Grounded Experimental Delta Printer

Quote
nicholas.seward
I already have the electronics for the beta kits but all of those will be getting paired with the mechanicals so I can do some consistent beta testing. It won't be until December when I will be doing this for real.

So does this mean that there will be kits? That post seems a bit old, so I'm not certain. I was just starting to figure out what it would take to buy the parts and find someone to print it for me for a reasonable price, but if you're planning on doing all that...

And if you are doing kits, is there an ETA/backlog? Or would someone else be willing to print me a set of GUS parts?

Also, is the csv on the older Simpson wiki page still valid, or is the one in the github the correct final version? Just wondering, because the one in github doesn't seem to have links for all of the parts.

@Samuel: I have canceled any plans of offering kits until future notice. The margin is just too small if I was to print the parts and the volume is just too big if I wanted to injection mold the parts. Use the GitHub BOM. The other one is for the original version.

I am trying to work up to being able to offer kits at volume but I have a lot of things to get in place first. It could be a year or more before I get to that point.

No worries. I was already looking into acquiring the parts myself. Wouldn't want to duplicate effort if you were already working on it though.

On another note, how seriously do you think it would affect the design if I used a set of NEMA 23 motors I already have left over from an older project? Obviously the mounts would have to be enlarged, but would the extra weight cause any effects I should take into consideration?

@Samuel: It depends on the weight of the NEMA23s. If you use ones like the 400oz-in ones that I use on my CNC mill it will severely limit your top speed. If you are only talking about 2 or 3 times the weight then you are probably fine. However, NEMA23s can be as much as 6 times the weight.

Nicholas, that mod looks great. Why not make that the default? I don't see any disadvantages. If there's a part you're on the fence about, you can alway push it to a branch on GitHub. Then everyone can have access.

Also related to the part, where does the string go from the tuner spool? This is what I'll call the part that the string goes in and then turns, on the guitar peg winder. Looking at the Simpson Flickr set (e.g. [www.flickr.com]), the positioning of the tuner spool is quite different. On the most recent files I'm looking at, the tuner spool and a screw are both popping out of the big flat side, just like the motor. On the pics, the tuner spool goes into the channel between the two gear halves. I have to assume this was some design improvement; do you have any pics with the (presumably newer) motor arm showing the line winding? Thanks.

Quote
brandonh
Nicholas, that mod looks great. Why not make that the default? I don't see any disadvantages. If there's a part you're on the fence about, you can alway push it to a branch on GitHub. Then everyone can have access.

Also related to the part, where does the string go from the tuner spool? This is what I'll call the part that the string goes in and then turns, on the guitar peg winder. Looking at the Simpson Flickr set (e.g. [www.flickr.com]), the positioning of the tuner spool is quite different. On the most recent files I'm looking at, the tuner spool and a screw are both popping out of the big flat side, just like the motor. On the pics, the tuner spool goes into the channel between the two gear halves. I have to assume this was some design improvement; do you have any pics with the (presumably newer) motor arm showing the line winding? Thanks.

To partially answer my own question, just came across a post right before the files release where Nicholas explains the reason for the changes. In the picture, the rounded edge of the motor arm part makes it clear that the line wraps to to the edge and then around. Still, from there, I'm not sure. Nicholas or Owens, could you describe the flow of the line wrapping? Thanks

Edited 1 time(s). Last edit at 12/02/2013 06:07AM by brandonh.

I just ordered all the parts, but one more question, then I'm done for awhile.

About the Bowden connection - interesting to not see a pushfit here. I take it the m4 nut must be threaded onto the PTFE tube, which is then secured with the bowden nut. Did you use RichRap's method to externally thread the PTFE tube? Never done this, so wondering if there's an easier way; it seems slippery. Did you encounter any issues with push fits to motivate the nut approach, which requires a bit more PTFE prep? Thanks again.

Edited 1 time(s). Last edit at 12/02/2013 08:11AM by brandonh.

@brandonh: Added the branch. Great idea. I am new to source management. [github.com] I will merge it once we can verify that it doesn't cause any problems.

The string goes from the pulley around a motor/slave bolt pair and then around the back edge and to a spring or the tuner. Sorry no pictures yet. I have the switch over in my queue. :-)

Taper the end of the PTFE tube and then thread a M4 nut directly on. I have yet to test it myself but owens assures me that it is super strong as in it can't be pulled off by hand. Try it out and let me know. I opted to not go for the pressfits because they are a specialty item and I figure I already exceeded my quota with the tuning pegs. I originally had a pressure cone. That worked but if you over tighten the cone decreased the ID of the tube and added drag.

I just realized I probably should make the BOWDEN NUT taller so it can keep the tube perpendicular to the nut. I will test this soon. No need to optimize unless there is a problem.

Quote
nicholas.seward
@brandonh: Added the branch. Great idea. I am new to source management. [github.com] I will merge it once we can verify that it doesn't cause any problems.

Thanks! Looks good here. Let me know if you have any Git/GitHub questions. I'm sure you'll be seeing Simpson pull requests shortly.

Quote
nicholas.seward
The string goes from the pulley around a motor/slave bolt pair and then around the back edge and to a spring or the tuner. Sorry no pictures yet. I have the switch over in my queue. :-)

Got it. If you or owens can post a quick wrap video, that'd be great; when wiring the Legacy Kossel a year ago, I found a video extremely useful and so much better than any text or diagram.

Quote
nicholas.seward
Taper the end of the PTFE tube and then thread a M4 nut directly on. I have yet to test it myself but owens assures me that it is super strong as in it can't be pulled off by hand. Try it out and let me know. I opted to not go for the pressfits because they are a specialty item and I figure I already exceeded my quota with the tuning pegs. I originally had a pressure cone. That worked but if you over tighten the cone decreased the ID of the tube and added drag.

I just realized I probably should make the BOWDEN NUT taller so it can keep the tube perpendicular to the nut. I will test this soon. No need to optimize unless there is a problem.

Right, but then you've got to thread an m4 nut on slippery PTFE for 3 full inches, which seems like a bit of extra effort. If the nut was at the bottom of Hub Bottom, next to the hotend, and inserted first, you could eliminate the Bowden nut entirely, as well as chop 20 mm of plastic off of this part. Yes, you might need to thicken Hub Bottom to do that, and yes, it might be harder to remove the tube.

Also, seems odd to have a 5mm diameter hole through Hub Bottom for a 4mm OD tube. This seems too tight for a 3mm-capable PTFE and too loose for the specified 4mm/2mm PTFE tube. If this was a loose slip fit for the 4mm tube - something like 4.2mm - then you could insert a carefully sized piece of PTFE into the hole in Hub Bottom, lock it down with the nut screwed onto the main-length PTFE, and avoid the 3-inch threading. Without this I doubt the filament would necessarily find its way in. When your hole _to_ a pushfit (or other PTFE tube) is larger than the push fit's hole, I find the filament does just this - get stuck on the lip - making loading frustrating.

Random note to other forum users looking to mod the GUS parts: if you're using 32-bit windows, the GUS Inventor files can quickly use up the default 2GB per-app Windows max, and then Inventor complains about being low on VM and/or freezes. Making the change to "3GB mode" made a huge difference, and now Inventor no longer complains about being low on virtual memory. Plus, I didn't need to upgrade to 64-bit windows to get more memory available.

Edited 1 time(s). Last edit at 12/02/2013 09:08AM by brandonh.

@brandonh: I terminate the PTFE tube at the top and then have another stub tube that goes through the hub.

You are right that 5mm is odd. It should work but I will downsize it to 4.2mm.

I will work up a video this week if I can.

Edited 1 time(s). Last edit at 12/02/2013 09:14AM by nicholas.seward.

New pics added to the Flickr Simpson set. The two arms have all of their major hardware installed (the endstop is missing, but that's because I haven't printed the new eccentric bit yet). Some tips on that:

- I didn't need to drill out any holes, but I did have to clean out the flash and strings pretty well in order to get things to fit.
- Despite not *needing* to drill them out, I did clearance the M3 holes for the motor with a #31 bit, just so the screws wouldn't bite into the plastic and make it harder to get the motor installed.
- Installing the four M8 bolts is a little bit weird, though you'll get used to it quickly enough (you have twelve to do). First, take a nylock nut and thread it onto the bolt the wrong way. This takes a little bit of doing; I found that the best way was to put the nut in a 13 mm closed end wrench and set it on the table, put the bolt into a deep 13 mm socket, then carefully line up the bolt and work it into the nylon. The socket gives enough grip to hold the bolt without having to use a ratchet handle. Get the nut about 3/4 of the way along the threads, then work the bolt into position in the arm (if you thread it all the way, it'll be hard to line up with the hole). Grab the nut with an open end wrench and turn the bolt with another wrench. Once you have the nut threaded as far as it will go, push the bolt through the arm all the way and put another nylock on the outside. Don't rely on the plastic to hold the nut; it might work, or it might split.
- The guitar tuner might need just a little bit of enlarging on the smaller hole, the one the shaft fits through, but the large hole should be a nice snug fit.
- The bolt on the end of the motor arm was a nice fit; the bearings on the slave arm, as expected, needed heat. I used a hair dryer for <30 seconds, then pushed the arm onto the bearing while it was sitting at the edge of my workbench.

I ran out of time before I could try stringing the arms, but I am not certain I understand how they're going to be set up and I'm also not sure how the pulley is supposed to go - screws closer to the motor makes sense to me, but I could imagine doing it the other way so you'd have a chance of removing it without dropping the motor (not that it's a big deal either way). I'm also trying to figure out whether the pulley is supposed to have grooves or holes for the string on the outside; I think grooves, but my printer pulled strings across them and filled them in so I'll have to clear them out again.

Quote
brandonh
Right, but then you've got to thread an m4 nut on slippery PTFE for 3 full inches, which seems like a bit of extra effort. If the nut was at the bottom of Hub Bottom, next to the hotend, and inserted first, you could eliminate the Bowden nut entirely, as well as chop 20 mm of plastic off of this part. Yes, you might need to thicken Hub Bottom to do that, and yes, it might be harder to remove the tube.

There's no way you could possibly thread the M4 nut onto the tube that far; it's barely possible to get the tubing all the way through the nut in my experience. However, that is enough to hold the tubing extremely tightly. I put nuts on both ends of a short section of tubing and couldn't pull them off again.

My GUS prototype actually has pushfits on both ends, because I had a bag of them and because I got into trouble with the force cone. It held the tubing perfectly well, but also caused increased friction and with all the crud we were dealing with at Maker Faire any additional friction was too much. I cut a short piece of tubing to slip into the hub, chamfered the top end to let the filament feed in easier, and centered it with a cut-down force cone. I'll probably stick with that since I'm much too lazy to reprint those parts unless I absolutely have to

I hope this clears up any questions about the string routing. (It looks like a subway map to me. I hope it will help people find their way.)

Edited 1 time(s). Last edit at 12/02/2013 03:04PM by nicholas.seward.

As you guys have noticed, this thread is getting out of hand and will just get worse once a lot of people start making Simpsons. For those that are building, I suggest that we move the party.

http://forum.conceptforge.org/

So now that the files have been released for GUS, I have one question: Is the GUS Simpson a viable option as a first 3D printer? Why or why not?

Obviously I know things are still in the beta stage, but I'm thinking about once the rough ends are fixed. I really love the simplicity of the design, and I have no doubt that it will become a popular design given the high percentage of reprapable parts. However, I know that I am not the only person in this thread who has been looking at GUS as an option for a first printer. Before too many of us bite off more than we can chew, I'd like to know how big of a project it would be to try and build a GUS from just the github files, and this forum.

@ pjoyce42

Quote
pjoyce42
Is the GUS Simpson a viable option as a first 3D printer? Why or why not?

I think Simpson is a viable printer if you are only creating objects that fit air (vase, art) or can work with large tolerances (400 microns = .016 inch).
If you need to produce objects with a higher precision then you'll have to wait for the Marlin firmware to be updated, or consider another printer.
My goal is to print my first printer with a delta printer, so I'm holding out as long as I can for the bugs to be worked out.
But I'm getting close to buying a Cartesian one

Quote
pjoyce42
I'd like to know how big of a project it would be to try and build a GUS from just the github files, and this forum

We all have different skill sets, abilities, and access to machinery.
Presently the Simpson printer looks to be the easiest printer to make, and assemble of what is available.

Quote
nicholas.seward, November 21, 2013 12:58PM
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.

[forums.reprap.org]

Quote
nicholas.seward, October 09, 2013 05:03PM
As you can see from the graph that I was off as much as .4mm. While 400 microns isn't a whole lot it should still be accounted for. You can see on the graph that when I go to 2.5 wraps like I plan to do it should bring the max error down to 230 microns. However, math is cheap.

[forums.reprap.org]

Nicholas is printing/replicating Simpson parts with a Printrbot, because the precision of Simpson is too large for self replication (400 microns = .016 inch).

As I understand it to improve the precision of the Simpson printer the Marlin firmware needs to be modified with some new math.
Nicholas mentioned that at the outer limit of the Simpson print envelope that the precision decreased to
(I'm going from memory, correct me if I'm incorrect) to .062 inch or was it .016 inch?

Quote
nicholas.seward, October 11, 2013 12:19PM
2) The math is not hard. See annirak's equations above. I haven't implemented yet because I need a few hours to make absolute certain that I make the second order adjustments in the right direction.

[forums.reprap.org]

If I understood the math I would be all over it, but unfortunately I don't
I'm slowly, and independently learning about the software and math that is involved for 3D printing.

I'm hoping that Nicholas starts a GitHub request to fund finishing the Marlin firmware, with stated goals.
As I don't see any one else stepping up to bat to solve the math problem (unless they have a skunk works project in the making?).
Nicholas, I'll kick in some money if you start a GitHub project that would allow Simpson to accurately print.
I prefer an accuracy of .002 inch, as my product designs tend to require higher precision.

I'm looking forward to the day that I too will have a Simpson printer on my desk, it just needs a few more tweaks before I can commit.

Edited 1 time(s). Last edit at 12/03/2013 01:58AM by A2.

@A2: The new design has fixed all the errors that you linked to. Testing still has to be done. Time will tell. None of the errors are due to the firmware. The main hurdle is implementing a more appropriate calibration routine. I forgot some obvious stuff the first go around. There should be no reason we can't get repeatability under .002"/50um.

@pjoyce42: We are working to make it appropriate as a 1st printer but the reality is that it is probably 1 year from that goal. Not only do we need to fix the problems, document the design, and document the build but we also need to grow the community. I have no doubt that all of those things will come to pass. I am guessing since you are on this forum that you are wanting to start before then so I would suggest finding any other RepRap project that you like and also has good documentation and a good following.

New pics and video of the fully assembled arm, linked here at the ConceptFORGE forum.

Tks Nicholas for the update, that's good news!

Calibration - Validation:
I'm hoping that the calibration works out for you, and I'm looking forward to the validation phase!

Print envelope size (new arms):
What is the largest X,Y square box length that Simpson can print with a height of 63.5 mm (2.5 inch Z)?


@Owen,
The new arms look great! Tks for posting the video.

Hysteresis, Brownian motion, White noise:
Am I seeing hysteresis (string lag) at the start of the first movement?
I don't envision it being an issue with all 3 arms in motion, and printing a raft will also account for it.

Simpson might be a candidate for the addition of noise into the system.
WWII airplane instrumentation was found to be more accurate while flying vs. on the ground.
The noise (vibration) averaged out the mechanical components, providing improved map coordinates.

Idea, Gear tooth polishing:
I wonder if the fit up between teeth, and the smoothness of the rolling motion could be improved with a chemical vapor polish?
Is there a chance that the chemical polish will relax internal stresses and distort the gear profile?
If stress relaxation is an issue, the majority of the arm can be masked off with a plastic bag.

@A2 You can fit at least 180x180mm at that height. The problem is that you would touch the edge of the print platform. (It is possible to print bigger than the platform. If you could print bigger then you could print 182x182 for the height that you listed.)

Also, the gears will smooth each other over time. To me they are super smooth to begin with.

Edited 2 time(s). Last edit at 12/04/2013 02:21AM by nicholas.seward.

Quote
A2
@Owen,
The new arms look great! Tks for posting the video.

Hysteresis, Brownian motion, White noise:
Am I seeing hysteresis (string lag) at the start of the first movement?
I don't envision it being an issue with all 3 arms in motion, and printing a raft will also account for it.

I think you're just seeing a pause before the arm began moving. I was filming with my iPhone held in one hand and reaching for the mouse with the other, so there was a brief wait while I entered the move command. If you watch closely you can see the top of the pulley turning, and I think that's exactly in sync with the arm motion. Certainly there doesn't seem to be any hysteresis at all.

@A2: Thank you very much for the summary. I have a pretty good idea of what would be involved with building a Simpson, I just wanted to get a definitive answer for myself, and others. Thanks.


@nicholas.seward: Thank you as well for the summary. I do feel that Simpsons will be a great beginner printer, and I feel that a 1 year timeline is a good goal. As for me getting a first 3D printer, I will probably get a QU-BD OneUP because of the small form factor, low price, and QU-BD's reputation. Once I get that I will (around March or April) familiarize myself with its capabilities and limits, then I will probably start working on a GUS, or Lisa Simpson.

@nicholas.seward

Simpson 3d printer calibration:
What does calibration involve, after thinking about it I don't have a clue

If I were to guess calibration would entail some of the following:
Accurately measure the center to center distance of pivots (arm lengths).
Temperature control of hot end and platen.
End stop firing.
Stepper motor voltage, amps.
Platen leveling.
The amount of nonlinear movement of the end effector.
Range of motion (envelope perimeter).

Or is it about testing sleeve pulleys for the string, and things of that nature?

Edited 1 time(s). Last edit at 12/05/2013 02:09AM by A2.

Quote
pjoyce42
I will probably get a QU-BD OneUP [...] then I will probably start working on a GUS, or Lisa Simpson.

The max travel for the OneUp is approximately 100x100x125mm while the TwoUP has 175x175x125mm.

You will need to build a smaller GUS if u get the OneUP. I don't think it's a small task to downsize a printer...

I'm not sure about LISA but the arms look the same length as the Simpson has. Maybe someone with more experience can check if you can print the parts of LISA on a 100x100x125mm bed

Edited 1 time(s). Last edit at 12/05/2013 02:00AM by TBog.

LISA's arms are currently 200mm from center to center so 228mm total. There are a few options. You can split the arms in half. You could also make them out of wood in 2 seconds if you have a 22mm forstner bit.

I have ordered the Two-up, so it looks like I may just be able print the GUS's arms diagonally. However, it will be close. By eye, it looks like Lisa should be able to build a GUS. Also, Lisa looks eminently scalable by extending her arms. Changing the arms to printed end pieces, and supplying the length with plastic, wood, extrusion, etc would appear to be relatively easy. Getting non-printed arms absolutely equal length might be hard. Can the math deal with unequal length arms?

Quote
dpharris
Getting non-printed arms absolutely equal length might be hard.

Method to drill two holes a specified distance apart with reasonable accuracy:
Using a 2d paper printer, print a blueprint with center marks 200 mm apart. An 8.5x11 inch piece of paper should work fine.
Glue the paper to your piece of wood.
With a awl (sharp point), create a center mark in the wood.
Then drill a pilot hole with a small diameter drill that fits into the center mark created by tha awl.
Verify that the drill is perpendicular in two planes to the wood. An assistant will make this easier.
The pilot hole will work as a guide for the larger drill.

You can perform these steps on three piece of wood, or alternatively you can use one piece of wood (3X thickness of one arm plus 2X the thickness of the saw kerf)
Cut the wood in thirds to create 3 arms with identical hole spacing.
Mark the end of the arms before cutting, and assemble with like ends in the same orientation.
Doing it this way it's really important to drill perpendicular holes.

@dpharris: I just checked the GUS arms, and they will fit on the Twoup with no problems as long at your print diagonally. As for the OneUp, it is not at all possible to print the GUS arms without major redesign, and breaking them into 3-4 sub-parts. As for the LISA Simpson you could easily print the arms in two pieces on the OneUp, and then couple them using some embedded nuts and threaded rod. This give you the advantage of being able to adjust the arm length so they are all exactly matching.

@nicholas.seward: Regarding the arms for GUS, what is the orientation for printing? There seem to be a large unsupported area on the motor arm that would prevent it from being printed as shown in the wiki.

@pjoyce42: You can print it in the orientation given. Most printers can bridge that far. It may sag but you have tons of layers until you get to the gear part so most printers will self-correct.

However, there is an untested version here. I am not sure if the tuner can be inserted. I think it can.

I would think that build quality (especially the strength of the inter-layer bonding) could depend strongly on the ambient temperature in the build volume.

I've tried searching for information on that subject, but get buried in other "temperature" hits, and haven't figured out how to find relevant stuff.

Does anybody here know of some relevant work on this topic?

I could imagine that even the temperature vs Z profile could matter.

Bridging ability should also be sensitive to ambient temperatures. Blow cool air upward??