Grounded Experimental Delta Printer

VDX Wrote:
-------------------------------------------------------
> Hi Nicholas,
>
> really well done!
>
> It shows, the stability of the head is good enough
> to stay in place with low torques.
>
> Next question - how fast can you go with some
> 'beefier' motors?

The motors are more than beefy enough. I could easily pull off NEMA14. I actually might be able to go faster with them because inertia is the limiting factor here. However, NEMA14 motors would increase the price.

Currently I have the acceleration set really low to make it gentle on the design. I have yet to do speed tests. I have had it going as fast as 300mm/s but with the current acc settings that couldnt happen in a normal print.

I don't want people to freak out when the elbows flap around. However they can be displaced up to 15mm and not affect the effector's location.

Nicely done dude!

Looks great!

First, do you guys think I can make my drive pulley fit very snugly on the stepper and in the process remove the set screw? If so, that would simplify assembly. My wife also suggested that I put a spiral groove on the pulley. I face palmed because that is so obvious. I should talk through the design more often with her.

As you can tell, I am part reduction mode. The lasercut version has lots of little parts to keep the lasercut parts from having to have any crucial dimension. The printed version will have about 50 printed parts and about 15 unique printed parts. It took me 12 hours to assembly my current version. That was with a few problems and reprints. I am shooting for a sub 4 hour assembly time. It takes about 2 hours to laser cut so 2 people could assemble Simpson as fast as the machine can churn out parts.

I am pushing ahead with what I hope will be the final redesign (excluding minor tweaks) before I publish. Everything is going to be parametric so it should be as easy as changing a number or two to get a drastically different Simpson. I have killed off all the non-metric hardware that I used for prototyping. (For those that are purest, give us Americans a break when we use non-metric stuff. I went to a specialty screw shop which happened to be the only place in town with a good selection of metric hardware and I ended up spending 10 times what it would have cost to use non-metric hardware. Luckily, the internet exists and on the internet there is not much difference in price.) The grand total of a build is looking to be less than $400 dollars if you go with all the bearings I use. Loose bolted connections should work just fine and that could bring the cost down quite a bit. I don't want to experiment with no bearings right now but I will put a toggle in OpenSCAD that will let builders turn on or off the bearings. (The no bearing option will probably perform the same but will be louder. Plastic squeaking sounds will abound. A few drops of oil on each joint will fix this for a period of time.)

I just put a order in for more springs and M8 button head cap screws to replace my 5/16" stuff. I actually am going to use weaker springs but more of them. Each arm will have an array of 4 springs. I will also make it so these springs will just hook onto pegs instead of having to bolt threw them. This will make assembly much easier. These springs can go from 2.5-7.5 inches instead of 3-7.25. This will help protect the spring from overstretching and will keep the force in a more acceptable range.

I also am going to put limit switches on the machine. The limit switch will detect when there is no tension on the string. This will let me do some pretty neat stuff. I will be able to put Simpson into a mode where he takes out any slack in the string. (I am going to call this the "posing" mode.) This way I can push the head down to a calibration position and the motors will follow. Also, once I put this protection into the firmware, having mess ups that cause you to restring Simpson will be few and far between. Additionally, homing will now consist of slowly running the motors until the tension is off of all the strings.

Calibration can now take place in less than a minute. I can have a piece of paper with lots of numbered dots that you put on the bed and you can push Simpson onto each dot. Send some gcode commands for each to remember each "pose". Use all this data to solve for the geometry of the machine in the firmware and away you go. (Normal Equations, Simplex, Gradient Descent, etc. One of these things will be able to solve this problem.) The extruder's calibration will be harder.

This is all very exciting. (At least for me.) I haven't even talked about some of the other optimizations I have planned. Stay tuned. This week should be very exciting. I almost have the calibration algorithm worked out so I should be able to print the arms for baby Simpson. I also just realized as I was typing this that Simpson can probably print out 12 inch arms when his longest arm is 10 inches. This will allow Simpson to be able to easily print a larger Simpson.

Very exiting indeed!

Congrats on getting it going. Such a magical moment when you see your creation actually starting to do what it was made to do, and it actually works!

Cheers,
Quentin

Beautiful machine, you have skills

Really looking forward to more video!

- Marinus

Nevermind, I found it!
[youtu.be]

Edited 1 time(s). Last edit at 06/08/2013 06:51AM by Ohmarinus.

I did a complete rebuild. I stiffened the springs, stiffened the central hub, increased the mechanical advantage, got rid of no go areas on the bed, made a workable leveling algorithm. It was a lot of work but there isn't much to show. It looks almost the exact same as before. Trust me that it is better.

I currently have some scaling issues and I have to perfect the leveling algorithm. Once I fix these two things, Simpson is going into reproduction mode.

Video Update!

Edited 1 time(s). Last edit at 06/09/2013 06:34PM by nicholas.seward.

i really like this project, thanks for the updates

Awesome update, the machine is SO CUTE!

Incredible... Initially I had really negative thoughts about the outcome, you proved me wrong, really, thats worth something

If math is going to be an issue just use a beagle bone with sysbios you get real time at 720Mhz

nicholas.seward Wrote:
-------------------------------------------------------
> I also
> just realized as I was typing this that Simpson
> can probably print out 12 inch arms when his
> longest arm is 10 inches. This will allow Simpson
> to be able to easily print a larger Simpson.

Ok, to me this printer design was already EXTREMELY interesting, but this point just made it even more so. Every cartesian RepRap design I've seen has the limitation of not being able to print parts for another printer of teh same design with a build volume larger than itself (without "cheating" by buying larger metal parts, that is). Frankly, this has made me think of the name "RepRap" as a bit of a stretch. Rostock delta based designs could possibly do this vertically, but they require such great height themselves. And while they are a great design, too, and arguably using fewer parts than cartesian designs, they still require a lot of metal extrusions/smooth rods.

THIS design is potentially superior on these points--key points, IMO:
1. Most of the parts are relatively flat. This can lead to fast, reliably printed, precise parts.
2. Requirements for precision machined metal are virtually eliminated and now just nuts, bolts, extruder, and motors (and bearings, but they might be printable).
3. It can print children that are larger than the parent: this means even the smallest RepRap can be used to bootstrap to a larger one of this lineage!

This design deserves the moniker "RepRap" more than any other design so far (well, in my opinion at least).

GREAT WORK!

---

1st working printer: Printrbot Plus v1 (built from kit, not pre-assembled) with Super-Z upgrade.
2nd printer currently under construction: Printrbot Plus v2 frame, with RAMPS electronics and TrinityLabs.com Magma hot end.

This sure has progressed while I've been away. It's looking very good!

Looking forward to your release!

@Annirak: I am going to refer to the arm's drive mechanism as the Annirak Drive. Better yet, I will call it whatever you want me to call it and attribute you with it. Do you want me to use your real name? What is it?

@xvart
If math is going to be an issue just use a beagle bone with sysbios you get real time at 720Mhz

How it works (Firmware or application for reprap Printer)?

The math is no more complex than the Rostock meaning that this can be done in standard firmware. I however have not gotten around to doing the firmware. I do have a gorgeous Python script that will transform the coordinates and break long lines into segments. It will even go so far as incrementally adjust the sent feed rate to keep the real feed rate constant.

After I get baby Simpson done, I will revisit the firmware. I know it can be done so I am currently focusing on things that I need to prove that Simpson can do like replicate.

When you were at QU-BD, did you see their printers?

---

[www.matter-replicator.com]

I actually started working there part-time (You would think working on Simpson would take up all my time. The trick is not sleeping.) to just be around some cool people and machines. (Hopefully, it will help me get my printer faster) They are currently building up a lot of XL printers. I can assure you that the machines are extremely well engineered and they are being assembled with love. The attention to detail is incredible. I am looking forward to getting the one my school bought. Simpson is sweet but I can print at least 4 times faster with an XL.

The XL and Simpson have about the same build volume. My build area is about 75% bigger which will let me do more little parts in an unattended print.

Hopefully, they will get all the kinks worked out on the XL and we can see about bringing Simpson to market. I just finished my final BOM. The cost if you part it out yourself is $388. I would be curious to sit down with QU-BD and see what prices than get if we order in bulk. If I had to guess, the end user will have to pay $400-$500 for an mostly assembled Simpson. (I think I have gotten the partial assembly down to 9 bolts that hold together 5 sub-assemblies. End user assembly time should be less than 30 minutes.)

Side note: Those of you lucky enough to be able to attend the Kansas City Maker Faire, I will be there with Simpson at the QU-BD exhibit. I hope I get to meet some you in the flesh.

Here is a photo:

---

[www.matter-replicator.com]

It is a thing of beauty.

Did you see any RPMs moving?

There are some completed RPM's but I have never been there when they are running. I have only seen the XL in action.

Simpson now has a wiki and is on Thingiverse.

I hope I can start sleeping now.

I assume you figured out the scaling issue. Have you managed self-replication?

The scaling issue was from a typo that I have since fixed. I pushed really hard to self-replicate but ran out of time to get it done for the Gada prize. I am going to take a week off and then hit it again. I don't see any reason why Simpson won't become a true ReRap.

My tirbute to Simpson

That was one of the best articles that I have ever read. Good job making me sound smarter than I actually am.

Thanks! That's a nice thing to say!

No way, jose. You're the one having conversations about mathematical formulas and referencing quantum evolution. You designed your own printer!

It is incredible to see what a community of smart and well-spirited people can do in a month.

Congratulations. Now put it on kickstarter and take my money

Hey everyone, 100% newbie here. I would like my first 3d printer to be one of these open source, self replicating kits. I really love your design here Nicholas. Could you go into some detail about your printer vs the Tantillus? Where do you see the Simpson a year from now? Does your printer natively support Blender files? How about ABS plastic?

As a newbie to both 3d printing and CAD would you recommend diving in using your Simpson? Should I wait for a few revisions of the Simpson?

Also, I would love to donate to keep you working on this beautiful printer. Do you have a Bitcoin address?

Thanks for the wonderfully clear youtube videos and dedication to 3d printing and open source!

John McKinney Wrote:
-------------------------------------------------------
> Hey everyone, 100% newbie here. I would like my
> first 3d printer to be one of these open source,
> self replicating kits. I really love your design
> here Nicholas. Could you go into some detail
> about your printer vs the Tantillus? Where do you
> see the Simpson a year from now? Does your
> printer natively support Blender files? How about
> ABS plastic?
>
> As a newbie to both 3d printing and CAD would you
> recommend diving in using your Simpson? Should I
> wait for a few revisions of the Simpson?
>
> Also, I would love to donate to keep you working
> on this beautiful printer. Do you have a Bitcoin
> address?
>
> Thanks for the wonderfully clear youtube videos
> and dedication to 3d printing and open source!


First, Simpson isn't beginner friendly yet. I would wait a year and if I don't iron everything out someone will.

The Tantillus costs about twice what mine does and has an order of magnitude smaller build volume. That said I would suggest it or a Mendel as a first machine.

The 3d printing tool chain will allow for the use of anything that can produce an stl. Blender can do that.

ABS would be no prob.

I will have to get back to you on donations. Thanks for the offer.

I have only tentatively talked with QU-BD aboutcarrying a laser cut kit for Simpson. It is at least a few months out.