meet Lander, the multi-configuration 3d printer

[realthor]

Might the winch line get caught up in the gear teeth and not spool nicely?

I can't really tell that until there is a prototype tested in the wild. Z is moving so slowly I guess the line is not going to get caught in all those teeth if it is tight enough. The weight of the bot will keep it tight enough with this design.

As for your KISS variation, what is your solution to avoid triangulation? I've been battling this since the beginning and that is how I ended up with the spured-bolt solution. Are you thinking to compensate for that in software?

I have a solution similar to yours for the actuation of the SCARA Arms but that requires the drums/pulleys of the SCARA to be designed in such a way as to avoid triangulation. It has been discussed in the "Ideas for next wally thread":

screenshot.153.png (13 KiB) Viewed 69107 times

With strings, triangulation is a bitch and there are few ways around it. No simple solution will do. Unless one is smart enough to compensate in software, but that has its own challenges to accurately predict what happens mechanically.

[realthor]

I've been trying quite a lot to come with a worm gear assembly that would engage a larger surface of contact but it seems to me that a worm gear (the driven gear) with fewer teeth that are taller than the norm (to allow the spooling thread) is not helping.

I might be doing it wrong to have an identical tooth profile on the worm as I have on the worm gear but with this approach the point of contact between these two is effectively on only 2 radial teeth (first half of the image below). There is no contact on more than one tooth on the axial plane whatsoever (you can see it better in the second half of the image below):

screenshot.225.png (263.49 KiB) Viewed 69106 times

And this is the end result with an 8deg tilt in the worm.I can't do more degrees unless the profile at the worm is thinner than the one on the worm gear. Is this how they design worms? And how thinner of a plastic profile can take the stress?

I am starting to think that worms are working as expected when there is a high number of smaller teeth that it can engage on the worm gear, so the total contact surface adds up to a sufficient amount. I will have to experiment in CAD to see how can I accommodate the spooled string with smaller teeth.

If you have experience with designing wormgears please weigh in.

[Nicholas Seward]

Wow. There is a lot here that I haven't had a chance to respond to. I will start by saying that your worm has the wrong profile for optimal contact. The worm is essentially a rack with a twist. The worm will look like an ACME screw.

You are correct that there will be minimal contact. With this design it can only have a point contact. That is from math and can't be avoided. Normal worm setups allow the spur to arc around the worm but that can't happen here. I wouldn't worry about it until you test it in real life.

I personally would go with a larger diameter worm. The bigger it is the more the system will behave like the rack and pinion system we are trying to emulate.

[realthor]

Yayyy!!! Nic is back . I can't even remember what questions are there left behind but feel free to answer, I've been banging my head around without your constant advice. Don't do that again:P ... kidding

I personally would go with a larger diameter worm. The bigger it is the more the system will behave like the rack and pinion system we are trying to emulate.

I tried several variations of worm diameter and, if the thread profile is designed in such a way as to mesh with the worm gear's teeth (proper size), then the diameter of the worm seems to only be relevant to its strength. Maybe some adjacent axial teeth can be engaged when tilted... is this the advantage of larger diameter?

I'll try some ACME for the worm. What about the worm gear? It is a normal spur profile like the one I have here right? I've got this profile from Me-Bac.com .

Edit: here's an 29deg ACME profile for the worm (25mm root diam for both worm and worm gear). This allows about 5deg tilt of the worm. I don't see much of a difference, in section at least. It seems to be a point (more likely line contact) in both cases. The trapezoidal ACME seems to allow beefier thread.

What should I see improved with ACME compared to the previous involute profile?

screenshot.228.png (160.06 KiB) Viewed 69105 times

Regards.

[realthor]

A little update: I started designing in CAD since Inkscape can't help past a certain stage and I want to learn at least one CAD tool. I am currently using DS Mechanical as it is free but plan to start on ProE at some point.

I am back to square 1 with the SCARA actuation and will try to come up with a solution for the triangulation issue before I decide it is too bulky and go back to the extending couplers idea. So far I know that the drums/pulleys have to be quite tall to accommodate for the distance the string coils move on the M8 bolts.

I imagine that for every mm that the string coil travels up or down the bolt the same amount must be achieved on the big drums so that triangulation is kept to a minimum. I can't yet imagine how many coils are needed on the drums and the actual calculation is based on the assumption that there are going to be about 6-7 coils on the bolt.

Question: are we using the number of coils as a way to rotate the drums via friction?

screenshot.234.png (23 KiB) Viewed 69079 times

Another area that is giving me headaches (which one doesn't :-s) is the DLCJs. As long as the ones presented by Nic on the wall-clamped Wally video are nowhere to be found, I am sure I am over-complicating the design. Will break the design up in the following posts.

[Cozmicray]

Since the KISS mechanism is rotating and moving back and forth,
the winch line should be in the vertical all the time
with the right rotation and thread sync.

I just separated the two functions of your spur worm thread?

I guess I don't understand where the triangles are?

Might the winch line get caught up in the gear teeth and not spool nicely?

As for your KISS variation, what is your solution to avoid triangulation? I've been battling this since the beginning and that is how I ended up with the spured-bolt solution. Are you thinking to compensate for that in software?

[realthor]

In winch terminology it is called Fleet Angle and is the angle the winch line makes with the spool. At each turn this angle increases until the spool is full of coils, when the line starts creating a second coil over the first one and so on. At each change in angle, the distance on Z is a trigonometric function and must be calibrated from software so that the printer knows how many stepper steps must turn the z motor to increase z with the desired resolution.

screenshot.235.png (157.37 KiB) Viewed 69074 times

The only other way to avoid the change in rope length is a tapered spool drum, which I don't like either:) .

Edit:
@Cozmicray: I just now realized what you meant with the sketch. Yes you are right, you are separating the functions, the same way I did by having half spur and half threaded bolt in one of my early sketches. In your case the whole motor moves by sliding on the rods.

This approach works but adds a whole bunch of undesired vitamins to the list (linear rods, linear bearings, anti-backlash nuts). And you are not listing all: the other end of the shaft should have a nut of sorts that would let the spooled string pass through or any other means of avoiding the shaft to bend and wiggle.

I started this whole project to avoid those. For some people this might just be the solution. I want a bit more rep-rap-ability and there is a chance we could print the whole Z mechanism (without motor and some nuts/washers/what have you).
Also with the printed "threaded gear" increasing Z is just a lengthier bolt away, another print, maybe with recycled plastic from this one:).

[realthor]

This was exhausting and I hope I haven't done it wrong .

I think I am starting to understand the logic of the correlation between the bolt coils travel and the drum coil(s) travel so that triangulation is kept to a minimum.

It goes like this: the coil on the bolt moves up or down with "P-pitch" mm at every bolt rotation so the coil on the drum must travel the same Z distance to keep the line between the two entities parallel at all times. Since only so much string can be accommodated by the bolt at a complete turn, the compensation on the drum is in the steeper angle of the "imaginary" groove (capstan groove?).

For example, an M8 bolt typically has a pitch of 1.25mm (coarse). The threading is a helix with a radius smaller than the 4mm, about ~3.325 actually, plus the radius of the spectra line used (ex: 0.32/2). I used the calculator at https://www.easycalculation.com/physics/classical-physics/helix.php and got a spectra helical length of ~23mm for a 3.45mm radius helix over 1.25mm of height at an angle of 60deg.

So for each rotation of the bolt the coil on the bolt travels 1.25mm on Z and winds or unwinds 23mm from the larger drum.

To keep up with this travel, the angle of the spectra line on the drum should accommodate 1.25mm in height for each 23mm, which translates to about 3deg of an angle (http://www.pagetutor.com/trigcalc/trig.html). Also to be noted that the "pitch" on the drum is about 12.5mm which equals to 10 coils.

Here's a sketch of some of the math:

screenshot.239.png (280.46 KiB) Viewed 69068 times

With 120mm drums, the Spectra helical length is about 422mm for 180deg of rotation. So the bolt will rotate 18.33 times for this 180deg swipe of the shoulder from side to side. Over these 18.33 rotations the coil will travel on the Z of the bolt 18.33*1.25=23mm. With an initial coil height of 12.5mm, it totals up to 35.5mm in Z height on the bolt, which should be the height of the shoulder drum.

@Nic: I know I got to a number we both previously agreed upon but I needed to understand the process mathematically. I hope I didn't screw up again . My sketch finally resembles yours provided in the Ideas for Next Wally thread but I didn't get it at the time.

I have two issues with this approach.
1) It's pretty bad news that 35.5mm+ will be the height of the drum, a lot more plastic than a solution with extended motor couplers but the bonus of less moving and potentially fragile parts.
2) The spectra line in the example above is a closed loop, resembling a belt and it is pretty difficult to tighten it or at least I don't know a good way to tension it because it moves up and down and no line end is fixed.

I got an idea though, which is inspired by your explanation in the other thread. If I was to split the coil and have half wound up in one direction at one end of the drum and the other half at the other side I would be left enough free space in the middle to use for the redirect line that drives the elbow pulley. I would also be able to use a tensioning mechanism that is away from the pulleys. Something like this:

screenshot.240.png (51.14 KiB) Viewed 69068 times

[realthor]

Here is a CAD attempt at the previous stringing concept. I've modeled the drums/pulleys with the proper angle grooves and tried an arrangement to see how they fit on the bot. Looks pretty ugly so far but at least some proof of concepts can be visualized.

I got rid of the separated strings for the pulley redirect (that would have needed another tensioning system and another drum section) so I am left with two off-SCARA tensioning systems that will let the drums be very simple in design.

screenshot.243.png (149.9 KiB) Viewed 69061 times

Edit: a similar concept to the shoulder drum rotation principle can be applied to the Z-winch for a backlash-free solution (gravity should take care of that but anyways, this is quite simple solution):

screenshot.255.png (104.93 KiB) Viewed 69010 times

...and a little 3D candy:

screenshot.256.png (67.68 KiB) Viewed 69007 times

[realthor]

I went back the thread and gathered all the questions that are not answered and that I would need some advice for.

Here's one last printscreen collage with my WIP (sorry about the low quality, my laptop really struggles with CAD software running):

screenshot.253.png (212.45 KiB) Viewed 69024 times

Double Parallelogram / DLCJs:

1) @Nic: why did you say that probably letting the bot roll up and down the wall is better than relying on DLCJs?
2) The DLCJs would double the number of bed platform lifting arms in Wally. Considering that the DLCJs have 2 bearings for each end and the non-DLCJ use a similar design, the number of bearings for the Z alone is 16 bearings per side, that is 32 bearings for both double parallelograms. Are so many bearings needed? Maybe the non-DLCJ arms can only have one bearing per end and be thinner (say 12mm instead of 25mm). I need some advice on this one.
3) Should the bot be wider rather than narrower to increase stability? My current design has 150mm double parallelogram arms and 150mm SCARA arms. The volume of the bot with SCARA parked to side and the Z double parallelogram retracted is 326X340X207mm.

Threaded Spur Gear:

1) @Nic: why would there appear artifacts if the worm is tilted the same way the spectra line is rolling on the bolt?
2) how is a left handed thread helping and how would that look? maybe you can show a render of that, I don't really understand and my CAD skills are 1 week old ;
3) What are your concerns about ultimate accuracy? Are you still planning to build a testing rig?
4) Considdering that the teeth on the threaded spur are quite wide, how many threads are the printed nuts going to accommodate?2-3 would get the nuts alone to be 30mm wide. If the threaded spur gear is 15cm long with 5cm in between nuts, if the nuts are 6cm altogether, then only 4cm are left for play.
5) How would you approach the tilt in the motor mount? Embed it in the printed mount? I don't really see other option and a variable tilt motor mount was nowhere to be found during my searches.

SCARA:

1) what would be a simple tensioning mechanism that I can do off the drums. I have sketched an idea of how to take the tensioning off the arms. I find this useful for a number of reasons, one being that you don't have to touch the arms. Another is that the tensioning system can be bulkier, etc.
2) Is friction alone going to be enough for the two string loops that I have sketched in the previous post or should I think of embedding some sort of waved helix to provide more friction? I've noticed that Wally doesn't rely on friction and that the string loops are clamped to the drums, but with my designI can't really clamp the strings as they move position on the drums and there's also the redirect which I now have in one single piece.

Work Area:

1) The working area of this 1-Arm-SCARA is quite large compared to its compact dimensions. It is ~1300 square cm for printing on both sides and ~1037square cm when restricting the motion in one side only. Considering that Morgan FW has the capability to support the one-side-only configuration I will use that one for now. My question here is if there is any support in firmware for other-than-square working areas. The biggest square I could print with this bot is 180X180mm but I could take advantage of the rest of the surface if rectangle at least is possible.
What about a free-form shape? Is it possible in software to define an equation that would result in the actual possible printing area?

Overall design and simplification:

As I have little to no experience with fabrication I will gladly accept design suggestions that would simplify and pretty-fy the current bulky sketch. Experience is key here and I have nothing to base my decisions upon but my imagination. Please pitch in with any suggestion.

I have also created a project thread on reprap forums to potentially get more coverage and maybe faster answers and advice to advance quicker through the design. It seems so lonely here lately, Nic must be very busy with his other projects .