LISA Simpson

Another interesting variation from the prolific brain of Nicholas. For additional stability and appearance enhancement here is my contribution
to this design progression.

I envision the corner post to be constructed from lightweight composite construction. I could put together simple infusion mold
and pop out several sets of these pieces.

the top and bottom pieces could have machined receptacles for easy and positive alignment with the composite pieces.
I think this is a simple method which would yield stiff structure with a very minimal parts counter.

@maboo - How rigid is your Rostock with strings on just two sides?

Nicholas
I know these rods are not ideal but I thought it would be okay for proof of concept bot ?

3/8 -10

@Dannydefe: 2.54mm per turn is not enough. For most steppers and drivers this will top you out at 25.4mm/s. I think the top speed should at least break 50mm/s. For $60 you can get 25.4mm per turn. I am also considering buying some lead nuts from Roton to keep me from printing them. They are only $10. [www.roton.com]

That brings the total to around $90 but that isn't much compared to all the other hardware we save on.

I need some help from you guys to figure out the home sequence. The problem is you can't just run each end into an end stop. The z height can change for a shoulder nut without its stepper turning.

Here are the options I have so far...
1) Manually home it. This can be pretty fast. Provide a crank and it would be super fast.
2) Run everything up until one hits. Back it off a safe amount. Repeat with the other two and then with the one last remaining one. At that point you go through the process again but this time you don't back off as much. After a few iterations with shrinking back off amounts you are good.
3) Just like 2 but you have it solve for the position of all the carriages after the first iteration. This would be rough to program and would be rough on the processor.

Watching the youtube video again, it seems like the bottom few inches of travel on the rods are wasted. If the steppers were a touch higher than the bed, you could increase build area for free.

That might also help with the end-stop design?

@jason.fisher: I thought about that. There are a few concerns.

1) It makes the stepper mount more complicated. Remember that you have to tension the rod between two bearings to isolate from stepper end play.
2) The steppers would intersect the build volume.
3) If you wanted to make a heated chamber you would probably want to keep the steppers out.

All of that can be designed around. However, the rod is relatively cheap so wasting a few inches for form isn't too bad.

Question: how would changing the stepper position change the homing sequence?

I was thinking that if you had the entire rod length to travel without interference, you could use some kind of conductive tape on the resting point of each rod end and run through a movement pattern to determine the length of the rods. If everything were level, you would only need to do this from the perspective of one of the rods .. I suppose where the stepper lies doesn't really matter, as long as the the distance from the top of the bed to the end of the stepper can be made consistent.

I picture a corkscrewing hula-hoop-type motion in the tool head that lets you iterate through the arms to raise them in sequence and find top, then drop back (quickly) to just above where the algorithm began when it started the search for top (run motion algorithm in reverse?), then continuing down to find bottom. You might have to configure the specifications of your threaded rod and arm lengths for the algorithm to work? But on your reference machine, with threaded rod X and the standard arm lengths, I think we could find a repeatable motion that raised/lowered each rod's arm by X on each swing/pass?

In my raised-stepper thought process, I was visualizing that you could reliably hit the bottom of a thread with two rod arms sequentially while dropping, and stop right before or as the third rod arm in the pattern hits bottom. They wouldn't hit bottom simultaneously, but I think the right motion could cause them to hit sequentially, and that you probably wouldn't actually crash into the bed until one or two more iterations of the pattern after you find bottom from first contact.

Then it is a matter of the math of rotations/rod specs against the reference distance for each pattern iteration, where the math on down starts calculating when the first rod arm makes contact, calculating a travel distance when the second rod arm hits and the first lifts slightly, then using that to stop travel right before we think it would hit the third on the next pass.

If you know the thread specs, could you guess at the length of the arms based on the distance/travel time between two arms hitting top? Finding top seems safe, and math is our strong suit, so maybe we can learn something in our quest for top that lets us safely find bottom?

Edited 5 time(s). Last edit at 11/18/2013 03:39PM by jason.fisher.

Quote
Quazzer
@maboo - How rigid is your Rostock with strings on just two sides?

Well not really "rigid". t helpes to reduce the wobble of the tower, but it is still visible. I will exchange that construction and put acrylic walls to two sides and see how that is.

Here is a picture, as one can see I used a triangular base, because I really didn´t like the original solution with the additional wall.
Btw. : the movement of the Rostock does not affact the quality of the prints that much, unless they are rather flat, even if it looks dramatic. Still I am not satisfied with the setup.

Here my Rostock on my messy desktop...



Edited 1 time(s). Last edit at 11/18/2013 04:44PM by maboo.

---

Blogs:
Meine 3D Druck Abenteuer
[3dptb.blogspot.de]
FLSUN Delta Drucker für Deutschland
[flsun-deutschland.blogspot.com]
Books on 3D patents:
[goo.gl] (english)
[www.amazon.de] (deutsch)

How about a hard wood ply frame like this?
The idea with having two boards on the bottom is to give it extra rigidity by putting in boards in between the layers.

Screws + glue, nothing will move. Hard plywood has almost no flex.
The hardwood vertical pieces could probably be a third wider than in the picture.

EDIT: I'm a bit off topic, I posted in the wrong thread, sorry guys.


Edited 7 time(s). Last edit at 11/19/2013 01:16PM by TheTechnicalNoob.

I like it Nicholas,

I laughed by you designing this because you felt bad, haha. Anyway it's a much better approach due to not using ball joints. Not as printable as GUS but seems easy to assemble, which is also good. Perhaps LISA can be a better marketable product (perhaps I'm going too fast). I want to see how this design matures.

I fail to see how the arm bases can rotate. I see that the threaded rod would make them shift Z position, but they also need to free turn on their own Z axis, how do you achieve that?

Quote
Guizmo
I fail to see how the arm bases can rotate. I see that the threaded rod would make them shift Z position, but they also need to free turn on their own Z axis, how do you achieve that?

It looks like the arm rotation is achieved by simply having the tracking nuts rotate on the threaded rod. This should be effective, but the Z-shift associated with the XY rotation needs to be accounted for in the kinematics.

Additionally, without the supports, what would prevent the top plate from twisting when all three rods are rotated in the same direction?

Regarding the motor mounts, I would say that the most elegant solution is to make a set of leveling legs that have a motor sized opening in them, or take the pvc piping and extend it to below the base to cover the motors. Having the motors hidden inside the PVC would allow you to hide almost all the wiring, leaving you with only a power cord, and a data cord showing (if you were to hide the controller under the bed). Here is a mockup of what I am talking about in case it isn't clear.

Edited 1 time(s). Last edit at 11/19/2013 12:51PM by pjoyce42.

@pjoyce42: Without support you would need a pretty hefty screw to square up the design. I might try without supports on a miniLISA but PVC is too cheap not to use.

I really like the PVC extending below to be feet. I am just playing around with a few details before I post the mockup. I switch between wanting all the electronics up top or all the electronics on the bottom. I think for weight reasons you have to keep as much as possible on the bottom.

The design with the PVC is about 20" at its widest and LISA is about 2' tall. That is a slightly bigger footprint than Wally or GUS Simpson but it is still easy to fit on a desk.

I like the thought of the PVC around the screw but that is problematic from the standpoint of the swinging arms. I thought that I could make the slot big enough but it is almost impossible with the way I am going to allow the hub to side step the screws.

Quote
pjoyce42

Quote
Guizmo
I fail to see how the arm bases can rotate. I see that the threaded rod would make them shift Z position, but they also need to free turn on their own Z axis, how do you achieve that?

It looks like the arm rotation is achieved by simply having the tracking nuts rotate on the threaded rod. This should be effective, but the Z-shift associated with the XY rotation needs to be accounted for in the kinematics.

Additionally, without the supports, what would prevent the top plate from twisting when all three rods are rotated in the same direction?

Regarding the motor mounts, I would say that the most elegant solution is to make a set of leveling legs that have a motor sized opening in them, or take the pvc piping and extend it to below the base to cover the motors. Having the motors hidden inside the PVC would allow you to hide almost all the wiring, leaving you with only a power cord, and a data cord showing (if you were to hide the controller under the bed). Here is a mockup of what I am talking about in case it isn't clear.

I needed to be more clear. I agree with you, when you rotate there would be a small shift in the z position. I bet Nicholas has already considered that.

okay I'm in, I'm going to attempt building a LISA Simpson. Here's what I ordered today from Roton (www.roton.com)
3x Flange Nut, 3/8 X 1.000 Right Hand Torqspline Part Number: 92054 Price: $18.55 \ea.
6 ' Screw, 3/8 X 1.000 Right Hand Torqspline® Part Number: 60998 Price: $11.87 \ft.

I'm not quite sure what the final structure will look like, but my primary concern is rigidity.
As far as software modification and calibration go, I'm positive I'll need some help on this end.

I just talked it over with my robotics class. They have 3 weeks to spare and I think this will be a quick build so they are taking it on. For consistency, I am ordering the same stuff Dannydefe ordered.

I contacted Roton about the backlash of the flange nut and they say it is 50-150 microns. That was to be expected. We have a few things going for us.
1) I will set the acceleration to .25g-.5g. That combined with a low jerk value should keep the nuts from taking off as we change directions.
2) There will be a slight side load on the shoulder from the offset arm attachment. See drawing. This will effectively decrease the backlash.


There are some antibacklash nuts but they aren't designed to be loaded like we are going to load them. I would have to get another follower nut so I can put a moment on the shoulder. I think I would try to print my own before I spend all that money.

Motivation for not printing the nuts in the first place: I try to design so that a printer doesn't have to be too precise. A printed nut would be one of those gotcha parts that could hold up a build. If for $60 we can get some high quality nuts that we don't have to mess with then I say go for it. I certainly will experiment with printed nuts. I am going to have to make printed thread clamps which are threaded on the inside. It won't be too much of a jump to print off whole nuts to see how good they are compared to purchased ones.


Edited 1 time(s). Last edit at 11/20/2013 11:04AM by nicholas.seward.

I hope this has not been discussed before: I just skimmed the thread.

On both my Solidoodles, and especially on my Trinity Labs Aluminatus, over-constraining the Z threaded rods ended up in wobble in the X and Y axes. Or at least that was alleged by various owners - I never personally had serious problems. At the moment, the Aluminatus has the steppers at the bottom, with the X carriage resting on the support nuts. No bearings at the bottom, and the top is free to flop about. This seems to produce better results than attemping to constrain the rod in three points (lower bearing, upper bearing, and moving nut).

I would advocate that the design not use the threaded rods for support, and allow for the possibility that the top of the rod is allowed to flop around. This might then require steel guide rods (which are present on both of my printers). More complex, but also more flexible should you encounter some X/Y variations.

@jon_bondy: The current plan is to tension the rod between the top and the bottom plates by adding PVC compression columns or something else similar. The tension will be used to pull the rod straight. Note: the arrangement in the video will more than likely lead to extreme wobble. To make that work the screws would have to be very large. (probably too big for the motors to quickly accelerate them angularly.)

Also, without smooth rods we will not be have the luxury of leaving one end of the screw free. With steel rods the design will get much more complex. The carriage has to rotate.

Neat design.

So how many tons of tension will be necessary to straighten the leadscrews?
Are there thrust bearings or taper bearings at top and bottom?

This requires tension or very straight leadscrews?

Do they tension leadscrews at the factory to straighten them?

I don't understand how the nuts and fixtures on the leadscrew are prevented from rotation?
Does the geometry of the center spider make it manageable?

Quote
cozmicray
Neat design.

So how many tons of tension will be necessary to straighten the leadscrews?
Are there thrust bearings or taper bearings at top and bottom?

This requires tension or very straight leadscrews?

Do they tension leadscrews at the factory to straighten them?

I don't understand how the nuts and fixtures on the leadscrew are prevented from rotation?
Does the geometry of the center spider make it manageable?

You are right about requiring a lot of tension. In fact, it requires infinite tension to theoretically straighten a curved rod. I am going to shoot for something a lot less than that. The R6 bearings I am using will support around 75lbf of axial load. In reality, I will tighten until the screw is noticeably harder to turn. Bottom line: the screw better be straight to begin with or this won't really work. The tension in the screw will only reduce any bends by a fraction. I have currently decided against thrust bearings or taper bearings for the time being. If 75lbf isn't enough then this design probably needs to be scrapped.

I am not sure how they manufacture the lead screws?

Theoretically, the youtube animation would be self-supporting. The nuts on each side of the plates will square the design. If all the parts are perfectly rigid then no other supports would be needed. I think in reality this would only work for large screws. (larger screws then we want to turn with NEMA17 steppers) However, I am planning on using PVC pressure columns that will more than square the design.

I would expect the load on the arms segment to eliminate any possible backlash.

---

Blogs:
Meine 3D Druck Abenteuer
[3dptb.blogspot.de]
FLSUN Delta Drucker für Deutschland
[flsun-deutschland.blogspot.com]
Books on 3D patents:
[goo.gl] (english)
[www.amazon.de] (deutsch)

How will the leadscrews be tensioned?
Leadscrew nut top and bottom tightened between bearings in top and bottom platforms?
How will they be prevented from unscrewing?

Screws with this type of lead -- not made for tensioning?

@cozmicray: You are very perceptive. I plan on using printed thread clamps. I plan on tightening the clamps until I can just barely turn them. At that point, I will tension and then finish tightening the clamps. The jury is out on whether this will work. My gut says this will work.

Nick what printer are you using to make all your parts?

What kind of fill density, what polymer will you use on printed thread clamps?
Might this require nylon or polycarb?

Have you thought about heating up a blank thread clamp (No threads) -- then clamping around leadscrew.

You could heat up threaded rod then clamp (even with external clamp right at center)? Is this against the repRap rules?

Quote
cozmicray
Nick what printer are you using to make all your parts?

What kind of fill density, what polymer will you use on printed thread clamps?
Might this require nylon or polycarb?

Have you thought about heating up a blank thread clamp (No threads) -- then clamping around leadscrew.

You could heat up threaded rod then clamp (even with external clamp right at center)? Is this against the repRap rules?

I use a Printrbot Plus. I hope I can get Simpson and Wally dialed in enough that I can use them also but for now it is just the Printrbot Plus. Also, I have a QU-BD Revolution XL. It is amazing but there is some problem with the thermistor I can't track down. I am going to take it back to QU-BD today for them to take a look.

I will probably do 75-100% infill for the printed thread clamps. I only print with PLA. That is just a personal choice. Since, I work around kids I want the printers to smell good at least.

I thought about doing some heating tricks like you mentioned. There are no rules against that. However, if you can print and then use without post processing then you have a design that others can more easily duplicate.

Originally for the lead nut I was going to print a nut that was just slightly undersized. I was then going to heat up the nut and the rod and thread it on. I will probably still try to do something like this. If it works then other can use the technique. If it doesn't, I have the purchased nuts.

Well, it is official. I purchased all the parts I didn't already have. I am matching what Dannydefe ordered so we can closely collaborate. He has some equipment that I don't have so I expect to see some cool composite bot from him. I on the other hand am going with PVC pressure columns.

I am handing my drawings over to my students. They are going to do revise, print, and mill the necessary parts. I am also going to have them log the whole build. I am still deciding on what is a realistic level of documentation I can ask of them. They are all very talented and gifted but they have what I consider more than a full-time college load. (This is a residential high school for gifted students.)

***WARNING: I AM ABOUT TO BRAG ON MY KIDS.***

If you guys think I am crazy for giving this project to high school students who only have 3 weeks to complete the project then check out this awesome cardboard robot. They decided on their own to restrict the arm to cardboard as a fun design challenge. The cardboard robot was designed and built in 6 weeks while I essentially ignored them and did work on Simpson and Wally. :-) (That is actually underselling it. They built a prototype with a jigsaw over a 3 week period. Redesigned it. Laser cut that version. Found more problems. Completely rebuilt the robot again from the ground up in a few day.)

I walked them through the inverse kinematics. I explained that you essentially just have to find the z value of the top intersection of a circle (with the radius of the arm) with a vertical line that is the length of the arm's xy projection away from the center of the circle. Instead of saying "huh" these guys immediately point out that we also have to account for the rotation/translation coupling of the arm.

I just thought I would share a quick photo of the Roton Torqspline rods and nuts. They look pretty good although the nuts have a small amount of play in them.
I think I might just do some rough experiments and see if I can't cast some nuts in epoxy after waxing the shaft .

Pic looks great!



You could use the epoxy nut to take up the backlash of the steel nut.
I'm not sure how I to go about maintaining a relationship between the two?

Add graphite to the epoxy to reduce the friction.
West System sells it, I've used it, and it works great.

Straight epoxy would have too much friction.
Ideally you want the contact surface to be mostly graphite.
You could mix two batches, the first one you go overboard with the graphite.
The second coating is just a backup.

You could even make split epoxy nuts, such that you can adjust the grip, and clearance.
If you haven't cast over a threaded rod, conduct an experiment.
The type, and amount of mold release will dictate how easy/hard it will be to remove the cast nut.

WEST SYSTEM
#423 Graphite Powder - 12 OZ
423 Graphite Powder is a fine black powder that can be mixed with WEST SYSTEM epoxy to
produce a low-friction exterior coating with increased scuff resistance and durability.
Epoxy/graphite is commonly used as a bearing surface,...
[www.westmarine.com]

Edited 1 time(s). Last edit at 11/24/2013 02:57AM by A2.

Seems to me you could create a printable tube section which could be somehow stacked to allow you to accommodate whatever length of acme rod you chose to purchase? Heck, might even be possible to put the motors inside the bottom mount of each tube with a fan on top of each one for airflow. would certainly make for an aesthetic printer hmm.. might obstruct the bottom of the print envelope now that I think about that though. nm.