ROS Discussion Group, Tonight! Tuesday, Oct 21 @ 7pm PT

8 views
Skip to first unread message

camp .

unread,
Oct 21, 2025, 4:52:46 PM (2 days ago) Oct 21
to HomeBrew Robotics Club
    It's Tuesday again! Whether you're a beginner or an expert, please join us if you're interested in ROS.  :-] 

Join Zoom Meeting
https://us06web.zoom.us/j/88983478865?pwd=bRdjh1GJpslwHJBANe46VCTlSewyNy.1
Meeting ID: 889 8347 8865, Passcode: 935511

Thanks,
Camp

camp .

unread,
Oct 21, 2025, 9:07:14 PM (2 days ago) Oct 21
to hbrob...@googlegroups.com
Tonight! - cp
--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/2132625130.5412080.1761079952851%40mail.yahoo.com.

Sergei Grichine

unread,
Oct 22, 2025, 2:41:57 PM (yesterday) Oct 22
to hbrob...@googlegroups.com
Here is a follow-up for my demo yesterday (sorry for jerky/lagging video) - the photo album link in notes below leads to a good one.


The idea of the project is to understand BLDC motor wheels (Ninebot Segway miniPRO, hoverboards) and create a universal ROS2 "base" (articubot compatible, of course ;-)).
So far, the base works great, moves precisely and has minimal current consumption (12V ~0.5 A idle, close to 1 A normal driving).

SimpleFOC library has means of monitoring currents and speeds. There are no traditional encoders, Hall sensors are used for velocity control feedback. I am not sure that available rotational measurements can be easily used for ROS2 odometry - I might have to use LIDAR odometry package instead.



--
Best Regards,
-- Sergei

Chris Albertson

unread,
Oct 22, 2025, 3:48:32 PM (yesterday) Oct 22
to hbrob...@googlegroups.com

On Oct 22, 2025, at 11:41 AM, Sergei Grichine <vital...@gmail.com> wrote:

Here is a follow-up for my demo yesterday (sorry for jerky/lagging video) - the photo album link in notes below leads to a good one.



The idea of the project is to understand BLDC motor wheels (Ninebot Segway miniPRO, hoverboards) and create a universal ROS2 "base" (articubot compatible, of course ;-)).
So far, the base works great, moves precisely and has minimal current consumption (12V ~0.5 A idle, close to 1 A normal driving).

SimpleFOC library has means of monitoring currents and speeds. There are no traditional encoders, Hall sensors are used for velocity control feedback. I am not sure that available rotational measurements can be easily used for ROS2 odometry - I might have to use LIDAR odometry package instead.

SimpleFOC.  Does have support of all kinds of encoders, quadrature and serial as well as the built-in Hall type.     You will actually get better performance from the motor if you use a better encoder with at least 1,000 counts per revolution.   

I have a few test stands that are just a motor and an encoder.   I use them to test various motors.     I have some larger drone motors and some tiny gimbal motors.    While you can of course get a motor to spin using just the Hall effect sensors, you really can't do good position servoing like you would need with a robot arm or leg.     In the case of an arm or legs, we care about performance at zero or near zero RPM.   Good control at zero RPM is nearly impossible with DC gear motors but pretty easy with SimpleFOC.   

Hall works because you are spinning at near constant RPM and getting only a few “ticks” per rev, but at constant RPM, interpolation works well. Interpolation will not work well with non-uniform speed or near zero speed.   If cost and space are an issue, AS5600 magnetic encoders are very good.  Just attach the magnet to the shaft end with CA glue, and a 1mm air gap works.

Hall sensors were invented so you could drive a BLDC motor with no computer, you just wire the hall to a MOSFET gate and drive the motor with square waves.  But FOC wants better control of the waveform.   

Best practice is to use two encoders, one on the motor and the other after any gear reductions of the wheel or arm.   Then you use a double PID controller.    This will handle mechanical play and flex.  



Sergei Grichine

unread,
Oct 22, 2025, 7:36:33 PM (21 hours ago) Oct 22
to hbrob...@googlegroups.com
Chris,

"Just attach the magnet to the shaft end with CA glue, and a 1mm air gap works." - would you elaborate on your suggestion, given the actual design of the wheel motor? 


image.png
image.png


--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.

Alex Sy

unread,
Oct 22, 2025, 9:17:11 PM (20 hours ago) Oct 22
to hbrob...@googlegroups.com
The wheel has the 3 wires which is the BLDC motor coils and the small connector should have 5 pins (+5V, GND, U, V, W)  The U, V, W are the hall effect output and there is no standard for the wire color code but if you have it's driver you can trace or measure it out. 
Like what Chris said, those have low resolution, ie 4 cyles per rotation since these wheels were meant for hoverboards and are for speed control.   You could glue the diametrical magnet on the center of the outside of the wheel and have the sensor (AS5600, MT6701 or MA730) mounted on a bracket to sense the magnet rotation.  The MA730 will give you very high resolution.  The AS5600 will have a hysteresis jump around the 358 to 2 degree transition and is relatively slow (meant for electronic pots).  It is quite easy to get MT6701 on Ebay already on a board.
Note that although these chips all say they have 12 to 14 bits (4096 to 16384) per revolution, in reality they are all only specified to 1 degree, which may be a problem for accurate arms even if they are sufficiently rigid.  They are perfectly fine for dogs and humanoid legs.  Optical encoders are going to be the most accurate but expensive.
Alex
image.png
image.png
robots_bringup.png

Michael Wimble

unread,
Oct 22, 2025, 10:07:47 PM (19 hours ago) Oct 22
to hbrob...@googlegroups.com
If you want odometry, you could use something I just saw on YouTube from the FRC society, a dead wheel odometry device:


On Oct 22, 2025, at 6:16 PM, 'Alex Sy' via HomeBrew Robotics Club <hbrob...@googlegroups.com> wrote:

The wheel has the 3 wires which is the BLDC motor coils and the small connector should have 5 pins (+5V, GND, U, V, W)  The U, V, W are the hall effect output and there is no standard for the wire color code but if you have it's driver you can trace or measure it out.  
Like what Chris said, those have low resolution, ie 4 cyles per rotation since these wheels were meant for hoverboards and are for speed control.   You could glue the diametrical magnet on the center of the outside of the wheel and have the sensor (AS5600, MT6701 or MA730) mounted on a bracket to sense the magnet rotation.  The MA730 will give you very high resolution.  The AS5600 will have a hysteresis jump around the 358 to 2 degree transition and is relatively slow (meant for electronic pots).  It is quite easy to get MT6701 on Ebay already on a board.
Note that although these chips all say they have 12 to 14 bits (4096 to 16384) per revolution, in reality they are all only specified to 1 degree, which may be a problem for accurate arms even if they are sufficiently rigid.  They are perfectly fine for dogs and humanoid legs.  Optical encoders are going to be the most accurate but expensive.
Alex
----- Original Message -----
Sent: Wednesday, October 22, 2025 4:36 PM
Subject: Re: [HBRobotics] ROS Discussion Group, Tonight! Tuesday, Oct 21 @ 7pm PT

Chris,

"Just attach the magnet to the shaft end with CA glue, and a 1mm air gap works." - would you elaborate on your suggestion, given the actual design of the wheel motor? 


<image.png>
<image.png>


On Wed, Oct 22, 2025 at 2:48 PM Chris Albertson <alberts...@gmail.com> wrote:

On Oct 22, 2025, at 11:41 AM, Sergei Grichine <vital...@gmail.com> wrote:

Here is a follow-up for my demo yesterday (sorry for jerky/lagging video) - the photo album link in notes below leads to a good one.


Reply all
Reply to author
Forward
0 new messages