Grounded Experimental Delta Printer

I like these simple designs, but that non linearity seems like an issue to me. Correct me if I'm wrong, but resolution or speed will always be at a minimum. Why I say this?

Because (depending on the specific design) almost always when printing close to the center (at 0 height) the arms are close to 90°, so speed=max and res=min. When close to the extremes some arms will be close to 180° and some close to 0°, so res=max and speed=min. And then, in certain positions (more probably with z > 0), when an arm is close to 90° and other close to 180° or 0°, both resolution and speed will be limited to the minimum value.

To me, that is an issue very well solved with the Annirak drive or the equivalent PGDJ. I see those mechanic designs as a need for resolution and speed, and not only to have easier firmware (which is great too).

@Guizmo: You are quite right. However, there is another way to look at this. My guiding thought is...

What is the simplest bot I can make that is good enough?

What I am saying is that it is okay to give up on having 25 micron resolution and 1000mm/s speeds if you get a simpler cheaper bot in return. This is more than good enough. Everyone probably has a different bar but mine is around 50 microns and 200mm/s. No one prints faster than that and you don't start getting resolution artifacts until the resolution gets worse then this.

That's ok for me too. But, can this design achieve it?

If the veredict is yes, then I guess your original design was already finished, which that takes us back to the goals question one more time. According to your guiding thought, you have been already there since the beggining (simplest bot). With some slight improvements the original Simpson would be the simplest Simpson bot with enough precision and speed. So, you have achieved it, you reached your goal. Then, from that point of view, Annirak drive and PGDJ are variations that in certain way just introduce more points of failure to the mechanics (more complicated design) with the benefit of improving software, which is not so good, as software is cheap and very easy to replicate. Please, please, please don't get me wrong, I'm not being ironic, I'm being very objetive and I'm analyzing from another totally valid point of view.

I believe, specially for the Beta testing program, you should define the goals you're aiming to, so the team can work towards them. Unless you do so, many "improvements" will be useless. And of course, the goals have to be broad so not to damage the creative process.

Anyway, to me, all the ideas and diferent designs that have born from this concept are amazing and worth trying. I believe my goals are just different. And again, thank you for letting us be part of this. I'm still on

Edited 1 time(s). Last edit at 11/08/2013 01:46PM by Guizmo.

@Guizmo: I will see how the print volume fairs on GDR and SPUD (Simple Pulley Using Device) Simpson with these requirements. I have a script that can be modified to output an STL directly so this is a good excuse to use it.

Well, I think I will just let my imagination flow. No boundaries. Not even my own. Just for the pleasure of DESIGN.

@ Solidus Labs

A few questions:

1. How many teeth on the worm wheel?
2. Does the code need the quantity of teeth, or a ratio in order to work correctly?
3. I'm curious how rigid the worm drive delta printer is, please make a video.
   Power up the bot, and with your hands push on the end effector, does it feel solid?
4. It looks to me like the forearm is longer than the upper arm.
   With unequal length arms, are you making compensations in the software, or is this a non issue?
5. Interesting use of the FR4 fiberglass on top.
   I wonder if you could etch the copper, modify it in some way, and use it as a heater element?
6. Are you modifying Marlin? If so what version of Marlin are you modifying, can you post a link, and where did you insert the new math?
7. I also would like to add encoders to each of the joints to further compensate for backlash.
   LoboCNC is working on a design to add encoders to stepper motors.
   Do you envision any problems adding encoders?
   Closed-loop stepper controller progress
   [forums.reprap.org]

Tks!

A2

Edited 1 time(s). Last edit at 11/08/2013 05:53PM by A2.

Quote
A2
@ Solidus Labs

A few questions:

1. How many teeth on the worm wheel?
2. Does the code need the quantity of teeth, or a ratio in order to work correctly?
3. I'm curious how rigid the worm drive delta printer is, please make a video.
   Power up the bot, and with your hands push on the end effector, does it feel solid?
4. It looks to me like the forearm is longer than the upper arm.
   With unequal length arms, are you making compensations in the software, or is this a non issue?
5. Interesting use of the FR4 fiberglass on top.
   I wonder if you could etch the copper, modify it in some way, and use it as a heater element?
6. Are you modifying Marlin? If so what version of Marlin are you modifying, can you post a link, and where did you insert the new math?
7. I also would like to add encoders to each of the joints to further compensate for backlash.
   LoboCNC is working on a design to add encoders to stepper motors.
   
   
   Do you envision any problems adding encoders?
   Closed-loop stepper controller progress
   [forums.reprap.org]

Tks!

A2

1. It's a 90 tooth worm wheel.
2. Since the output of the code is in degrees, the steps per unit in the config are translated into steps per degrees. So 200*8*90/360= 400 steps per degree.
3. I will work on making a video; I have noticed that the rigidity changes depending on its position.
4. The l1 and l2 arm lengths are set in the config. They are handled in the formulas so it is not an issue.
5. I think this has been done, if the heat bed is a single sided pcb, you can print on the back. .60 FR4 is not the most rigid material so it does need some support.
6. I started with the marlin delta branch and replaced the delta formulas [github.com]
7. Closed loop is always nice, I personally prefer the dc servo setup but an encoder on an elbow could help the accuracy.

I hope I can get the calibration finnished and start test printing by the end of this week.

Quote
Solidus Labs
1. It's a 90 tooth worm wheel.
2. Since the output of the code is in degrees, the steps per unit in the config are translated into steps per degrees. So 200*8*90/360= 400 steps per degree.
3. I will work on making a video; I have noticed that the rigidity changes depending on its position.
4. The l1 and l2 arm lengths are set in the config. They are handled in the formulas so it is not an issue.
5. I think this has been done, if the heat bed is a single sided pcb, you can print on the back. .60 FR4 is not the most rigid material so it does need some support.
6. I started with the marlin delta branch and replaced the delta formulas [github.com]
7. Closed loop is always nice, I personally prefer the dc servo setup but an encoder on an elbow could help the accuracy.

I hope I can get the calibration finished and start test printing by the end of this week.

@ Solidus Labs, thank you for your reply.

Comments:
2. 400 steps without using microsteps is pretty good resolution.

3. Interesting observation of the change of rigidity based on the arm location.

3.1 I'm wondering if there should be a torsional preload on the pivotal end effector, or a preload on the worm wheel?
Maybe a split worm wheel with a spring to preload both sides of the worm gear teeth?

3.2 I found an example of a polar bot that uses a spring to account for the singularity of the end effector.
I think there was a paper written up on it as well, and I'm unsure if a patent is being pursued.

I believe it was a torsion spring located between the pivotal end effector, and each of the arms.
The idea is to add a torque moment between the end effector and arm, such that it's always under load.
If you want to see the example, let me know and I'll search my notes for it.

4. Arm length: I'm hoping that the code can accommodate equal arm lengths, such that the delta bot can be scaled up and down in size.
Is the center to center distance of each of the arm pivots of equal distance, or is one arm shorter than the other?

7. If stepper motor encoders become a reality, I'm going to make an effort to incorporate it.


New questions:
8.1 How much axial end play is there on a NEMA 17 single shaft, and a dual shaft stepper motors?

It was pointed out to me that if there is axial end play in the stepper motor shaft that the end play will add to the backlash problem.
Axial end play could increase due to wear, this would create a serious problem for print accuracy.

8.2 Possible solution: With the teeth of the worm gear fully embedded into the worm wheel,
axial end play could be eliminated by pushing the stepper motor axially fully in one direction (i.e. towards the worm gear).

8.3 I'm thinking to minimize this potential problem a dual shaft stepper motor should be used, and the ability to push the motor forward axially.
I'm guessing that a single shaft stepper motor is only pocketing the stub shaft end in a bronze bushing,
while the exposed shaft is press fit into a roller bearing.
I would like to hear what others think of this potential problem.

8.4 Is axial end play at fault for the lack of rigidity of the end effector in some locations over the print bed?

9.0 Do you or any one else know where I can find a specification of NEMA 17 stepper motor shaft end play?
FYI: I don't have a motor to test.

10.0 I think it's time to start a new forum thread specifically for “Worm Drive Delta Printers”.
Would you like for me to start a new thread, I'll move my questions to the new thread.


I'm looking forward to following your development efforts, and the video, good luck with your first test print!

A2

Edited 2 time(s). Last edit at 11/12/2013 04:40PM by A2.

New topic for worm drive grounded delta printer.

[forums.reprap.org]

I'm just wondering if a few vitamins (end stops) can be eliminated with software.

Bed leveling, and end stops are they really required, or can we with precision manufacturing, and software eliminate them?

Assumptions: perfect equal arm lengths, a flat bed, and 120 degree arm separation:

Because the arms make a 3 point contact with the table, is leveling really necessary for grounded deltas?
I'm also thinking that for end stop elimination that the nozzle zeros on the table, I'm unsure if that could be accomplished without a vitamin.
But it might be an improvement?
We already know the length of the arms so the upper reach of the arms can be calculated.

I assume a portion of the code would have to be modified for this.

What are the flaws in my reasoning?

Edited 2 time(s). Last edit at 11/23/2013 07:37PM by A2.

Quote
A2
I'm just wondering if a few vitamins (end stops) can be eliminated with software.

Bed leveling, and end stops are they really required, or can we with precision manufacturing, and software eliminate them?

The software can know how long and what range of motion the arms have. But the software needs to know what position the arms are in. Thats why it moves the arms all the way to the ends and when they hit the endstops we know the position of the arms.

You only need one end stop per arm because you move the arms in one direction till you hit the switch and then location of arms is know.
If you can find a way to tell the software what position the arms are in when it's turned on with less vitamins used then a switch, then you're in luck. But to be honest, a switch is about as basic as it gets.
The same goes for bed leveling. A switch is pretty basic and only cost a dollar or two.

P.S I'm eagerly awaiting the release of the GUS simpson.
Nicholas can you give us an estimate of how many kg plastic is required?
Any development updates you can share with us?
I'm very grateful the world has people like you.
Your work is fantastic!
Thank you so much for sharing.

Edited 1 time(s). Last edit at 11/26/2013 04:11AM by TheTechnicalNoob.

Quote
TheTechnicalNoob
But the software needs to know what position the arms are in.

Thanks for pointing that out!

Idea to eliminate all end stops:
A centering guide is laid on top of the heat bed that has a cone shaped hole in the middle.
You place the hot end nozzle into the cone, which centers the arms equally.
If the nozzle profile is inadequate, or heat is an issue, you place a cone over the nozzle.
Then turn on the printer.
Now you know the location, and the angle of each of the arms without using any end stops.
The maximum altitude can be calculated with software.

A2

@ TheTechnicalNoob

Simpson cad files I believe are available?
Nicholas said last week that he is working on fixing the software.
I think he said that it's off by 1.5 mm at the outer limit of the print envelope?

Yes the cad files for the simpson are released: [www.thingiverse.com]
I'm after the cad files for the GUS simpson, stands for Geared Union (the mathematical symbol) Simpson (source for acronym: [forums.reprap.org] )
As far as I'm aware they're not released yet.
EDIT: sounds like he will put the files in his repository ( [github.com] )when they are ready.

The GUS is the simpson that uses the gears, not plexiglass.
Like this:


EDIT:Your endstopless calibration method should work, but I think I like endstops better: more precise and less error prone, less work.

Edited 2 time(s). Last edit at 11/26/2013 05:37AM by TheTechnicalNoob.

TheTechnicalNoob: Aww, you got me all excited for the GUS files! Ahh well, I won't have a second controller board until December, so I can wait.

Looking at the GUS design, I'm a little worried I'll have a tough time printing the larger parts. Any idea how large the largest part is?

The longest parts are 175mm from end to end. I just finished the new arms. I will be uploading TODAY!

Can't wait for the day when you will start shipping beta Simpson.

That's terrifyingly large, but I think I can print it if I do it sideways.

Whoo! I'll be printing ASAP then!

Quote
TheTechnicalNoob
Your endstopless calibration method should work, but I think I like endstops better: more precise and less error prone, less work.

Tks!

I agree that utilizing quality end stops is more desirable.

I'm thinking of people in tough financial situations, and people in locals that don't have easy access to all the hardware that we have available.
Reducing the vitamin count is I believe some thing that reprapers strive for in their design efforts.

A2

Quote
nicholas.seward
The longest parts are 175mm from end to end. I just finished the new arms. I will be uploading TODAY!

Which new arms Nicholas? did you improved the stringing?

I have the two terminating ends attach on the back of the motor arm. A spring will attach to the protruding bolt head. Stringing should be much simpler and less nerve racking. We will only have to do one loop between a pair of bolts.

*Better access to the termination of the strings
*I enlarged the clearance for the pulley. (You can't see that from this view.)
*I decreased from 1.5 loops to 1 loops. (Less stringing. Less confusion.)
*This setup is perfectly proportional.
*Double herringbone gears for slightly better registration. (I never had a problem with the single herringbone gears but these will be better.)

Quote
GeorgeH
That's terrifyingly large, but I think I can print it if I do it sideways.

Whoo! I'll be printing ASAP then!

One recommendation, having printed the first version of the arms: even if you're using PLA, you need a heated bed. And it should be a stiff bed, like a fairly substantial piece of glass, not something that can flex (ie. don't print right on your PC board heater). The end without the gear is going to try to curl, and it will pull the blue tape right off, or warp your bed if it's flexible enough.

Quote
nicholas.seward
[attachment 21846 newarm.jpg]
I have the two terminating ends attach on the back of the motor arm. A spring will attach...

Looks good!

Didn't you have interference problems with the center of the herringbones? Once you told me to make a channel there instead.

@Guizmo: You only need a split between the two sides if you are going to use the same arm blank for both sides. The gears are off by 1/2 a tooth from side to side but there is a 10mm gap.

ConceptFORGE's GIT Repository

GUS's STLs have been published!

Edited 2 time(s). Last edit at 11/27/2013 01:46AM by nicholas.seward.

YRS!!!! Can't wait to start working on a gus! Thank you Mr Seward!

Tks! Looking forward to seeing what the RepRap community does with your design!

Can RepRap software slice STEP files?

To import Inventor .ipt files into SolidWorks:
[help.solidworks.com]

Just posted the git link to Reddit and to #reprap IRC.

Shane

Wow!

Thanks! What about the firmware? is it working properly now?

Ps: You even published the Inventor files! Great!

Edited 1 time(s). Last edit at 11/27/2013 11:07AM by Guizmo.

FYI here's the number of parts to print: [dl.dropboxusercontent.com]

I'm going to slice them tonight and report back on filament amounts and time to print like I did before with Wally.