Helium Frog HF06 "SCARA" Robot

I have just begun a new SCARA robot design, which will be reprappable.

Heres the blog if you want to have a look and comment.

HF06 SCARA


I think it would be a useful addition to the reprap stable as it has features that the Mendel / Prusa doesn't have such as:-

Tool changing becomes relatively easy as the area around the tool head isn't cluttered with the machine frame.

A fixed work table which enables you to do research on powder printing etc. The reason is that you could easily add powder bins under the tooling arm.

The possibility of multiple arms working in one workspace.

It is more suited for assembly work as it has a much larger build area relative to its size. It could be 2 or 3 times the area.

Its going to use conventional reprap electronics and motors. I have done a few calculations and it looks like I can get the resolution in X and Y close to a Mendel, so my goal is that is can also self replicate.

Has anyone got experience of microstepping a 400 step per rev Nema motor?


As far as I am aware there hasn't been a SCARA robot completed by the reprap community, but if anyone knows of one could they post some links.

Edited 1 time(s). Last edit at 01/07/2012 03:22PM by martinprice2004.

There is nothing special about a 400 step motor that I am aware of so it should microstep between poles just as well as a similar price 200 step motor. There are special motors optimised for microstepping but they are expensive and presumably lower holding torque.

---

[www.hydraraptor.blogspot.com]

Just thought I would post an image of the SCARA build so far. Most of the parts are printed and the main frame assembled.



More details at Helium Frog website

Edited 1 time(s). Last edit at 02/11/2012 04:34PM by martinprice2004.

Very nice work, i love it. Further on, I think implementing software coordination might become a pain, but worth it in the end.

I have a question though. In this config the last arm actuator motor is placed on second arm, that makes it movable, and its heavy, which increases the load on arms. An alternative is to put it in a static position, and the only possible place like that is right in the middle of the second arm axis, in the picture looks like a M8 or M6 nut on top of a bearing. Because its also center of rotation of middle arm, it will be equidistant to the engaging pulley of the last arm also regardless of its position. Downside probably only 3 screws will affix the motor instead of 4. Wouldnt that be better, e.g. avoiding motor becoming a load itself.

Edit: On a second thought, could put in that place a doubled gear-reductor, and the motor can be far behind. And the other motor same thing with another double gear reduction then both motors can stand far behind and give assembly a better balance and center of gravity. What you think, worth considering?

Edited 2 time(s). Last edit at 02/14/2012 04:57AM by NoobMan.

@NoobMan

Yes the best place for the motor would be further back towards the frame above the first pivot as this would reduce the arm inertia. Unfortunately I couldn't do this as the motor would then clash with the top of the Z axis frame. I could move the whole assembly outward a little, but this has its own problems as it increases the cantilever loads on the linear bearings. I have seen SCARA robots which use two belts in series for the secondary arm motor and both motors are placed at the back of the frame , but this further complicates the design and firmware.

I have all the firmware calculations worked out and they are not too complex, but the software compensation in Z might be troublesome. My plan is to adjust the bed so the primary arm runs parallel and then just compensate for the secondary arm in software if necessary. I may do this by just having 2 values of z compensation one at max secondary arm angle one at min angle to start and just use a simple linear equation. Its a fudge, but might be good enough as the arm only moves about 120 degrees in my setup.

What i was thinking about is that's hard to tell from pictures by i was under impression if the right side motor is reversed one side up and to come down from there, the second arm motor could take roughly its place. But indeed it kind of complicates the build considerably, even would need a tensioner solution between the arm pulleys.
Nice to hear firmware is to be worked out. Would the firmware fairly modular, e.g. be usable for robots with different arms sizes, e.g. a parameter for arm lengths? I'd also contemplate making some mechanics someday, if i can find some feasable firmware to work out.

Yes the firmware is modular. It will be based on the delta robot firmware on my site here which is very basic but does work. It will just need a change of the equations in the delta routine to get it working. Ill also work on a few extra G codes and maybe add acceleration. The SCARA robot formulas are actually easier because the Z axis is conventional cartesian configuration.

The delta firmware worked in a slightly different way from most of the reprap firmware as follows:-

Reprap firmware pre calculates the steps for the move and then increments them. Cartesian robots have constant steps (acceleration not considered) for linear moves so this is the best and fastest way to do it using Bresenhams algorithm. You could however accumulate timing errors for long moves.

Delta and SCARA robots do not have constant steps for linear moves as they move in arcs. I compute the expected position at each firmware cycle and adjust the actual position to match . Its more accurate, but requires more computing which could be a problem for very fast robots. They would still work, but the path becomes less accurate.

I'm sure there are better ways to do it, but that is what I have so far.

... if you're interested in other kinematic systems and especially the comparison between a 'normal' scara and a parallel scara setup with much higher positioning accuracy, you should read through the links in this post: [forums.reprap.org]

Or search the forum with 'parallel scara' and 'all dates' ...

---

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

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

Thanks for the links Viktor

The parallel SCARA looks a better concept as the motors are mounted further back and it would be more rigid with two arms.

The only downside is the part count is a little higher. I think this one is worth considering as a possible replacement for the current prusa. With the use of a Sarrus linkage derived rising table, all linear bearings could be eliminated. Is it patented?

I did work on another concept a few months back and there is an image here if you want to have a look. It might also overcome the rigidity problem. It eliminates the elbow, but uses linear rails. My thoughts are to use a rising cantilevered table like the Ultimaker and maybe wall mount it. All motors would be fixed. I may possibly split the linear rails and offset them in Z to make the machine a little more compact.

What would you call that machine.....A delta cartesian hybrid?.....Cart Delta....!

Edited 1 time(s). Last edit at 02/14/2012 03:16PM by martinprice2004.

... there are some other parallel kinematics with different counts of linear or angular drives.

Look at the atached images in this post: [forums.reprap.org]
- the links are dead, but i've saved the images.

We had some more talk about this and other sorts of kinematics - simpler in buildup than the cartesian frame, but more complex in software ...

---

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

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

Just uploaded a video of the robot on YouTube

SCARA Robot Video Link

I havent implemented any acceleration yet, but it is capable of moving at 100mm/sec in X and Y without any problems. Z axis is very poor at only 5mm/sec. I might try to implement a belt driven Z axis or try a larger motor.

... looks fine

Have you measured the stiffness of the arm structure?

---

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

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

Look very nice, I make a project like you now. Can you post your Scara Code please?

Thanks a lot

Michael

It looks like development on the SCARA is on hold at the moment as I work away from home at the moment. I will put the code up on my website as soon as I can and also post all the models.

All details including 3dmodels, partslist and firmware has now been uploaded. They are avaialble at my site or on thingiverse.

Thingiverse

Helium Frog SCARA

Thanks you very much for your Code. I will update my project as soon as possible.

@victor

With reference to the hexapod design debate, It also got me thinking that modifying the SCARA to a parallel arm design might be an improvement. You suggested this configuration before, but maybe using RC servos might be an option.



The resolution could be an issue, but I did spot a hexapod milling robot which does seem to produce relatively detailed shapes.

Hexapod Milling

Perhaps a future project.

... yes, the parallel scara setup is much more precise (look at the images in the text comparing single-scara vs. parallel).

Matt's 'milling hexapod' has an 'averaging' resolution, as the actual position is an combined average of all six legs - so likely more precise, than a two-leg-scara made with the same servos could be.

What you can achieve with servos, can be calculated - simply divide the max. outer border line through the step-count the servos will perform when turning the possible angle. This will be the coarsest resolution - when drawing near to the center, it will be accordingly finer.

I have super-fine resolution steppers with controllers, that will perform either 10000 or 51200 steps per revolution, what's much better, compared to common servos ...

---

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

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

I can see the averaging using several legs improves things.

I wonder whether there is a plane horizontal configuration of say 4 servos in a line with multi links that would do a similar thing.

... look at ABB's 4-arm-delta - you have 4 motors, 4 arms and 4 (8) 'passive' joints positioning the too-head in the forced averaged position in 3D.

In a plane you'll have a slightly higher accuracy,than with two motors, but your frame has to be big enough to hold all motors.

Look at 'Jonis toy': [wiki.linuxcnc.org]

Here you have 3 strings moving the ring around - you can replace the strings with arms, that have a central joint, this will add a Z-constrain.

How much more precise this 3-arm setup is than a 2-arm, depends on the angular- or step-resolution of a single drive ...

---

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

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