Helium Frog Delta Robot

I am continuing the design of the Helium Frog Delta Robot and I have uploaded the latest models on the website. I am still working on the central tool platform. I'm going to put on a 3 bolt flange so the tooling can be swapped easily from the top by unbolting and lifting it off.

Initially I think I will design a simple pen holder so the robot will be able to draw diagrams on a flat sheet. This will be the best method of finalising the software . Drawing squares and circles is a good method of checking the logic is correct and also for measuring the accuracy of the mechanism.

Secondly I would like to add an extruder head, but which type? I think a stepper motor one would be nice, but which is the most reliable design.
The delta robot head is best if kept quite light in weight, so this should also be considered.

Helium Frog Website

Edited 1 time(s). Last edit at 10/26/2010 05:44AM by martinprice2004.

---

Helium Frog Website

Martin,

I was wondering two things:

Would you like to document your RepRap development in here at RepRap?
We actually have a documentation stub for some of your previous research:
http://reprap.org/wiki/WormBot-Deutsch
(via an image kicking around in the german language forum.)
It's probably more efficient and more fun this way.

And, secondly, I'd be honored if you join us in reprap-dev, reprap's open mailing list.
http://lists.reprap.org/mailman/listinfo/reprap-dev

Since you're a developer.

---

-Sebastien, RepRap.org library gnome.

---

Remember, you're all RepRap developers (once you've joined the super-secret developer mailing list), and the wiki, RepRap.org, [reprap.org] is for everyone and everything!

I have updated the Helium Frog Wiki Page and created a new development page for the helium frog delta. I'll add more details when I can.

Helium Frog Delta Robot Wiki

Why do the diagonal braces go to the mid points of the top bars? It will be much stiffer if they go vertex to vertex because any movement of a vertex would require one of the diagonals to stretch. The way you have it only requires the top bar to flex at its midpoint, which is many orders of magnitude less force than stretching it.

This is the same mistake as made on Mendel where the circuit board is supposed to brace it but doesn't as it doesn't go vertex to vertex.

---

[www.hydraraptor.blogspot.com]

Good insight. Another improvement that just occurred to me would be a sheet of polycarbonate/plywood/etc. mounted to the top as a reinforcement. May be unnecessary, but seems like it would improve rigidity.

Awesome design. The shelled out parts show you were incorporating printability/economics in to the design. I really like it.

nophead Wrote:
-------------------------------------------------------
> Why do the diagonal braces go to the mid points of
> the top bars? It will be much stiffer if they go
> vertex to vertex because any movement of a vertex
> would require one of the diagonals to stretch. The
> way you have it only requires the top bar to flex
> at its midpoint, which is many orders of magnitude
> less force than stretching it.
>
> This is the same mistake as made on Mendel where
> the circuit board is supposed to brace it but
> doesn't as it doesn't go vertex to vertex.

nophead Wrote:
-------------------------------------------------------
> Why do the diagonal braces go to the mid points of
> the top bars? It will be much stiffer if they go
> vertex to vertex because any movement of a vertex
> would require one of the diagonals to stretch. The
> way you have it only requires the top bar to flex
> at its midpoint, which is many orders of magnitude
> less force than stretching it.
>
> This is the same mistake as made on Mendel where
> the circuit board is supposed to brace it but
> doesn't as it doesn't go vertex to vertex.

Thank you for the feedback, please keep it coming as it helps me very much.

Yes you are quite correct in your assumption that two cross braces may be stronger and if I fastened them to each other at the middle they would certainly be stronger. The diagonal braces would be quite long and may not give the strength you expect, so a centre connection would be required I think. The top brace is really strong with the centre block, in fact there is only about 120mm of exposed bar each side of the centre portion so I'm not too worried about flex.

The assembly I settled on was done this way for two main reasons.

In order to make the assembly easy to adjust, the major issue I have is that the top triangle is twisted relative to the bottom one. Having the assembly that way it is easier to adjust out any twist. Setting the pitch between vertex is also simplified. It could be done with your configuration, but a lot more fiddly.

The other consideration is that as the delta robot moves around to its limits some of the arms would foul the bars if positioned as you suggest. The way I have them gives more room at the base where you need it. I checked this in CAD and this seemed the best way to get a 200mm diameter working envelope in the smallest footprint.

I did consider putting plywood on the back two faces to strengthen things, but this is going away from the reprap philosophy of self replication. Maybe when I get it up and running it will shake about so much I will have to rethink, but I just want to get it built and see how it works.

Edited 2 time(s). Last edit at 10/26/2010 05:15PM by martinprice2004.

... when using steel wires instead of bars you could get away much lighter and the fixing could be simpler too.

Even the arms can be made from a single bar in the middle with two balljoints at the ends and two wires at the position of the bars in the current design ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Martin,
A beautiful, elegant design. I like the way you have carried the Darwin/Mendel approach of smooth and threaded rod with RP end connectors for the structure rather than the more specialized and custom machined pieces I saw in a video clip of a different polar robot.

The first polar robot I saw working had the paired rods from each vertex aligned vertically, making a parallelogram to provide vertical stiffness to keep the extruder from leaning sideways. You have chosen to use side-by-side bars to provide the stiffness. I am not sure that it will work as well on a real world model, but it may be fine and does give a larger fraction of vertical usable space. However, it relies heavily on the rotational stiffness of the each vertex vertical. You have built the verticals from 2 bars, which is good, but one bar is a drive threaded rod, and is hard to get as stiffly mounted as the vertex-to-vertex bar. You would get more stiffness and keep the mount for the extruder more rigidly level if you used two vertical bars on the sides of each vertex, bolted as directly to the diagonals as possible. The diagonals not only stiffen the square of each side, but also reduce twisting where they connect. You should take advantage of that and run the two vertical bars in each corner from them.

You also have a lot of bearings on the inner most bar of each vertex. The stacked triple bearings make it very rigid vertically, but do not help with horizontal twist, which is what you center support needs. The existing stacked triple bearings would work great for the vertical pairing of support bars making the parallelogram. Or, if you switch to two side-by-side bars, you could use stacked pairs on each side and get the horizontal twist stiffness you need with only 2 more bearings per vertex. It would look something like this:

.................................____________________
................................|............................................ |
................................|......\\...............................//....| <= bearings
................................|........O..........................O.....| <= two structural bars
................................\......//...............................\\.../ <= bearings
..................................\......................................../
.....................................\................(O)............./ <= threaded drive rod
........................................\............................/
...........................................\__________/

Mike

Edited 3 time(s). Last edit at 10/27/2010 08:03PM by rocket_scientist.

Have you thought about using the delta to move the platform instead of the extruder? That way you could actually have multiple extruders in fixed positions, or you could have the multi-colour extruder that someone suggested in another thread that would require multiple steppers. That would be way too heavy to swing around with the delta, but if it was fixed, it would be no problem.

If you join the platform to the arms with two perpendicular hinges on two sides(one horizontal attached to the platform and one vertical attached to the first hinge), and a single horizontal hinge on the other, the platform will be constrained and level. I have been thinking about such a design, but with triangular side faces instead of rectangular (i.e. the top frame is 60 degrees rotated from the bottom frame). I was thinking threaded rod to push and pull the platform, but by rotating the nut rather than the rod (too much inertia in the rod).

Greg,
I find the concept of moving the table interesting. You have to be careful to let the hot plastic harden before you make any sudden moves, but the ability to have several fixed print heads sounds interesting.

If you rotate the nut instead of the rod, then you need to have the stepper motor mounted on the moving surface, which probably adds more mass than you save by not rotating the drive rod. And you would also likely need a gear to rotate the nut since it can not be directly shaft mounted and still have the threaded rod go through it.

Yes you need a gear, but they can be easily printed. The stepper motor would be "fixed" at the point where the rod, which is attached to the platform meets the top frame. It wouldn't actually be FIXED, because the rod will need to pivot, but the motor would stay mounted on the top frame and would not need to be accelerated with the platform (hence its weight is not an issue).

When I was investigating bevel gears for my version of the Heart Shaped Conical Gear Cluster, I did come across a spherical gear for an articulating toy. I should look for that again. With such a gear, it may be possible to fix the drive motor.

Edited 1 time(s). Last edit at 10/28/2010 03:40AM by Greg Frost.

Mike

The bearing arrangement you suggested is an interesting one, but on a reprapped machine probably wouldn't work. Your bearing arrangement means that the linear rails have to be perfectly aligned and spaced or the carriage mechanism will either be tight in places or go loose, it could also twist. If you look at the Mendel Reprap design the bearing arrangement uses one bar as the primary axis with 3 bearings and the other to stop twist (2 bearings). There is a video on youtube to explain this principle and it is a good one as the reprapped assembly is very inaccurate compared to conventional machine tools with machined slides. I made a cartesian robot (see website) with a two rail bush design and it suffered terribly from tight spots on the rails. You should avoid this bearing setup if at all possible. The setup you have seen probably used extruded aluminium or similar which is dimensionally very accurate (around 0.05mm or so).

Here is the Video.

Mendel Bearings

The carriages on my setup are allowed to rotate a little on the rails (The lead screw should accomodate a little mis alignment as the lead screw nuts are floating). I am hoping that the 6 bars are the constraint for the twist.

Greg

The inverted delta design is one I have considered, in fact there is a discussion on it on the German reprap forum.

Inverted Delta

I wouldn't worry too much about moving the part and having it collapse. I own a Makerbot which moves the part under the extruder. The inertia of the plastic part doesn't cause it to collapse. I wanted to have a static part as I will add a heated bed and probably use a silicone rubber heated mat. These are 240v supply so I wanted to keep the cabling to the mat fixed which is why I did it this way. I may try your idea as it would be easy to rebuild the Helium Frog Delta that way and it may mean that the lead screws could be shorter. I did think that 2 delta robots in a diamond configuration with a sliding table between them would be a neat design.

Edited 1 time(s). Last edit at 10/28/2010 05:15AM by martinprice2004.

True, I keep forgetting that it is easy to draw perfectly aligned rails, but not as easy to build them. I also wonder about adding X and Y offset screws to the top and bottom of one of each pair of vertical rails.If you get enough constraint with the design you that is great. If not, you have some ideas on how to add more.

If they are printed parts then I would make the holes under sized, clamp the parts together and ream them as a pair. Maybe add extra holes to do the clamping.

Edited 1 time(s). Last edit at 10/29/2010 07:22AM by nophead.

---

[www.hydraraptor.blogspot.com]

I've been looking at the way the base is attached to the printed parts. I don't have a printer yet so thinking about just mdf for the top and base leads to:

From Helium Frog

I have completed the assembly of the delta robot and it looks like the twist in the centre platform may be an issue.This is only controlled by the lead screws, if the lead screws were not there, the vertical carriages could rotate around the vertical rail which cause the centre tooling platform to tilt a little. The M8 lead screws are quite long and can flex, so I may have to add a second linear rail the other side of the lead screw and another two bearings to stop the twist. I think front to back rather than side to side rails as Mike suggested would work better in the design I have, but the principle is the same.

I think the robot would be OK for pick and place etc as it is now, but I think it needs improving before it could be used Rapid prototyping or running at high speed.

It might only need this modification to one corner, or if I go to M10 lead screws might not be necessary.

Arvin

Your MDF design sounds a good one, but I did see a reprap design made from flat sheet which was held together with tie wraps (used like stitching at the edges. It looked a really neat idea and I could imagine a triangular stricture made this way would be very rigid. It could be laser cut from plywood quite cheaply. You could also sew it together with nylon chord.

Edited 1 time(s). Last edit at 10/30/2010 12:08PM by martinprice2004.

Actually, if I were to make it out of steel I'd use some 1" wide strips and let the corner threaded bolts clamp them tight. Each corner would need at least one more connection to make them solid.

MDF is easier for me since cutting steel is a sweaty job with a hacksaw and I have a table saw, skilsaw, etc.

I work hard to make a job easier...does that mean I'm lazy? Maybe...

Work continues on the delta robot design, I am currently printing some new corner modules to incorporate an extra linear rail. This should eliminate the twist in the vertical carriages and improve the accuracy of the mechanism.

There is another concern with the design, The work space above the centre tool platform may be too small to get an extruder with motor in. It may have to be positioned some way above the tool platform to give sufficient clearance when the delta is fully articulated. I could use a fixed remote feed mechanism (Wade extruder) and down a tube to the heated nozzle.

Whilst thinking of this and looking on the net, there is a robot configuration where the delta is laid flat on one side (A bit like an old style "A" frame tent). This would overcome the issue and may also stiffen the design up as the base board now is bolted to the longer side of the frame. Two of the linear rails are low down on the base plate.

What do you think of this and has anyone had experience of using a remote wire feed at some distance from the heater nozzle?

You are referring to a Bowden Extruder.

Quite a few RepRappers are using this technology.
For the Huxley it is especially interesting.

---

Bob Morrison
Wörth am Rhein, Germany
"Luke, use the source!"
BLOG - PHOTOS - Thingiverse

Bowder extruder, yes that will probably work on the delta. It would benefit from a lighter tool head a well.

Bowden extruders do have significant issues with ooze because the process of releasing pressure by reversing the extruder has a lit more hysteresis. That, and the possibility of multiple extrusion heads/multi-colour extruder was what lead me to thinking of a stationary extruder with a moving platform.

I think you will probably limit the achievable print quality if you use bowden extruders (although clever design may be able to overcome its issues)

Yes I have been reading up on bowder extruders and hysteresis seems to be a drawaback, but most seem to be able to compensate in software for that. I think it will be sufficent for what I want on the HF Delta. Perhaps the best solution is a bowden extruder with a solenoid "pin" gate at the nozzle. Solenoids are lighter than a motor so you would still have a fairly light head.

I was having a route around the net and found a bowden extruder design with a Nema 23 motor. It drives directly on the filament without a gearbox. It had very few parts which is an attractive proposition.

Nema 23 Extruder

By the look of one of the pictures it also uses a splined insert to pinch the plastic filament, but its not mentioned in the parts list. I may try and use this with the off the shelf splined nut insert. This saves a lot of time on the lathe or machining the motor shaft. The insert as described by Adrian Bowyer here. but modified with a couple of grub screws etc.

Adrian Bowyer Spline insert

Edited 2 time(s). Last edit at 11/06/2010 08:26AM by martinprice2004.

Just a quick post to let you know that The Helium Frog Delta Robot has made its first baby steps.

The robot firmware has been written in basic form and it now accepts G1 and G92 codes only. This is enough to test out the delta algorithms. I have had the robot drawing a few square boxes and all seems OK. I'll try to post a video on Youtube in the next few days and get the models and firmware uploaded in case its of any use to someone.

The robot feed rate is quite slow using M8 lead screws, so I'm on the lookout for a cheap alternative. Maybe go to M12 or use toothed belt drive.

I used a stripped out version of Hydra MMM firmware and rewrote the linear move routine. I had some fun getting the stepper motor timing correct as its vital on a delta robot otherwise you get wonky lines!

The latest info and some pictures are at.

Helium Frog Delta Blog

Uploaded video on Youtube

Helium Frog Delta Robot - First Steps

Pretty cool!

---

Help improve the RepRap wiki! Just click "Edit" in the top-right corner of the page and start typing. Anyone can edit the wiki!

ACME rods as an option?

Is the Helium Frog Delta Robot still under active development? I haven't seen any updates in several months, but I really like the idea. So now I'm thinking about making a new derivative prototype with the following changes:

1. GT2 high torque timing belt (2 mm pitch) instead of threaded rod, to increase positioning speed.
2. Four vertical slides instead of three, to increase positioning accuracy.
3. Hollow carbon fiber diagonal rods from the kite shop, to reduce the weight of moving parts.
4. Laser-cut plywood for the top and bottom frame, to dramatically reduce the number of parts.

[www.thingiverse.com]

The Helium Frog Delta Robot is sort of still in development! I have nearly completed another design which uses vertical belts and has a few other improvements. The threaded rods were the worst idea on the delta, and belts would be much better.

I have also been working on another design, which uses two linear rails mounted horizontally with two arms in a triangular configuration for X and Y axis. The Z axis bieng vertical linear rails. Ill post some pictures if anyone is interested.

As the delta robot is huge, I might revise the design to a 150x150 equivalent envelope and make a slightly smaller desktop friendly machine.

Having successfully used LM8UU linear bearings on my prusa derivative, I would also like to swap to those on the delta as this simplifies things. The original delta has now been stripped down and the parts salvaged.

I am currently on another project (SCARA robot) but I hope to return to the delta robot soon.

The four columns in your design, sounds like one too many, but It might be good as the frame is large and if you want to achieve accuracy I think it might be worth doing it that way.

Don't worry about the mathematical equations for the firmware, Theres a copy on my website and I also have an excel spreadsheet which explains the equations, which are just trigenometry and pythagoras theorem. It wouldn't be too hard to modify them for any configuration.

martinprice2004 Wrote:
-------------------------------------------------------
> Ill
> post some pictures if anyone is interested.

That would be nice.