DUET3D rotary delta --> motor problems ?

Hi to all.
Am setting one rotary based delta robot with duet3d ( Ethernet ) board. Really good board with great web interface.
As test am using delta firmware from duet3d, but we are planning to add our rotary firmware ( maybe duet will release support in few months ? )

The delta robot that we are testing looks like this:
[ibb.co]

TECHNICAL STUFF:
Motors --> NEMA 23 double axis, 3A ( 23HS8630 ) , green/black , red/blue
Smaller arm length --> 30 cm , weigth 80 gramms
Longer arm length -->100 cm , weigth 60 gramms
Full weight of head and arms --> 1,5 kg
Power supply for motors --> 24V / 5 A

DUET CONFIGURATION
Power for motors limited to --> 2 A
Microstepping --> x16 ( interlaced )
Delta radius --> 200 mm
Steps per mm --> 80

PROBLEM:
I dont know if this is the problem regarding rotary setup or is this something else but it looks like that motors dont have enough power to lift robots arms.
NEMA23 motors should be able to lift this without problem ( even if there is no belt or lead screw ).

I made few tests and result is different every time.
E.g. if i dismantle all longer rods to reduce the weight i can see movement but the movement is not the same every time.
As soon i add smaller weight, one of the arms "breaks down"...

This is the video without load ( 45 grams )
[sendvid.com]

With small load ( 150 grams )
[sendvid.com]

Is this a problem with my configuration settings or just problem with rotary setup.
I found it hard to believe that nema23 can not lift cca 300 grams without belt or screw lead...
Specially when i look at this slider : [www.youtube.com]

Arnix

Edited 3 time(s). Last edit at 06/28/2018 05:27AM by Arnix.

Quote

Power supply for motors --> 24V / 5 A

I'm sure its partly responsible for the weak motors.

Hi O_lampe !

I can add stronger power supply but current will be still limited to 2A.
Are you suggesting to connect some external power supply ?

If so, how ?

Thank you for your fast reply !

Arnix

I could not see the video. It did not work for me. Anyway here are some thoughts.

The power source can be actually enough for small speeds. It will not be enough at high speeds.

You did not specify how much torque your stepper motors have. Based on a web search it is 1.89 Nm @ 3A. In your case it is only 1.26 Nm since you drive it with only 2 A. At the end of the short arm your motor can generate force of about 1.26 / 0.3 = 4.2 N. That corresponds to about 420 g. If your head has 1.5 kg there is no chance it could work when short arms are about horizontal. You need a stronger motor or a gear box ... or a lighter head ... and possibly also arms.

0.) Video... strange i can open it without problem...

1.) Is it possible to connect external power source to motors ? So i dont need to be limited to DUET's 2A supply ?

2.) hercek, thank you for calculation. If i add gearbox, how much would this slow down whole movement and on the other side,
will the power supply be enough for "implementing" this gearbox ?

3.) What other stepper motors would you guys suggest ( of course, the ones that i can use with DUET3D ) ?

4.) Btw. if i change x16 to x32 and i can move biceps ( screwdriver is attached to it ), without problem.

After that, i have try again with rods but it's just too weak...

Arnix

Edited 1 time(s). Last edit at 06/28/2018 05:09PM by Arnix.

Your gear box needs to slow down the movement at least by a factor of 4 but you may want to go to a factor of 10 or so.
The factor of at least 4 is needed just to move your 1.5 kg heavy arms/head. (1.5 kg sounds more heavy than what the picture indicates though.) You probably want to slow down the movement also to get higher resolution.

Your power source should be enough if you use it only to power the three stepper motors and you use it only somewhere below 5/6 of the maximum possible speed. The maximum possible speed depends especially on stepper inductance and resistance and stepper driver loses. In other words, I cannot tell you what is your maximum speed and acceleration without quite a bit of computing. You can just use the current power source at low speed and then increase speed till it starts to skip steps. Then you can consider stronger power source do get a bit more than 1/6 of additional speed ... well, I should rather say acceleration.

I do not really know answers to the other questions you have. I do not use duet. Maybe dc42 or somebody else will help you.

OK. Resistance of the motor is 1 Ohm.
Btw. if i change rotary to leadscrew design, do i still need to implement somekind of gearbox, to get higher torque ?
On the other side, vacuum pump needs to be implemented also, so this will add some weight to arms...
----------------------
Just for info, what kind of board are you using and does your board supports rotary setup ?

A.

Edited 2 time(s). Last edit at 06/29/2018 04:03AM by Arnix.

Lead screw typically means huge gearing to smaller speeds. You do not need gearbox if you change to lead screw.

I have a very old RAMPS v3. Support for rotary delta is a matter of firmware not a board. My old version of Repetier firmware does not support rotary delta (I do not know the status of the latest version).

I think i will stay on rotary design + planetary or worm reduction.

Support for rotary delta is a matter of firmware not a board
*** i agree with you. I meant that you have some out of box and ready to use solution...

Am trying to understand what is happening when motor lifts some weight.
To be more exact, in last tests ( without reductor ) holding torque was quite low and motor had problems lifting screwdriver.
So what is happening between steps ? Isnt the board (any board) holding torque in between steps ?
Am asking this because when i was trying to move one arm,in some smaller steps, arm was always falling down.

The weight of the arm with this screwdriver is definitely under 350gram, so i guess that the main problem was as O_lampe stated, power supply ?
I have ordered 24V 10A PSU so this should solve the problem....

Btw. when we see those big industrial high speed delta robots, what do you think, what kind of approach are they using ?
Servos, steppers, some other solutions ?


Arnix

Edited 1 time(s). Last edit at 06/29/2018 05:52PM by Arnix.

Quote
Arnix
Isnt the board (any board) holding torque in between steps ?

It is. The stepper will hold with with torque 1.26 Nm at worst at the nearest full step boundary.
You have some other problem as well if it does not hold the screwdriver.

Quote
Arnix
The weight of the arm with this screwdriver is definitely under 350gram, so i guess that the main problem was as O_lampe stated, power supply ?

Your current power supply should work at low speed. But maybe it is broken. You can verify it by connecting voltmeter to the power supply when it is loaded (the steppers are consuming current). If the power supply voltage is below 24 V in such a situation then it is broken. The point is that when the steppers are not moving then one stepper should take only around 250 mA from your power supply despite the fact that the current in the stepper phase will be 2 A. That is because you actually need only around 3 V to push 2 A through your stepper when it is not moving. The exact numbers depend on your stepper drivers.

A few suggestions:

1. By default, RepRapFirmware reduces the motor current to 30% of normal after 30 seconds of inactivity. This may be why the arms kept falling down. You can adjust this setting in the M906 command.

2. Is there any room for you to install counterweights?

3. There is a calculator at reprapfirmware.org for working out the maximum motor speed you can get before torque starts to reduce. It needs to know the steps/mm, which is not so easy to work out for a rotary delta.

HTH David

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I can not thank you enough for all your support.

Before i write down my answers, please look at this robot : [www.youtube.com]
This is the speed that i want to archive but with bigger robot dimensions.
I hope that this can be done with stepper motors.

---------------------------------------------------------------------------------------
You have some other problem as well if it does not hold the screwdriver.
*** I was thinking about that too. But i think this has something to do with SW setup.

DC_42 is expert on DUET so maybe he can share some light on this:
If i run the program with x16 setup, hand is lifted and next second it drops down.
BUT if i run it with X32 option, screwdriver is lifted without problem.
Interesting part is that if i choose X64 , arm acts like with X16 option
I have tested with x + 10 steps.


If the power supply voltage is below 24 V in such a situation then it is broken.
The point is that when the steppers are not moving then one stepper should take only around 250 mA from your power supply despite the fact that the current in the stepper phase will be 2 A.
***I will check this as soon as possible. Robot is not near me because college is working on gearbox adaptation...

Is there any room for you to install counterweights?
*** I can do that to, but i would like to avoid this part. The robot should work 24/7 so any additional HW could lead to some HW problems. IMHO.

By default, RepRapFirmware reduces the motor current to 30% of normal after 30 seconds of inactivity
*** this part is working OK, but torque drops almost instantly after lift. Please look at text few lines before...

There is a calculator at reprapfirmware.org for working out the maximum motor speed you can get before torque starts to reduce. It needs to know the steps/mm, which is not so easy to work out for a rotary delta.
*** College of mine has found one web side with calculator for rotary based system. Can this help ?

[www.marginallyclever.com]

This is printscreen of actual robot dimensions :[ibb.co]

Rotary firmware DC_42:
*** I saw on duet forum that you will not be "fully "awaliable for next week or two. Can we hear us after vacation so you can help us implement rotary delta firmware ?

Best regards !

Any suggestion guys ?

The speed on the video is not high. You can do it with smaller motors than on the video if you provide gearing.

You can describe your robot with newton motion equations and compute the needed torque from motors or (if you are lazy to do it analytically) you can draw the whole thing as an assembly in a CAD and let the CAD simulate the dynamic forces. As for as the motor strength needed, it is not so much about speed but mostly about acceleration.

You probably will not get an answer to the question when some feature will be implemented in some open source firmware. Especially, if the work did not already start. Also I doubt dc42 has a rotary delta. If so then there is not a chance he would implement it. He would not be able to test it. But it is open source. You can give it a try yourself.

You are already asking very specific questions which will be hardly answered since we do not know the answers without analysing it in a more detail. And analysing it sounds like a work. It cannot be compared with idle chatting on a forum and pointing out obvious problems

It would be good if you can chat with somebody who is already building a rotary delta too. You would get more useful responses.

Hi guys !
I was playing with some HW stuff and became obsessed with it :-)
We implemented spring in the middle of robot and reduce the weight to 200 grams.
Now we can lift the arms and its enough for testing purposes..

Btw.power supply is still weak one (24V/5A) but it will be replaced in few days.
--------------------------------------------------------------------------
This is what i get when i run G0 X600 from GCode console
[sendvid.com] --> or if this one is not working then:
[tinypic.com] ( you need flash 9 for this )

When i run G0 Y 600 ( after G28 )
[sendvid.com] --> or if this one is not working then:
[tinypic.com] ( you need flash 9 for this )

When i run on G0 Z100
[sendvid.com] --> --> or if this one is not working then:
[tinypic.com] ( you need flash 9 for this )
---------------------------------------------------------------------------
I installed eclipse on my linux box and we have tested the main steps that we need to take before new FW implementation.

As some base for rotary based design we are planning to use this :

[github.com]

recompile it and use it on eclipse.. I know very little about that source, but it's open source and i guess it can be update for DUET3d needs ..

Of course, we can not build this without DC_42 or someone who has done this part, so any input is more then welcome.
In the meantime i have to find some vacuum pump :-) that can be used with duet.

Arnix