearer 60 gr finally. Rulers are 13mm wide x 0.6mm thick, very flexy!

 

Given a max tension of 1.4N as we accelerate the printhead, one axis , say the X axis, has to accelerate 2 rulers  printhead, so about 100 gr so far. If that was the total final mass, 1.4N could accelerate the 100 gr at 14 m/sec**2, or 14000 mm/sec**2. I'm currently running Cura at 750 mm/sec**2, so there's plenty of tension and I won't be in danger of overcoming the tensioners, even when the heater block & tube & fan & support plates etc are fitted. I might even be able to fit low tension springs to the X & Y flat rulers and reduce the friction by another 5 gr weight per axis!

 

Here's the rigidity & mass calculation to justify this design as potentially "better" than 6mm circular rods: (I don't yet know if this works!)

 

Stiffness (I) of 6mm dia Rod is Pi * (r**4) /4 = 3.14 * (3**4) /4 = 63.6

Stiffness (I) of 13 * 0.58 mm Bar is b * (h**3)/12 = 0.58 * (13**3)/12 = 106  so it's nearly twice as stiff as the rod (but only in 1 plane).

Mass of Rod is proportional to X-sectional area = Pi * (r**2) = 28.27

Mass of Bar is proportional to X-sectional area = b * h = 0.58 * 13 = 7.54

but I need 2 Bars to replace 1 Rod, so total Rod X-sectional area = 7.54 * 2 = 15.1

 

So my stiffness increases from 64 to 106, nearly double, and my mass reduces from 28 to 15, almost half.

There's some added weight from the bearings, washers and retaining screws, but these are small, 3mm i.d. x 7mm o.d. flanged FR683ZZ, 6 per ruler. I save weight by removing the 6mm linear bearings in the original head, so I think bearing masses will cancel out.

 

Cost looks cheap so far. Plastic bits printed in ABS on my UMO. Springs from my recycled scrap bin. Bearings cost me £13 for 30 on eBay. 4 Rulers from kinexmeasuring.com cost me £15 which included shipping &  £0.37 bank cost converting £ to Euros. So under £30.

 

Problems so far are slight.

1: The rulers have engraved markings, so I've used a very fine Arkansas oilstone (my ArkanStone! )to remove any burrs and get the marked face really smooth & polished. The bearings are from China, and have some tiny remaining swarf sitting in the grooves of the seals, so these need cleaning out.

2: The bearings have a groove where the flange meets the outer race, and this is wide enough for the edge of the thin ruler to fall into. So I've made 0.6mm thick washers to fill this groove.

3: The inner race on these bearings doesn't stick out further than the flange, so either you need to add small spacer washers, or you can do as I did which is machine 0.2mm off the 0.8mm thick flange to provide clearance to stop rubbing & friction. Only 13 of my 26 bearings needed this doing.

4: Each bearing has at least one 3mm i.d. washer transferring the clamping force to the plastic, and these are stamped out 0.5mm thick, so are very slightly dished. I've been careful to assemble these with the sharp, burred edge against the plastic, and they do not appear to rub against the outer race when put this way.

 

Now for some pics! I'm finding it very hard to put these where I want them in the text, and can't see how to rearrange them in an edit, so apologies for the mess following.
 

Two design schematics showing the basic idea.

 

Flanged bearing FR683ZZ as supplied.

 

Bearing with 0.2mm skimmed off outer race. Plastic washer fitted.

 

Original bearing is 3mm wide. Flange is 0.8mm wide.

 

Ruler edge falling into the outer race groove. You can also see the raised surface of the ruler where the 1mm markings are.

 

0.6mm washer fitted to hide groove.

 

0.6mm washer fitted to hide the groove. This bearing has been skimmed.

 

Slider parts before assembly, top & underside views

 

Slider parts fitted together, top & underside views

 

Schematic view of how bearings & rulers will go...

 

... final assembly. Some washers later got trimmed a bit more as too large outer diameter.

 

Reverse side of slider. This is X or Y axis, stronger springs out of my recycle store. You can see a small snake-shaped S-piece forming a loop for the spring. This has half a halved-torus to reinforce the contact point and make it easy to print flat with no supports.

 

Front side. Bearing screws at the bottom are loose.

 

Edeg on view shows where ruler will go. Note thinner flanges where flange is close to plastic.

 

Smaller of the 2 ABS printed chassis parts.

 

Final assembly without rulers in. The item just above fits at the left side, dropping downwards.

 

With rulers fitted. You can just see a tiny 1/2 mm gap opening up where the flat ruler in the middle at the bottom has gone through the slider.

 

Edited February 5, 2018 by andywalter
typos and add comments.

[gr5 2,172]

2 hours ago, andywalter said:

So I've made my own system, using 4 cheap-but-accurately-made steel rulers,

 wow!

 love this!

[andywalter 7]

Construction of the first Extruder Carriage is complete, I've made the heat-block down-pipe, brackets, heater block, and a couple of semicircular grippers which will hold the bowden tube firmly to the top of the downpipe in such a way that I can unscrew it easily from thare and not have to pull it out from the extruder-end of the tube. I've made the 4 sets of bearing-blocks & clamps which hold the helical plastic nuts & the 8mm linear bearings. So here are some pics!

 

Front view showing the Bowden-tube gripper pieces.

 

The two semicircular grippers fit inside the thin-walled steel cup, and that lot is held down onto the heater-block downpipe.

 

With bowden tube pieces screwed onto downpipe

 

View from above, gripper pieces sitting inside aluminium screw-cap

 

Finally! First sight of the whole thing assembled. Everything moves freely and feels fine so far.

The outer plastic pieces locate the helical nuts, the smaller pieces just inside sit on the standard 8mm rails. This assembly weighs 317 grams.

Edited February 25, 2018 by andywalter
rearrange pics and add text.