Prototype Scara Robot Arm

Well even though I built this months a go I only managed to find time over the Xmas break to wire it up and give it a whirl.



Please find video for anyone interested in first run.

[www.youtube.com]

Cheers

Sean.

---

******************************
[www.cncdudez.co.uk]
******************************

Looks cool!

Can you describe how the joints are designed.
Do you think a NEMA 17 driven extruder could be mounted on the end effector?

Quote
CLaNZeR
Our marketing targets are the small business and home/hobby markets.
Our foundational goal is producing machines of high quality at an affordable price.

[www.ukcnc.info]

Are you planning on selling or sharing your design+software?

Hey Sean!

I remember you from the Steorn days

Looks like a lovely piece of work, as always. I'm already intrigued by the possibilities.

---

What is Open Source?
What is Open Source Hardware?
Open Source in a nutshell: the Four Freedoms
CC BY-NC is not an Open Source license

Hi A2

The Steppers have 50:1 planetary gearboxes attached which sit inside the lathed down aluminium joints.

These sit in bit flat bearings to keep it rigid, so very simple.

Still very much a work in progress and still not happy with the gearboxes durability as I have just stripped the gears inside again while playing around yesterday.
Also the planetary gearboxes have a little backlash, which I can compensate for in software but not ideal.

Would love to use Harmonic drives/gearheads but they are so expensive. Now working on using multi belt and pulley setup built into the arms.

Not sure what we way are going to go with this one yet as early play time.

Cheers

Sean.

---

******************************
[www.cncdudez.co.uk]
******************************

Hey Bobc !

Thanks and will push out more details as it develops.

If we run the stepper drivers at 16th micro step then that gives us 3200 steps per revolution on the motors.
So if we did direct drive then this would only give us 0.1125mm resolution.

By adding the 50:1 gearboxes to the steppers we now get 0.005mm resolution which now gives it possibilities such as 3D printing, Icing adapters not to mention pick and place etc etc.

Another goal was to get the speeds up and RPM of the Stepper motors.
usually a stepper motor, even with good ramping will top out at 1000 RPM, where a new circuit we have been working on for the last year that uses low cost dedicated signal generators gets the stepper up to 5000RPM if needed. But of course you loose torque at these high speeds.

Cheers

Sean.

---

******************************
[www.cncdudez.co.uk]
******************************

Hi

The first thought looking at the picture was "nice crane". Then i realized that i only saw a half of this picture. But then the ideas came flowing. And yes it really looks like a crane.

First thing i would test is to rotate the whole crane and eliminate one stepper on the arm. Second i put the other stepper in the tower and drive the axle with a pulley reducing the need for the gearbox and weight but leaving a mechanical advantage in the range of 1:5 to 1:10. That said the math looks a bit more complicate than now and the singularity is near the tower.

Just 2 cents

Edited 1 time(s). Last edit at 01/10/2014 08:58AM by magicworx.

Yep does look like a crane

But the reason behind the skinny arms was to get as much movement as I could and not restrict it to just 180 degree arc.

This will probably change with the next design as the arms are going to be beefed up to take the new gear system I have in mind.

If the arms are strong enough and the joints mounted on wide bearings then having the weight of the motors on the arms should not cause too much flex.
If you check the you tube video link on first post, there is not much flex going on, even on rapid movements and that is with the skinny arms.

If you look at commercial Scara Arms, most have the motors on the arms.

Cheers

Sean.

Hi

I am sorry. I wanted to loose weight on the arm to be more flexible with the weight of the hotend and stuff. I believe that if you want to print faster you need a very rigid arm. To reduce the needed power to overcome the mass of the arm i prefer to go as light as possible. And yes i like the design of your skinny arm. But back to the 180 degree arc. If you turn the whole tower you got 360. Put the joint approx a bit further than half of the armlength and you got a very big build volume.

Yes a heated bed is kind of funny looking. But these are only ideas and i can't wait to see your upgraded design. So long

Magicworx

Sorry but my brain just screams "NOT STIFF" when I look at your otherwise very nicely built machine. The center tower is rectangular and not torsionally stiff. The arms are skinny in the load direction so are not stiff. I think you'd be far better off using two pieces of rectangular aluminum tubing with the wide side horizontal. Use a single crossed roller bearings at the joints between the two tubes, light and stiff. With the tubes offset from one another you will have a large travel even with fat arms. You can cut triangular openings on all sides of the tubes to reduce weight with very little loss of stiffness. Take a look at an Adept 1. Saw one run once, scary fast and accurate.

Actually it is not that bad. We loaded it up with 5 kilo weight to see what the torque was like on the gearboxes and it moved it fine. We had to bolt the base down tight to a bench, but was very surprised.
Of course it did bend, but then a Hot end does not weigh 5 kilos does it

But agree on some of the points and spent this weekend drawing up a new design that does away with the crappy gearboxes. The speeds we want just strip the cogs inside, even with ramping and the backlash is a pain.
So new design has been beefed up and gearboxes replaced with belt pulley system to get the resolution we want.

I think you are a bit optimistic comparing this machine to a commercial machine such as the Adept Scara LOL !

I want to make a machine for a couple hundred quid, not spend £K?????
I am only a small business and cannot afford luxuries such as that. This play time for me.

This machine has been a good first stage as it has at least helped getting the software right and the Inverse Kinematics calculations sorted and proved as such.

Will post pictures of Version 2.0 soon. But it will look very different

Thanks for the feedback, to me this is what prototyping is about, get the idea, build it and make tweaks or scrap it and back to the drawing board !
More fun than getting it right first time !

Cheers

Sean.

Edited 1 time(s). Last edit at 01/12/2014 04:53PM by CLaNZeR.

---

******************************
[www.cncdudez.co.uk]
******************************

I found a similar design that maybe helps out with your next design

Yep have seen that one before and nice build but very low resolution.

With the version1 I made above it has 50:1 gearboxes and the steppers were driven with micro stepping at 1/16th (3200) pulses per rev.

That gives the arm movement of 0.00225 degrees which equates to a resolution of 0.007mm

But the plus side was we are able to run the steppers motors at silly speeds which also gives us the rapid movements as seen in the video.

The version2 I am now working on has a gearing of 37.04:1 which again with micro stepping at 1/16 will give us 0.01mm which should be good enough.

---

******************************
[www.cncdudez.co.uk]
******************************

Well it has been nearly 4 months and in the background the Scara solution is actually coming together.

We ditched the Planetary Gearboxes after getting through so many, even with gentle ramping they cannot hold up of running for a few hours without a tooth getting stripped.

MK2 ended up being a 3 stage gear combination of T2.5mm pulleys and belts that were built into the arm and actually worked very well, but even with 3 stage drives it only takes one belt to wear and backlash gets introduced to the system.

I am really pleased to say that we have now sourced a supplier that can make us harmonic drives at a reasonable price. But we do have to order in quantity.
At the moment we have some 50:1 test gear drives being made to our specification and should be with us in a few weeks for testing.

But meanwhile we have sourced some 30:1 second hand harmonic gear drives which in testing are working out really good.
So MK3 has been drawn up this week with the 30:1 harmonic drives being the centre piece, but also accommodating the new 50:1 drives we have being made.

Will get some pictures done as the build progresses.

The biggest headache have been the firmware and software.

To get the resolution we have got over the mechanical hurdle, but now we are onto the inverse kinematics calcs to convert Cartesian moves into radial movements with interpolation.
Not a problem host side doing all the calculations, but a bottleneck with the USB stack and 10-100ms delays between packets.

Moving the calcs into firmware and wanting to get the resolution we want, I am not sure if the Atmel/Arduino hardware is going to be the solution, I just do not see it being powerful enough to do these calcs without us dropping the resolution.

So we are now split down two lines at the moment.

1.) We approach with custom hardware/firmware platform that will need host software written for it.

2.) Go for a established hardware platform that is man enough to do the IK calculations and the interpolation needed and can use existing packages.

Well number 1 is already being worked on.

But number 2 we have came across the SmoothieBoard [smoothieware.org]
This really nice open source controller/firmware on first glances seems to fit the bill. But it does not support Scara at the moment. But being based on the GRBL gcode engine seems ideal.

The communications with the guys from Smoothie are very positive and it does seem that Scara profile and configuration is something that the Smoothie is man enough processor wise and also it will work with existing hosting software for 3D printers.
But at the moment it is not implemented but it is possible.
Profiles for other delta machines such as Rostock and H-Bot have been implemented with success.

Meanwhile I see these guys are a planned launch for September which looks exciting
[www.fluxintegration.com]

Good luck to them and I see they have gone down the dedicated route with regards firmware/software.


Cheers

Sean.

---

******************************
[www.cncdudez.co.uk]
******************************

The FLX.ARM.S16 looks really interesting, but personally I like your design better with the arm and end effector moving up and down. That enables a larger workspace around the arm, if the print bed doesn't have to move.

Hmm I just got this crazy idea: I wonder if it's possible to design a scara arm that has both motor for primary rotation and up/down axis on the bottom. Rotate the whole tower at the base and have a pulley for up down movement. And have the third motor near the central axis and use a pulley to rotate the arm.

Please be gentle. My first cad design with OpenSCAD

Hi dejay

My thoughts enterly. Good drawing by the way. Put another bar from top to the edge of the first horizontal bar to improve rigidity and you can put on a drill at the end.....

Okay that's an alcohol influated idea...


Just want to tell.

Magicworx

Hi Dejay

Nice 3D drawing and idea.

Sorry to say with the MK3 prototype we have now gone for moving bed on the Z-Axis and keep the Scara Arm fixed up in the ait at around 300mm mark.

The reason was that we wanted to make the actual arm as stable as we could and take all means of flex out of it.

I have found time the last few days to get back to the MK3 and by the end of the week I will post some pictures.
It looks very different from the MK1 on the start of this thread and the harmonic gearboxes are great.

Today I was milling out a Bowden style extruder out of 20mm aluminium billet which makes it nice and sturdy inline with the new arm.
This will keep the weight off the arm also, even though I must admit the arm has turned out pretty beefy. But at least this way with MK4 we can look at thinning it down till we get a happy balance.

More in a couple of days.

Cheers

Sean.

---

******************************
[www.cncdudez.co.uk]
******************************

Quote
CLaNZeR
Sorry to say with the MK3 prototype we have now gone for moving bed on the Z-Axis and keep the Scara Arm fixed up in the ait at around 300mm mark.

Well that is probably the best way to create more rigidity. I'm really not sure if my rotating tower would work well since you'd probably need some "training wheels" to keep it stable. And of course you really don't need a 360 degree work space, although it would be cool and might be good for some applications. For stiffness you could go for two arms like the wally also.

Quote
magicworx
My thoughts enterly. Good drawing by the way. Put another bar from top to the edge of the first horizontal bar to improve rigidity and you can put on a drill at the end.....

Thanks, definitely add more triangles! I was thinking a bar and wheel to the bottom below the third motor, since you already have more space available at the bottom because of the extruder, and the overall bending force would be applied a bit lower on the vertical beam.

Found a bit of time to take some pictures today.

This is the MK3 Prototype with the 30:1 Harmonic gears



















Cheers

Sean.

---

******************************
[www.cncdudez.co.uk]
******************************

Wow looks really slick! Might look even better with the middle arm out of aluminium too, but it looks very clean and professional.

Is the harmonic drive something like this? Some pictures of it and how you made them would be really cool!

Hi Dejay

All the arms are made out of Aluminium. The black parts on the first arm printed lids to cover up the motors.

The Harmonic drives are not homemade ones but the commercial ones [harmonicdrive.de]

I picked up a couple for a good price on Ebay for testing. They are really expensive at £1K a piece new from Harmonic Drives AG

But since we have now sourced a supplier that can do them at reasonable prices and the two 50:1 samples should arrive next week.
These ones bolt directly onto Nema17 stepper motors so MK4 will look a little different with regards the arm design.

MK4 will also have thinner arms and a few other changes.

The MK3 now allows us to test our software routines and focus on that part for now.

Cheers

Sean.

---

******************************
[www.cncdudez.co.uk]
******************************

Well done! Is the software completely custom?

Hi Evil Monkey

Have been playing with Smoothie Board today which if it works out well would allow Open Source software to be used with the machine.

But next week I will be testing our own hardware/firmware as I think a dedicated controller is just going to make it easier to implement.

Will see how it pans out and update as we go along.

Cheers

Sean.

---

******************************
[www.cncdudez.co.uk]
******************************

I recently purchased two 50:1 stepper motors from a company called Phidgets out of Canada, they were around $45.00 a piece. I haven't tested them out yet, but they don't appear to have any backlash. Could this be a more cost effective solution to harmonic drives?

It says the backlash is 1.5 degrees. http://www.phidgets.com/products.php?category=23&product_id=3328_0 I wonder what it is in practice? Also, is that pre or post gearbox backlash?

---

ConceptFORGE
Wally, GUS Simpson, LISA Simpson, THOR Simpson, Sextupteron, CoreXZ

That is the same style of stepper+planetary gearbox we used in the MK1 arm at the top of this thread.

We lathed out some aluminium joints that had bearings top and bottom to take the pressure off the gearbox bearing.

1.5 degree backlash equates to a fair bit of movement by the time you put the load of your arms on them. The ones we tried did not have that much backlash but were more expensive.

With the Harmonic drives you are into how many arcs backlash and most are 2-3 arc- minutes (1/60th of a degree) in both directions. So nothing really at all.

With a 50:1 and 16th micro step as stated in a couple of posts above you are moving the motor 0.00225 degree per pulse and with two arms that have 200mm centres that comes out at 0.0078mm movement of the extruder hot end every pulse.
So as you can see 1.5 degree backlash will give a whopping 5mm potential backlash on both motors.
You can compensate for this in software if you know your exact back lash and it is constant.

But the next problem we found was that those planetary gearboxes have a max input speed and it does not take much to strip the small gears inside.

With the harmonic drives they are good for 3000 RPM input.

Cheers

Sean

---

******************************
[www.cncdudez.co.uk]
******************************

Good point. I intend to use them in a similar setup as the Wally. I'm going to see if 2 geared steppers opposed to each other can mitigate any backlash.

This project just popped up on 3ders.org