@gravis thanks for the feedback: I simply hope the current configuration is good enough for follow up experiments. Carl bugeja has better magnets. He uses a metal encasing and 10 magnets see https://youtu.be/zeGKYIcLMIs?t=1186. He ordered this via ali express. The metal encasing enhances the field, this is known from pot magnets. He indeed seems to use a triangular style segment. So probably my design is sub optimal. The QFN was a pain at the start, but I can do it by hand known. I can't handle current measurements at the FPGA level at the moment. Still, it's could to keep this in the back of my mind. I made a new design for the prism encasing, and will see in the coming weeks if the power from the electric motor is sufficient.
thanks!! reached out to him he seemed to have missed my work on pcb motors as well. His reply to the trapezoid is as follows;
To answer your question, we're not sure at the moment - but it does seem to be a popular choice. Our working theory at the moment is that with a circular coil you have to have the magnet aligned perfectly with the centre of the coil, but with the long trapezoid tracks you have a much wider magnetic field for the magnet to move across - but we're not sure to be honest.
I think moving to arc segments is a good idea but you also need to modify your coils to be trapezoidal (e.g. https://i.redd.it/jxrfg6pmdh611.jpg ) though if they are very tight they may look triangular (e.g. http://oldcomputer.info/media/disk35/motor.jpg ).
I've looked into arc segment magnets before and you have to pay close attention to all the parameters, especially the direction of magnetization.
Visual of parameters: https://www.r4ymagnetics.com/pictures/TWqym3.jpg
Visual of magnetization: https://www.r4ymagnetics.com/pictures/A7vWKH.jpg
Other than that, sometimes they measure using diameter instead of radius and you should definitely do a quick render in OpenSCAD to make sure it is as you expect as it's common for images to be non-representative of the product. OpenSCAD doesn't have built-in arc function but this is the easiest way to make them using the product parameters: https://www.xarg.org/snippet/circular-sector-and-arcs-with-openscad/
Also, it may be possible to use ferrite magnets instead of neodymium magnets. It depends on the strength of the coil output but I wouldn't discount the possibility without trying it because ferrite is dirt cheap in comparison.
It seems like you may not be utilizing the current sensing in the motor driver. This is something you should be utilizing as it can be used to determine the motor state.
It seems to me that you should ditch the motor driver and build your own. In addition to lowering costs, it would give you better control and feedback. The power requirements for your motor are so meager that it should only require a transistor to be a MOSFET driver. It will also make it easier to build the board because QFN is no fun and will result in higher failure rates.
Hexastorm
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.
@gravis thanks for the feedback:
I simply hope the current configuration is good enough for follow up experiments.
Carl bugeja has better magnets. He uses a metal encasing and 10 magnets see https://youtu.be/zeGKYIcLMIs?t=1186. He ordered this via ali express. The metal encasing enhances the field, this is known from pot magnets. He indeed seems to use a triangular style segment. So probably my design is sub optimal.
The QFN was a pain at the start, but I can do it by hand known. I can't handle current measurements at the FPGA level at the moment. Still, it's could to keep this in the back of my mind.
I made a new design for the prism encasing, and will see in the coming weeks if the power from the electric motor is sufficient.
Are you sure? yes | no
If you missed the coil generator on the HaD front page, here it is: https://github.com/atomic14/kicad-coil-plugins
Are you sure? yes | no
thanks!! reached out to him he seemed to have missed my work on pcb motors as well.
His reply to the trapezoid is as follows;
To answer your question, we're not sure at the moment - but it does seem to be a popular choice. Our working theory at the moment is that with a circular coil you have to have the magnet aligned perfectly with the centre of the coil, but with the long trapezoid tracks you have a much wider magnetic field for the magnet to move across - but we're not sure to be honest.
Are you sure? yes | no
I think moving to arc segments is a good idea but you also need to modify your coils to be trapezoidal (e.g. https://i.redd.it/jxrfg6pmdh611.jpg ) though if they are very tight they may look triangular (e.g. http://oldcomputer.info/media/disk35/motor.jpg ).
I've looked into arc segment magnets before and you have to pay close attention to all the parameters, especially the direction of magnetization.
Visual of parameters: https://www.r4ymagnetics.com/pictures/TWqym3.jpg
Visual of magnetization: https://www.r4ymagnetics.com/pictures/A7vWKH.jpg
Other than that, sometimes they measure using diameter instead of radius and you should definitely do a quick render in OpenSCAD to make sure it is as you expect as it's common for images to be non-representative of the product. OpenSCAD doesn't have built-in arc function but this is the easiest way to make them using the product parameters: https://www.xarg.org/snippet/circular-sector-and-arcs-with-openscad/
Also, it may be possible to use ferrite magnets instead of neodymium magnets. It depends on the strength of the coil output but I wouldn't discount the possibility without trying it because ferrite is dirt cheap in comparison.
Are you sure? yes | no
It seems like you may not be utilizing the current sensing in the motor driver. This is something you should be utilizing as it can be used to determine the motor state.
It seems to me that you should ditch the motor driver and build your own. In addition to lowering costs, it would give you better control and feedback. The power requirements for your motor are so meager that it should only require a transistor to be a MOSFET driver. It will also make it easier to build the board because QFN is no fun and will result in higher failure rates.
Are you sure? yes | no