High performance motor controllers

Hey Reprappers,
I would just like to announce my project, called Odrive. It is an open source 2-axis brushless servomotor controller, meant to be used with cheap hobby motors.
Let me know if you are interested in development, or want to get your hands on some early hardware.
Cheers!

Wow, that is scary fast!

I'd like something like that for the pick and place machine I am building.

Edited 1 time(s). Last edit at 05/08/2016 02:02PM by bobc.

---

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

Awesome! This project was in fact envisioned with pick and place in mind (:

Wow! That's stunning. And the lipo power reservoir is brilliant!

Quote
JamesK
Wow! That's stunning. And the lipo power reservoir is brilliant!

Haha thanks! ;D

Quote
JamesK
Wow! That's stunning. And the lipo power reservoir is brilliant!

Gotta disagree, seems like I'm always finding new ways to kill lipo cells. Can't you use something less fragile like supercaps?

Outstanding work! I am a bit concerned, however, that there are only two motor channels. I imagine most any application is going to need three channels. Pick and place probably wont need three channels, but I think you could hurt the marketability of the driver when it comes to CNC and 3D printing by only being able to drive two axes. I think it either needs to be offered in a single channel, a three channel, or a four channel. Maybe there is something that I am not following, but thats the way I see it. Again, great work.

Good luck. I will be following this one closely.

Quote
rmlrn
Gotta disagree, seems like I'm always finding new ways to kill lipo cells. Can't you use something less fragile like supercaps?

Power density/cost on supercaps is still far behind lipo isn't it? I know it's a shame to have to include a short lifetime part like lipos, but I don't know what else you could substitute in there with anything like the power density.

Quote
DRobs86
Outstanding work! I am a bit concerned, however, that there are only two motor channels. I imagine most any application is going to need three channels. Pick and place probably wont need three channels, but I think you could hurt the marketability of the driver when it comes to CNC and 3D printing by only being able to drive two axes. I think it either needs to be offered in a single channel, a three channel, or a four channel.

I'm not sure, for most printers the Z axis and extruder have very different characteristics than XY. I could see a 2 channel product being very useful. You could make a killer coreXY with this.

Quote
madcowswe
It is an open source

But, where source is? Only general description at the link exists.

Quote
madcowswe
...energy regeneration...

recuperation - more suitable term. (Yes, it is funny remark considering my english).

Quote
madcowswe
...Li-Po ... variable voltage DC bus ... DC-DC converter ...

With LiFePO4 a 3%-stability achievable without additional converters.

Quote
rmlrn

Quote
JamesK
Wow! That's stunning. And the lipo power reservoir is brilliant!

Gotta disagree, seems like I'm always finding new ways to kill lipo cells. Can't you use something less fragile like supercaps?

Actually, that was the topic of my university project. Odrive can support the following topologies:

• Lipo on the bus, DC-DC for power supply.
• Power supply on the bus, LC circuit on the DC-DC (with the C being an ultra-capacitor bank).
• Ultracapactiors on the bus, DC-DC for power supply.

I looked at the cost and performance of all the options, and decided that the first one was best by far. But if your application differs, you can use any of the other topologies.
Cheers!

Quote
DRobs86
Outstanding work! I am a bit concerned, however, that there are only two motor channels. I imagine most any application is going to need three channels. Pick and place probably wont need three channels, but I think you could hurt the marketability of the driver when it comes to CNC and 3D printing by only being able to drive two axes. I think it either needs to be offered in a single channel, a three channel, or a four channel. Maybe there is something that I am not following, but thats the way I see it. Again, great work.

Good luck. I will be following this one closely.

I was considering both 1 axis and 3 axes. Actually, v2 was 3 axis. The main reason I went for 2 axis was cost. Many applications will need 2 powerful axes. But some will only need 1 axis, so you would be paying for a wasted axis. Some applications need more than 2 axes. But then you can just use 2 Odrives. They can share the DC bus, (so only 1 battery, only 1 DC-DC connected (or both if you want)), and they can coordinate moves over CAN.

Quote
Andrey_SSh

Quote
madcowswe
It is an open source

But, where source is? Only general description at the link exists.

v3 Firmware is here
v3 Board design is here
v2 Firmware is here
v2 FPGA logic is here
v2 Board design is here

The main reason it looks barren is because v3 firmware is only just getting off the ground, and I plan to convert the v3 board files from Altium Designer (very expensive software) to CircuitMaker, which is free. Once that is done, they will be linked to on the hackaday.io page.
Cheers!

Well that goes as fast as any extruder could possibly cope with. Very impressive.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

I can see your point. So the plan would be to use an o drive for x and y and use a standard stepper driver for Z (and E if on a printer application). I could see that. The speed is great, but just as important as that to me is tbe closed loop system. That is more of an issue in my applications on x and y so perhaps an o drive would be fitting. Where this could really shine from an economic perspective is applications where you would be using very hefty NEMA 23s or 34s.

Quote
DRobs86
I can see your point. So the plan would be to use an o drive for x and y and use a standard stepper driver for Z (and E if on a printer application). I could see that. The speed is great, but just as important as that to me is tbe closed loop system. That is more of an issue in my applications on x and y so perhaps an o drive would be fitting. Where this could really shine from an economic perspective is applications where you would be using very hefty NEMA 23s or 34s.

Yeah that works. But you can also use 2 Odrives. It depends, as you say, on the application.
Exactly, in high power applications is where it really makes a difference (:

Cheers!

Hey everyone,
There is now an update about board availability and the roadmap for the ODrive features on the project page.
If you are interested in getting an ODrive board, you should check it out.
Cheers!

The time has finally come for the first manufacturing run of ODrive v3.1. They are now on the way, and should arrive early to mid January.

At this stage, around 20 board kits are going out the people who signed up to the "Inital development" phase. They have not been all allocated yet, you can signup here: Link

Since the boars are going out to just a small group of early developers, I will have the time to personally get you up to speed with the codebase and help to get going with the hardware. Then, together, we can prepare some stuff that is a bit more stable and a bit more documented for when the alpha testing begins.

The cost for me to get this small batch of boards manufactured was $96 per board, so that is the amount I need to ask for a kit, plus shipping.

The kit involves basically everything seen in the above picture, and consists of:

• ODrive v3.1
• USB Programmer
• A set of the optional large gauge wire screw terminals
• A set of pin headers
• Some nylon standoffs

I hope that ODrive will be able to help you make an awesome robotics project, thank you so much for your contribution to helping people have access to open robotics hardware and software.

When you say cheap hobby motors, how much you talkin'.... i paid £5 for a tiny drone motor...a more substantial stepper not too much, more.

Quote
MechaBits
When you say cheap hobby motors, how much you talkin'.... i paid £5 for a tiny drone motor...a more substantial stepper not too much, more.

I went over some appropriate motors on this post: Link
You can also see it illustrated below:

Board looks great - congratulations on getting this far. Those are some serious connectors!

Quote
JamesK
Board looks great - congratulations on getting this far. Those are some serious connectors!

Thanks!

Hello, I've waited and hoped for a project like yours. Thanks for making it open source.

How heavy is a board+4 motors?

I'm in desperate need of lighter motor alternatives to drive the Hangprinter Project forwards. If you think that a oDriven Hangprinter a good idea I would like to work with you!

PS! See my blog for pics and description of Hangprinter.

---

torbjornludvigsen.com

Quote
tobben
Hello, I've waited and hoped for a project like yours. Thanks for making it open source.

How heavy is a board+4 motors?

I'm in desperate need of lighter motor alternatives to drive the Hangprinter Project forwards. If you think that a oDriven Hangprinter a good idea I would like to work with you!

PS! See my blog for pics and description of Hangprinter.

Hey tobben,
I checked out your blog, and wow that hangprinter is so cool!
I think that any appropriate brushless motor will beat steppers in terms of power-to-weight ratio by about factor 10! For instance the 350kv motor that is one of my favorites has a power-to-weight of 45W/g.
Have you considered having the motors on the anchors?

I would be very happy to support this project!

Cheers,
Oskar

Hi, thanks!

Those motors look bad-ass. I guess 1/10'th of their torque is enough for Hangprinter. The torque requirement is only ca 5 Ncm at 2-3 revs/s.

The reason for the single-unit design is the possibility to park such a unit in the ceiling. Without discreet parking I wouldn't get my wife's blessing to install it in the kitchen. I
really wish I had a dedicated lab or workshop, but alas, what is engineering if not working around constraints.

With ~20 g motors, we would save 700 g compared to the current Hangprinter. Anchor-mounting motors would save an additional 100 g, but at a high cost in terms of engineering hours per machine. It would be worth it if print speed requirement was > ca 1 g/s, but it would be a very different machine than the current one.

I had BLDC motors in mind when I first designed Hangprinter, but I lacked the motor control knowledge and found no off-the-shelf BLDC solutions.

I saw on your blog (oh, and in this thread) that you've had a board production run. If it's possible for me to convert my little Marlin code and test BLDC as an (almost) drop-in replacement for my steppers, then I want one fast =D Bulding/soldering on my own is OK if you don't have any boards left. Using other types of commands than step/dir is also OK but will require more time and planning.

What motors do you recommend me to order for experimentation?
Do the step/direction input waste much power compared to other inputs?

thanks,

Edited 2 time(s). Last edit at 12/20/2016 10:27AM by tobben.

---

torbjornludvigsen.com

Quote
tobben
Hi, thanks!

Those motors look bad-ass. I guess 1/10'th of their torque is enough for Hangprinter. The torque requirement is only ca 5 Ncm at 2-3 revs/s.

The reason for the single-unit design is the possibility to park such a unit in the ceiling. Without discreet parking I wouldn't get my wife's blessing to install it in the kitchen. I
really wish I had a dedicated lab or workshop, but alas, what is engineering if not working around constraints.

With ~20 g motors, we would save 700 g compared to the current Hangprinter. Anchor-mounting motors would save an additional 100 g, but at a high cost in terms of engineering hours per machine. It would be worth it if print speed requirement was > ca 1 g/s, but it would be a very different machine than the current one.

I had BLDC motors in mind when I first designed Hangprinter, but I lacked the motor control knowledge and found no off-the-shelf BLDC solutions.

I saw on your blog (oh, and in this thread) that you've had a board production run. If it's possible for me to convert my little Marlin code and test BLDC as an (almost) drop-in replacement for my steppers, then I want one fast =D Bulding/soldering on my own is OK if you don't have any boards left. Using other types of commands than step/dir is also OK but will require more time and planning.

What motors do you recommend me to order for experimentation?
Do the step/direction input waste much power compared to other inputs?

thanks,

So smaller brushless motors have a lot less torque, but spin faster. So they still pack some power, but you may need a reduction gearbox, reduction pulley and belt, or something similar. Or spool your lines on very small diameter pulleys, but that carries its own problems.
You made a good point about the fact that some magnetic encoders do have quadrature outputs, so are indeed compatible with the current ODrive version! So that should make for some light weight encoders.

The first interface that will be available will indeed be step/dir signals, so it should be okay to do a drop-in replacement. It has its limitations, but they are mostly to do with speed, and I think for most 3D printer applications it should be fine.
In terms of motors, it would depend on if you think it is acceptable to have a reduction gearbox?

Cheers,
Oskar

Quote
madcowswe
The first interface that will be available will indeed be step/dir signals, so it should be okay to do a drop-in replacement. It has its limitations, but they are mostly to do with speed, and I think for most 3D printer applications it should be fine.

Nice!

Quote
madcowswe
In terms of motors, it would depend on if you think it is acceptable to have a reduction gearbox?

A gearbox is acceptable. Hangprinter's motors experience force from one direction only, so play is not an issue. We should be able to gear down without adding too much weight. We can always make lines double, triple or even quadruple to gear down more.

What do you think about using worm drives on all axes?

I have little experience with gearing down. The worm driven D-axis on the current Hangprinter is limited to 30 mm/s speed because the stepper+driver stalls/resonates.

---

torbjornludvigsen.com

Quote
tobben
A gearbox is acceptable. Hangprinter's motors experience force from one direction only, so play is not an issue. We should be able to gear down without adding too much weight. We can always make lines double, triple or even quadruple to gear down more.

What do you think about using worm drives on all axes?

I have little experience with gearing down. The worm driven D-axis on the current Hangprinter is limited to 30 mm/s speed because the stepper+driver stalls/resonates.

Okay thats cool then!
I think quadruple lines is going to be good, and maybe have a single reduction stage onto as small a pulley as you can get away with. Basically, if you want to use the small brushless motors, as much reduction that you can use is good, but without adding too much friction.
I don't really know very much about what kind of gearing is the best here. I have used belts and pulleys mostly, but maybe nylon gears may also work. I don't know if the friction in worm gears will be worth the extra gear ratio they provide. Maybe someone else in this thread can also comment?

Cheers,
Oskar

I have a question for everyone: What is the fastest extruder/hot-end that you know of?
I guess the reason I am asking is fairly self explanatory if you've seen the video ;D
Cheers,
Oskar

Quote
madcowswe
I have a question for everyone: What is the fastest extruder/hot-end that you know of?
I guess the reason I am asking is fairly self explanatory if you've seen the video ;D
Cheers,
Oskar

Forget about the hotend, the viscosity of the molten filament makes this kind of speed a non starter.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]