Ideas for the next Wally.

Re: Ideas for the next Wally.

Postby realthor » Tue Apr 14, 2015 4:39 pm

yes, quite a redesign ... I am trying to breed a Wally with a Morgan (maybe 1-arm scara) and a bed that doesn't need rails/linear rods/linear bearings. This might even be called something else ... but attaching to a wall still seems a great idea to me. I also explore concepts at this stage as I am far from any means to actually build one.
Unfortunately I don't have the necessary skills to fix the existing wally beside suggesting the platform that Nicholas proposed.
reprap Lander concept on Concept Forge
reprap Lander concept on RepRap Forums
realthor
 
Posts: 83
Joined: Tue Mar 31, 2015 7:45 am

Re: Ideas for the next Wally.

Postby Nicholas Seward » Tue Apr 14, 2015 5:46 pm

@realthor:

Case 1: both motors on board with a big redirect pulley.
Case 2: one static motor concentric with shoulder or a small redirect pulley.
Case 3: one motor on brachium

Case 1 uses existing firmware. Case 3 would be the easiest to build. I will see if I can write down the forward and inverse kinematics. I want to do an analysis to see which one is the simplest to put in firmware and which is the most resilent to calibration error. I suspect that case 1 and 3 are both the simplest mathwise and case 2 is the most susceptible to calibration error.

Bowden tubes require very gradual bends. As drawn, you would generate too much friction in your tube. Going under the shoulder is possible but I tend to think that all the extra effort is for a small gain. If you prevent the effector from going within 75mm of the shoulder then you only throw away 6.25% of the build area. I also would suggest avoiding the joint for routing. Put a hole through the forearm pull close to the joint and you will get a very similar effect for much cheaper.

Note: Bending the arm in only 1 direction only throws away 25% of the work area.

If the arm support is above the bed then and force on the bed will force it into the wall with no rail needed.

Feel free to make another thread here or anywhere else(reprap). I love it here because it is very relevant to Wally. Let me know if you go so where else. I love talking about this stuff.
Nicholas Seward
 
Posts: 738
Joined: Mon Nov 25, 2013 10:41 pm

Re: Ideas for the next Wally.

Postby realthor » Tue Apr 14, 2015 10:51 pm

@Nicholas: I am happy with this thread like it is, it is Wally that started my own take based on my own needs/situation.

1) I guess this latest design (with 3 pulleys) falls within your case no 1. I am quite happy with using existing FW, no need to reinvent the wheel at this point.

2) What tool do you use to calculate lost surface area and the like? If only that of a small percent is lost than yes, going under the shoulder makes no sense. I thought that maybe full revolutions would be possible but that would twist and stress the filament and the PFTE tube to a breaking point.

3) Is the positioning of the extruders, placed on top of the motors "shelf", ok? It seems to me again that they are bending the filament a little too much at the intake. But in this configuration it seems the perfect placing to keep a small footprint.

As far as I understand, keeping the two arms on one side while printing makes it basically a Morgan without the guide arms right? The advantage over the Morgan is that I can flip the elbow to the other side and continue with another print, and maybe some less material (?). In this case the Morgan FW should work and only little adjustments need to be made, like how to home to one side and then when a print is done home it to the other side (ready for next print).

screenshot.118.png
screenshot.118.png (98.44 KiB) Viewed 15104 times


I will follow up with a study of bed sizes/working area and advantages/disadvantages for different needs.
reprap Lander concept on Concept Forge
reprap Lander concept on RepRap Forums
realthor
 
Posts: 83
Joined: Tue Mar 31, 2015 7:45 am

Re: Ideas for the next Wally.

Postby Nicholas Seward » Tue Apr 14, 2015 11:07 pm

@realthor

2) I just did the area lost divided by total area. (pi*75^2)/(pi*300^2)~6% It isn't exact but it is pretty close. The real number may be less if you don't allow flip flopping of the elbow or more if you do. Many CAD programs such as AutoCAD can do area calculations but a rough figure is good for this.

3) Don't worry about filament bending as much as bowden tube bending.

Yep, this is basically a Morgan. (The reason you can't do a real Morgan for this is that the elbow will hit the wall and you will have very little print area left.)
Nicholas Seward
 
Posts: 738
Joined: Mon Nov 25, 2013 10:41 pm

Re: Ideas for the next Wally.

Postby realthor » Wed Apr 15, 2015 12:15 am

I wanted to put this in another post because will be quite large. This is a study of bed sizes/working areas and advantages/disadvantages for different needs. I wanted to have a visual of all possibilities that I see in using my design. Could be totally flawed though and that's why I am putting it here for a critical eye.

screenshot.122.png
screenshot.122.png (111.36 KiB) Viewed 15104 times


Specifics: a 3-nozzle extruder has been considered, causing the clearance to be specific to this case (but I guess it would be proportionately smaller with one nozzle and also it doesn't count much in the grand picture); this clearance is to allow all nozzles to be able to get to the perimeter of the work area - and causing the outer nozzles to leave the perimeter.

Case 1)
Advantages:
* The largest working area for a single print.
Disadvantages:
* Needs new code to allow for elbow flip
* Will not be precise with last position
* Low precision due to distance from wall, length of the lifting mechanism arms
* Needs clearance on the sides if a multiple extruder system is used (like in this example)

Case 2)
Advantages:
* 80+ percent more working area than the case1;
* Can start a second print as soon as the first one is done (needs a homing routine that homes to the other side so slight modifications of Morgan FW);
* Can have a hot-bed on one side and a unheated bed for the other print area at a lower cost than in case1;
* Does not need increased platform size for the clearance from the lifting mechanics;
Disadvantages:
* More than double the bed size for only 80% increase in work area;
* Work Area per print is smaller than case1 (probably around 85% - eyeballing);
* Needs larger wall support frame (about 80% wider)
* Needs beefier lifting arms to support all the added weight

Case 3)
Advantages:
* Most stable (precise) as it is closest to the wall, has the shortest lifting arms;
* Least expensive from the hot-bed pov but also the lifting mechanism
* Wall support is only ~10% larger than case1 and ~25% smaller than case2);
Disadvantages:
* Small work area but that depends on what is needed (less than 50% of the case1 working area and less than 33% of the case2 working area)
* The XY is quite large for such a small printing area and, considering that the price of the XY is many times higher than the platform lift/heated bed/etc, it could make no sense to not build it bigger (assuming the precision loss is not increasing too much)

What would you choose? Do you have to add anything else or maybe you don't see this useful ? I am eager to see what you think about it.
reprap Lander concept on Concept Forge
reprap Lander concept on RepRap Forums
realthor
 
Posts: 83
Joined: Tue Mar 31, 2015 7:45 am

Re: Ideas for the next Wally.

Postby Cozmicray » Wed Apr 15, 2015 1:59 am

WHY?

Why the crazy design?
Perhaps you should print on a sphere just to make it harder?
What's wrong with X, Y machine? With a linear Z axis

Perhaps someone should work on a slicer to produce polar g-code
so you don't need a Cray computer to do the math to control the machine?

Yes square wheels -- much better --- go back to cave -- paint on wall!

:?:
Cozmicray
 
Posts: 126
Joined: Tue Dec 03, 2013 10:23 pm

Re: Ideas for the next Wally.

Postby realthor » Wed Apr 15, 2015 2:30 am

@Cozmicray: because exploring concepts is free, because i need this to understand the factors that influence a good result, because going crazy without much self censoring is creativity food, because going straight to difficult makes much more sense to me than starting easy.

Morgan, which this design takes almost everything from, is one of the cheapest and easiest to build. I happen to dislike linear rods and linear bearings for obvious reasons. The double parallelogram platform avoids the maths that made Wally a too difficult project.

An XY machine (Cartesian I assume) I would only consider for CNC, if I'm not proven wrong that a Delta or SCARA can't do it better.

No need for sarcasm, if everybody would have stayed with XY and Linear Z there would be no Wally, GUS Simpson, DLP, Morgan especially and so on. To each his own, all should be respected for bringing variety and not designing patented stuff.
reprap Lander concept on Concept Forge
reprap Lander concept on RepRap Forums
realthor
 
Posts: 83
Joined: Tue Mar 31, 2015 7:45 am

Re: Ideas for the next Wally.

Postby Nicholas Seward » Wed Apr 15, 2015 3:14 am

warp.gif
warp.gif (821.29 KiB) Viewed 15100 times

I did the inverse and forward kinematics for all the cases I listed earlier. The truth is they are all about the same in terms of math complexity and the effect of miscalibration. The GIF above shows the effect of miscalibration from 95%-105% of the ideal calibration.
Nicholas Seward
 
Posts: 738
Joined: Mon Nov 25, 2013 10:41 pm

Re: Ideas for the next Wally.

Postby Nicholas Seward » Wed Apr 15, 2015 3:23 am

Here are the kinematics for all the cases. They are all pretty simple and as I said in the last post the effects of miscalibration all look about the same.

Variable Definitions
L=length of arm
ma=mechanical advantage (This is what could be miscalibrated if you drive the machine with string.)
(r,theta)=polar coordinates
(a,b)=stepper positions

Inverse Kinematics
Case 1:
a=(theta-acos(r/(2*L)))*ma;
b=(theta+acos(r/(2*L)))*ma;

Case 2:
a=(theta-acos(r/(2*L)))*ma;
b=theta+(2*ma-1)*acos(r/(2*L));

Case 3:
a=(theta-acos(r/(2*L)))*ma;
b=2*ma*acos(r/(2*L));

Forward Kinematics (Needed for end stop detection.)
Case 1:
r=2*L*cos((b-a)/(2*ma));
theta=(a+b)/(2*ma);

Case 2:
r=2*L*cos((b*ma-a)/(2*ma*ma));
theta=( 2*a*ma-a+b*ma)/(2*ma*ma);

Case 3:
r=2*L*Math.cos(b/(2*ma));
theta=(b+2*a)/(2*ma);
Nicholas Seward
 
Posts: 738
Joined: Mon Nov 25, 2013 10:41 pm

Re: Ideas for the next Wally.

Postby Nicholas Seward » Wed Apr 15, 2015 3:37 am

@realthor: I think I would go for a full half circle build area. There is no need to make it square. You might want square sides for the Z arms to attach to so maybe just trim a little off each side of the half circle. Check out the Morgan Pro new beds.

Don't worry about resolutions at full arm extension. That is a straight forward design problem.

Pick the build area you want. -> Calculate smallest arms that will give you that build area. -> Calculate the mechanical advantage you need to get the resolution you want.
Nicholas Seward
 
Posts: 738
Joined: Mon Nov 25, 2013 10:41 pm

PreviousNext

Return to Wally

Who is online

Users browsing this forum: No registered users and 1 guest

cron