BrushlessinSeattle
1 mW
- Joined
- May 5, 2010
- Messages
- 12
It’s about time I started a build log for a project that I’ve been working on for a while.
I’ve read a number of the threads on this forum; it’s an amazing resource, thanks to everyone for sharing their knowledge.
This is a sinusoidal ESC driving an RC friction drive. The friction drive part is fairly basic. I’m using an Aeolian motor that I converted from Delta to Y, Kv is around 100. Its static mounted and always engaged. Not good, I know. The swing arm designs that I’ve read about on this forum are superior for sure. When I can find the time I’ll hinge the mount so the motor can be engaged/disengaged, but for this project my main focus is on the controller. Ive used trap controllers in the past. For RC builds they work fairly well but are inefficient and the high pitch whine drives me crazy. When I noticed people (and dogs) half a block away turning their heads to see what the noise was I knew I had to do something. Sinusoidal drives come with a lot of challenges but, for me, it was worth the effort. I can run this controller in trap mode or sinusoidal. Noise-wise it’s night and day, hornets vs. butterflies. The fets run cooler in sinusoidal and there’s less cogging at low RPMs. I wanted to keep this as simple as possible so that I could get it up and running quickly. It’s a bit of a kludge but works surprisingly well.
Im using a heavily modified Microchip demo board (pic33xx). The driver stage was weedy and had to go. I built my own using 3077s and 5181 drivers. Im experimenting with snubbers (thanks Lebowski). I added internal halls to the motor. Phase is set 6 times per ecycle with interpolation between hall signals. I realize this is crude but so far it’s working well. Sensorless would be a better way to go especially at higher RPMS. FOC is worth investigating too but efficiency gains seem marginal and its an additional layer of complexity. The generated waveforms are sinusoidal into a resistive load (test rig) but get a little misshapen when driving the motor. I’m using space vector mod, I think I’ll need to use a different approach in order to generate waves that more closely match the motors bemf (sinusoidish). I’ve implemented a very sensitive over current protection circuit – shuts off all fets fast! and latches. This has minimized fet smoking but avoiding false triggers has been a challenge.
At this point Im wondering where to go with this. Its an operational prototype, I use it daily and haven’t had too many problems. I could (and should) at least have some PCBs made. I'm going to make an aluminum case that will mount in the triangle. Software-wise there's plenty of room for improvement. I'm also going to implement some type of BMS. Its far from plug and play. The controller is configured and tuned to work with this particular motor. To make it usable with other motors would require config changes and there's no user interface (unless you have a pic programmer). I think here in Seattle there may be a small demand for custom builds, reliability would be an issue.
If anyone has insights or wants to collaborate to help take this to the next stage let me know. I have an electronics background but could use some help with coding.
Thanks for looking
BiS
I’ve read a number of the threads on this forum; it’s an amazing resource, thanks to everyone for sharing their knowledge.
This is a sinusoidal ESC driving an RC friction drive. The friction drive part is fairly basic. I’m using an Aeolian motor that I converted from Delta to Y, Kv is around 100. Its static mounted and always engaged. Not good, I know. The swing arm designs that I’ve read about on this forum are superior for sure. When I can find the time I’ll hinge the mount so the motor can be engaged/disengaged, but for this project my main focus is on the controller. Ive used trap controllers in the past. For RC builds they work fairly well but are inefficient and the high pitch whine drives me crazy. When I noticed people (and dogs) half a block away turning their heads to see what the noise was I knew I had to do something. Sinusoidal drives come with a lot of challenges but, for me, it was worth the effort. I can run this controller in trap mode or sinusoidal. Noise-wise it’s night and day, hornets vs. butterflies. The fets run cooler in sinusoidal and there’s less cogging at low RPMs. I wanted to keep this as simple as possible so that I could get it up and running quickly. It’s a bit of a kludge but works surprisingly well.
Im using a heavily modified Microchip demo board (pic33xx). The driver stage was weedy and had to go. I built my own using 3077s and 5181 drivers. Im experimenting with snubbers (thanks Lebowski). I added internal halls to the motor. Phase is set 6 times per ecycle with interpolation between hall signals. I realize this is crude but so far it’s working well. Sensorless would be a better way to go especially at higher RPMS. FOC is worth investigating too but efficiency gains seem marginal and its an additional layer of complexity. The generated waveforms are sinusoidal into a resistive load (test rig) but get a little misshapen when driving the motor. I’m using space vector mod, I think I’ll need to use a different approach in order to generate waves that more closely match the motors bemf (sinusoidish). I’ve implemented a very sensitive over current protection circuit – shuts off all fets fast! and latches. This has minimized fet smoking but avoiding false triggers has been a challenge.
At this point Im wondering where to go with this. Its an operational prototype, I use it daily and haven’t had too many problems. I could (and should) at least have some PCBs made. I'm going to make an aluminum case that will mount in the triangle. Software-wise there's plenty of room for improvement. I'm also going to implement some type of BMS. Its far from plug and play. The controller is configured and tuned to work with this particular motor. To make it usable with other motors would require config changes and there's no user interface (unless you have a pic programmer). I think here in Seattle there may be a small demand for custom builds, reliability would be an issue.
If anyone has insights or wants to collaborate to help take this to the next stage let me know. I have an electronics background but could use some help with coding.
Thanks for looking
BiS
