Grounded Experimental Delta Printer

@terribleperson: I hope that we can adapt this to do milling operations. Annirak is correct that there would be many design consideration that have to be addressed.

I think if I was asked to make this a milling machine I would use a worm gear drive on the elbow. As a quick idea for backlash reduction, I would use two steppers with worm gears for each driven gear. This will allow for the backlash to be adjusted out of the system or minimized from time to time. (I am not saying this is the best solution but it was the first thing that popped out when I started thinking about it.)

The second thing that really concerns me is resonant vibrations. The Simpson is very light and wispy by design. Generating the forces needed is not a problem but the machine would vibrate itself to death if asked to machine metal. I think after you go through the bulk up process to mitigate these issues that you will end up with a small envelope compared to standard metal mills of the same size and weight. I really hope I am wrong and will work hard to prove myself wrong in the future. The cost saving potential is great.

About metal, until we have big simpson to play with, maybe sandmold is what we can sidestep with current design: [youtu.be]
If 3/6 arms Simpson can do accurate aluminum milling, I think it is enough to be a game changer.

Updated : There's a technique to cut very tough metal without physical cutting force called EDM, but I didn't know how it really works. [fab.cba.mit.edu] and [fab.cba.mit.edu] from [mtm.cba.mit.edu]

Or very slow and precise micromachining end mill with turbine, possible ?

Or active vibration control as used in helicopters.

Light weight design with carbon fiber foam core / Heavy weight design with epoxy granite and steel.
Too much options, too little time.

terribleperson Wrote:
-------------------------------------------------------
> Is it just me or does the Simpson setup look
> pretty good for milling? I don't know much about
> milling (or even really 3D printing), but the
> Simpson appears to be far stiffer and more robust
> than most repraps. If you scale it up, I imagine
> the weight and the shape would negate most
> vibration, making it possible to mill steel (and
> potentially harder materials), rather than the
> usual 'wood and maybe aluminum' you get out of
> most hobbyist mills.
>
> edit: Is the ability to print arms larger than the
> printer is currently using inherent to the Simpson
> design, or just to a particular implementation?

Edited 7 time(s). Last edit at 08/21/2013 10:32AM by Buytaert.

@terribleperson: I forgot to address your question about the ability to print longer arms. I have stretched and skewed the geometry of Simpson in many different ways to see if the ability to print longer arms is inherent or I just got lucky. The answer is that for any reasonable geometry that you would pick leads to being able to print longer arms. I am going to coin the word "macrocreation" to describe the process of creating a larger version of oneself. To design a Simpson there is really just one ratio that matters. That is the arm length vs distance between the shoulder pivots. I am currently using the ratio 3:5. Mathematically, you get the biggest volume as the ratio goes to infinity meaning the shoulders get closer together. However, your print volume becomes flower shaped as the area from the shoulder pivots radiating out become a no man's land. The flower shaped volumes have the potential to print much much larger arms. It just wouldn't be good for everyday use. I moved the shoulders apart until I got the bulk of the theoretical print volume to be between the shoulders. Currently, Simpson has the shoulders spaced at 250mm and the arms are 150mm. Simpson can technically reach all positions on the print bed plus quite a bit off the board. I purposely truncated the print area because the available height goes down to zero at the edges. Bottomline, any Simpson reasonably designed will happily be able to participate in macrocreation. A 3:5 Simpson can have a macrocreation factor of .67. (67% longer arms) A 3:4 Simpson has what I consider the practical max for the macrocreation factor of 1.1. (110% longer arms)

Here you can see an animation where it goes from a 3:5 Simpson to a 3:4 Simpson. I scale the width of the possible arm print to be proportional to the length.


Edited 3 time(s). Last edit at 08/20/2013 10:16PM by nicholas.seward.

@nicholas.seward: That's pretty much I was hoping for. I mean, obviously it's possible to tweak something as unclearly defined as the Simpson design until it can't macrocreate, but I was hoping that macrocreation would turn out to be the rule, rather than the exception.

I suppose once the Simpson design is semi-finalized, the thing to do would be start finding the breakpoints. For example, a Simpson past a certain size won't fit in a certain size box (therefore any larger Simpson will cost more money to ship). A Simpson past a certain size might require heavier motors, so that's another breakpoint. If you found the largest Simpson that could fit in the smallest (reasonable) box while still using the same vitamins as a larger Simpson, and then tweaked the geometry for maximum arm size, it'd be easier to ship off one of those and have people scale it up to their desired size, rather than shipping off a large-scale Simpson. You could even provide designs for 'max arm' Simpsons as opposed to normal manufacturing Simpsons, so someone who wanted a very large Simpson could take the shortest path and only make the last one standard.

@terribleperson I think we should begin with smaller one, find a suitable design for first generations. We only need a few generations for 3:5 Simpson to reach 1 meter arm length.

To give rough idea how Simpson do macrocreation :
For 3:5 Simpson (cm)
1st-15 2nd-25 3rd-41 4th-69 5th-116 6th-194 7th-324 8th-541 9th-903 10th-1509
For 3:4 Simpson (cm)
1st-15 2nd-16.5 3rd-18.1 4th-19.96 5th-21.96 6th-24.15 7th-26.5 8th-29.2 9th-32.1 10th-35.3

These results are very different, so I think focus on design is better than production for now.

22 Simpson print beds. 3/8" water jet cut basalt. (For those mathematical nerds out there, this is a constant width shape.) The macrocreation factor will be limited to 50% with this print bed.

Edited 3 time(s). Last edit at 08/21/2013 11:16AM by nicholas.seward.

@nicholas.seward: A macrocreation ratio of 1.1 is absolutely astounding. I was expecting something on the order of 20% larger each cycle, but more than doubling the length of the arms makes it a whole lot more practical. How much larger does it actually get with arms of twice the length? Can you then produce arms twice as large as the previous generation could, or is it a smaller increase?

Also, those beds look really nice. Did a reuleaux triangle (star of the Poul Anderson story The Three-cornered Wheel) just happen to fit, or is there a particular reason it's that shape? I'm also sort of interested about the bed material. It looks very nice, but I don't think I've ever seen anyone use anything quite like it on a reprap.

@Buytaert: There's a reason I said after the design is semi-finalized. The reprap project as a whole seems to be a bit prone to fragmentation and occasional loss of focus; I happen to like Simpson and want to see it succeed.

Edited 2 time(s). Last edit at 08/21/2013 11:22AM by terribleperson.

@terribleperson: The reuleaux triangle matches the usable area. For production I am putting a circular heating pad under the plate so a reuleaux triangle is a good way to fit a circle between 3 shoulder mount points.

The basalt will act very similar to glass. It just happened to be cheaper to obtain and water jet cut. It also doesn't need to be tempered after being water jet cut like glass would.

@Buytaert: Your math is a bit off but understandably. When I say "3:4 Simpson" I mean the arm length vs shoulder mount separation is 3:4. The macrocreation factor for a 3:4 Simpson is 110%

Let's say I stick with the 3:5 Simpson with a practical macrocreation factor of 50% (theoretical factor of 67%).

Generation 0: 250mm (longest printable dimension)
Generation 1: 375mm
Generation 2: 562mm
Generation 3: 843mm
Generation 4: 1.27m
Generation 5: 1.90m
Generation 6: 2.85m
Generation 7: 4.27m
Generation 8: 6.41m
Generation 9: 9.61m
Generation 10: 14.42m
Generation 11: 21.6m
Generation 12: 32.4m
Generation 13: 48.7m
Generation 14: 73.0m
Generation 15: 109.5m
Generation 16: 164.2m
Generation 17: 246m
Generation 18: 369m
Generation 19: 554m
Generation 20: 831m
Generation 21: 1.247km

Let's say we dream about the 110% 3:4 machine a bit.

Generation 0: 250mm
Generation 1: 525mm
Generation 2: 1.103m
Generation 3: 2.32m
Generation 4: 4.86m
Generation 5: 10.21m
Generation 6: 21.4m (House)
Generation 7: 45.0m
Generation 8: 94.6m (Football Field)
Generation 9: 198.6m
Generation 10: 417m
Generation 11: 876m
Generation 12: 1.839km
Generation 13: 3.86km
Generation 14: 8.11km
Generation 15: 17.03km (Manhattan)
Generation 16: 35.8km (Marathon)
Generation 17: 75.1km
Generation 18: 157.7km
Generation 19: 331km (Width of Arkansas)
Generation 20: 696km
Generation 21: 1.461Mm
Generation 22: 3.07Mm (Width of Australia)
Generation 23: 6.44Mm
Generation 24: 13.53Mm (Earth)

When I said I wanted to promote the spread of 3D printing I didn't think about it literally taking over the world. (Where is Randall Monroe to make me some cute illustrations of Simpson taking over the world?)

Edited 1 time(s). Last edit at 08/21/2013 01:45PM by nicholas.seward.

It occurs to me (as I'm sure it has to all of you) that there's no way it's going to get past even a meter before we start encountering structural issues. Plastic does not scale up all that well. Machining suddenly becomes a whole lot more important. We need a machine that can both print and machine with no structural changes, just a head swap. If we can print and machine, Simpson can continue to scale up.

edit: I'm going to admit I'm a bit biased towards a string-driven Simpson. That said, I think it would actual be easier to solve the problem of the drive pulley spinning than to use gears and have to solve backlash. I'm not sure but I think the strings should somewhat absorb vibrations. The elasticity on a spectra line is below 1%, but it IS slightly elastic.

edit2: Wait a second. Why would the drive pulley spin? Isn't the line permanently affixed to the pulleys, no friction necessary?

Edited 3 time(s). Last edit at 08/21/2013 03:25PM by terribleperson.

This thread is getting wild, and very interesting.

nicholas.seward Wrote:
-------------------------------------------------------
> Announcing the Gear Drive. Guizmo made a
>
> new thread to discuss a bearingless Simpson
> among other things. (It is a great read if you
> like math and mechanisms.) One product of that
> thread is the Gear Drive or the Proportional Gear
> Drive Joint. It has smashed all my expectations
> and was so impressive that I am rolling it into
> the beta build.
>
> It is strong. It can apply more the 5lbf at the
> arm ends.
> It is fast. It can go faster than 500mm/s
> It is simple. 2 plastic parts. No bearings.
> It is a two sided drive. Gravity and springs
> don't limit performance anymore.
>
> src="//www.youtube.com/embed/PKkfn1GKuMo"
> frameborder="0" allowfullscreen>
>
> I look forward to modeling what beta Simpson is
> going to look like now.
>
> Todo: I am waiting on springs and guitar tuning
> pegs for the motor arms. I also need to add limit
> switches. Spring assisted auto-home is gone. I
> have the perfect place for the limit switch.


Wow, I feel happy this design is going to be used on the beta machines! I'm still working on a few ideas derived from that concept.

Just curious, how much additional error is introduced in each macrocreation iteration? Will it scale up at the same rate, or will we be seeing exponential increases?

nicholas.seward Wrote:
-------------------------------------------------------
> @terribleperson: The reuleaux triangle matches the
> usable area. For production I am putting a
> circular heating pad under the plate so a reuleaux
> triangle is a good way to fit a circle between 3
> shoulder mount points.
>
> The basalt will act very similar to glass. It
> just happened to be cheaper to obtain and water
> jet cut. It also doesn't need to be tempered
> after being water jet cut like glass would.
>
> @Buytaert: Your math is a bit off but
> understandably. When I say "3:4 Simpson" I mean
> the arm length vs shoulder mount separation is
> 3:4. The macrocreation factor for a 3:4 Simpson
> is 110%
>
> Let's say I stick with the 3:5 Simpson with a
> practical macrocreation factor of 50% (theoretical
> factor of 67%).
>
> Generation 0: 250mm (longest printable dimension)
> Generation 1: 375mm
> Generation 2: 562mm
> Generation 3: 843mm
> Generation 4: 1.27m
> Generation 5: 1.90m
> Generation 6: 2.85m
> Generation 7: 4.27m
> Generation 8: 6.41m
> Generation 9: 9.61m
> Generation 10: 14.42m
> Generation 11: 21.6m
> Generation 12: 32.4m
> Generation 13: 48.7m
> Generation 14: 73.0m
> Generation 15: 109.5m
> Generation 16: 164.2m
> Generation 17: 246m
> Generation 18: 369m
> Generation 19: 554m
> Generation 20: 831m
> Generation 21: 1.247km
>
> Let's say we dream about the 110% 3:4 machine a
> bit.
>
> Generation 0: 250mm
> Generation 1: 525mm
> Generation 2: 1.103m
> Generation 3: 2.32m
> Generation 4: 4.86m
> Generation 5: 10.21m
> Generation 6: 21.4m (House)
> Generation 7: 45.0m
> Generation 8: 94.6m (Football Field)
> Generation 9: 198.6m
> Generation 10: 417m
> Generation 11: 876m
> Generation 12: 1.839km
> Generation 13: 3.86km
> Generation 14: 8.11km
> Generation 15: 17.03km (Manhattan)
> Generation 16: 35.8km (Marathon)
> Generation 17: 75.1km
> Generation 18: 157.7km
> Generation 19: 331km (Width of Arkansas)
> Generation 20: 696km
> Generation 21: 1.461Mm
> Generation 22: 3.07Mm (Width of Australia)
> Generation 23: 6.44Mm
> Generation 24: 13.53Mm (Earth)
>
> When I said I wanted to promote the spread of 3D
> printing I didn't think about it literally taking
> over the world. (Where is Randall Monroe to make
> me some cute illustrations of Simpson taking over
> the world?)

If the bed were on a turntable, you could maximize the 3:4 design?

@nicholas LOL my math is a bit off. At these rate, it won't be too long we need carbon nanotube structures.

Turntable ? That's look like 3D scanner.

jason.fisher Wrote:
-------------------------------------------------------
> If the bed were on a turntable, you could maximize
> the 3:4 design?

So, I just realized this: Simpson doesn't have to print layers parallel to the printing surface. This should allow for optimization of part strength and print orientation. In fact, the layers don't even have to be flat! An example would be a wavy bowl where the layers followed the contours of the lip up and down in addition to sideways. While it may not be the optimal slicing, it would be an interesting design aesthetic. Maybe this is inherent in delta printers, and I just didn't know about it until now, or maybe I'm just crazy.

@Matt M

Something like this?
[www.thingiverse.com]

Yes! That is exactly what I was talking, thanks for the vid. So I guess it's not just for delta printers, but Simpson seems like it would lend itself nicely to this type of printing.

Are there any specifics on the resolution of Simpson? Specifically the Z axis? Is resolution a function of the extruder, the motors, the software, or a combination?

While I'm about as far from an expert on 3D printers as you can get, since no one else has responded yet, I thought I'd give an explanation a try.

The first thing to remember is that resolution can mean two different things with regards to a 3D printers. It can refer to the accuracy with which plastic is placed (dependent on the motors, the software, and limited by structural constraints that might introduce error), or the quantity of plastic that is placed on a given spot and the precision with which you dispense plastic (which is dependent on the extruder, and possibly the software, I think). The first type of resolution is effectively limited by the second in that it doesn't matter how accurate your placement is, if you're extruding 1mm globs of plastic, you're not going to be able to put stuff closer together than that.

With regards to the type of resolution you're referring to and given the design of Simpson and the stated opinions of nicholas.seward and Annirak, I imagine resolution is almost entirely down to the motors and drive system. I doubt they'd let software be a limiter on the performance of Simpson.

edit: Printing layers that aren't parallel to the surface and/or curve (and possibly abandoning the layer system or at least partially replacing it, although that'd be a major undertaking) is indeed really neat.

Edited 1 time(s). Last edit at 08/27/2013 12:44AM by terribleperson.

On the subject of resolution .. a stainless steel variable iris-type nozzle tip could be well-suited to a large format Simpson? Nozzle discussion might be off-topic for this thread, so I created a new one here .. [forums.reprap.org]

@Matt M: The resolution is anything that you want it to be. I shoot for 25 microns in every direction which would be great for 3D-3D printing. (Okay, I just did the math and the filament I use causes extrusion width inconsistencies of +/-4 microns.) Even if you wanted to shoot for 4 micron resolution, Simpson could still rapid at 100mm/s.

I have been wanting to do egg crate (sin(x)sin(y)) slicing for a while. I feel this would eliminate the possibility of layers shearing off.

@jason.fisher: Great idea! I posted on your thread.

Here is some awesomeness.

Edited 1 time(s). Last edit at 08/27/2013 03:22PM by nicholas.seward.

That is pretty damn neat. I wonder if you can do scratch holograms on basalt?

wow, it looks incredible! can't wait to build it. I'm gonna take a week off work to enjoy this build.

@terribleperson: I don't think the engraving has the right optical properties to make a hologram. A vinyl cutter like device could. :-) You probably need a diamond edged tool. (I never heard of scratch holograms before. Thanks for introducing me.)

Ah, I guess you looked them up? The best example of them is a video on youtube by wbeaty (of amasci.com fame). Everything the man says should be taken with a grain of salt, but he often has some interesting insights, useful ideas, and his list of "things that are wrong in textbooks" is great. There's quite a few more videos of them on youtube than there were when I first heard of them, but wbeaty's is still the best (and most complex) example of them. At the end he shows off an (I think) acid-etched copper plate with various test holograms on it.

Also, now I'm thinking it'd be neat to be able to have a CAD program in which you could custom-design the layers for a part.Print a sphere from concentric layers, stuff like that.

@nicholas.seward: I've been reading through this thread with great interest, and I really like the direction that this design is going. I am particularly impressed by the way you managed to find a great balance between cost, reprap-ability, print volume, and precision. The potential to do milling is very enticing, and it occurs to me that you could make almost every plastic part out of alumiun (on a standard cnc mill the first time) without any modifications. I would think that this should result in a Simpson design that can handle milling, and still be self replicating.

The only change that might be needed is to redesign the extruder mount to accomidate a spindle which is larger, more massive, and spinning.

On another note, have you posted the full cad files for the newest iteration of Simpson? I would love to throw the model into Solidworks during my lunch breaks and see how things might be milled.

One last thing, it might be possible to design the mounting base to be printed if you split it into three triangular sections that are bolted together. This would further reduce (albeit just slightly) the number of non-reproducable parts.

@pjoyce42: Sorry, all my development is in Autodesk Inventor. (I dream of there being a FOSS MCAD solution that can do the 10% of the functionality of Inventor or Solidworks that matters.) However, I am getting close to doing a full release of the files. I could release DXF files along side the Inventor files and the OpenSCAD files so you can experiment.

If you want a larger spindle, here is one idea. I am not a fan of the harder kinematics but it would work.

Not to discourage but Simpson has the potential to develop so many different (possibly catastrophic) vibrational modes that I worry about milling aluminum. I think it is worth an experiment. (I flash back to my attempt to mill aluminum on my MDF hobby mill. My fairly sturdy machine hit this natural frequency that caused the effector to oscillate by over 1/4". If you pushed with 200 lbf you wouldn't get a deflection like that. Hitting a natural frequency is bad news and in some cases can rip a machine apart. Think Tacoma Narrows bridge.) General engineering wisdom is to use one or more of the following techniques.

*Increase mass: This lowers the natural frequencies and reduces the amplitude of vibrations.
*Increase damping: This can remove vibrational modes altogether and will reduce the amplitude of those remaining.
*Passively Isolate: This is really just a specialized form of increasing mass and adding damping.
*Actively Isolate: As far as I know, no one has come up with a cost effective way to use an active control system to mitigate vibrations in a production machine. However, this has the potential to allowing a light, low friction machine to do unusual things like mill metal. However, however, the processing power needed to run a system like this would be outrageous along with the fact that reaction time needs to be a fraction of the period of the oscillation that you are mitigating. Assuming the hardware existed, the software would still have to be written. Hopefully, this will be an option someday.

As you can imagine that using any of the first three ways will reduce speed and/or increase motion control costs.

I have to admit, I was wondering how you managed to make such amazing models in any of the FOSS MCAD suites. It makes a lot more sense now. I actually have Inventor (and the simulation suites (I love that my autodesk accout still works from when I was a student)) on my personal computer, and I use Solidworks all day, everyday at work. I'm a mechanical engineer, so this sort of thing is right up my alley. I would be more than happy to have the inventor files if you are okay releasing them. Otherwise I'm fine waiting for the full release.

Have you been able to run any freq./resonance simulations to see where you might run into issues? My gut instinct is that your original elbows are actually perfectly suited for resonance damping because you can change out the spring(s) to provide some level of damping.

Is anyone working on the kinematics for marlin? This machine is truly amazing and would love to print one on my prusa but my math is not good enough to tackle it.

A man's got to know his limitations. (Clint Eastwood, Magnum Force)

Roger

I believe the Rostock firmware works with some modification?