Orange_Crush
1 µW
- Joined
- May 17, 2021
- Messages
- 4
Started this thread as a small experiment log/report, trying out an stm32-based m365 scooter controller with stancecoke's EBiCS firmware port on a geared motor bike build (xf19) aiming for 20-25A initially, throttle through an esp32 as laid out on the firmware's github page. A 3-position switch to select a max throttle level might be something I'll look into once it's rolling. Just headless, throttle-only operation to begin with - I have a BMS app to monitor the pack and speed with anyway.
To address the obvious 'why?' question, I have a pretty limited amount of space for a controller on this bike without significant redesign or a phaserunner/CA3-sized budget increase, and little need to exceed the 20-25A the motor is rated for without oil or a better connector for phase wires. The M365 controller is a bit more compact where it counts for me, inexpensive, also 52V-compatible with only a reflash, has a 32-bit processor to the KT's 8-bit and can do better FOC calculations the slower KT can't manage. Just looking to achieve smooth, maybe slightly quieter operation and a little efficiency bump as a bonus. The Lishui LSW-675 from BMSBattery, from my measurement/estimation would still be a fairly snug fit, and costs roughly double the shipped price of this controller.
As the plan sits now, I'll start with a v2.1/v3.0 board (slightly reinforced traces with extra solder vs earlier models), add wire to traces and improve fet-heatsink contact over the rubber-pad-and-case-lid factory method with individual screws and thermal silicone pads. The XT30 connector will come out in favor of 12awg wire leading to an XT90 to match the 52V DIY pack, and the spade terminal phase connectors likewise will get 12awg wire soldered on (after drilling the small center hole out on each) going to 4mm bullets. A bit of a list, but easy tasks as far as I'm concerned. Without this extra reinforcement the scooter guys are taking these controllers to 35A or so battery current before they seem to start giving out, and even then usually under regen (not applicable for me), we'll see what it can do in a slightly different application with a larger motor when the parts arrive and update here accordingly.
Following up from the questions in the KT thread to not muddy its waters..
I misread a little there for the status, sorry - it'd be convenient though the config.h file isn't otherwise too difficult to parse to hardcode values into for basic use with my ultra-rusty C++ knowledge, EBiCS documentation referencing and a bit of basic deduction, at least outside of what I've already asked. The M365 scooter community is still kicking, though it seems the many scooter-specific firmware configurators out there cover most of their needs with stock batteries and motors, the hardcoded 15-pole count being a bit of a speedbump for some modders iirc. I don't know if any KT/lishui display options (or the bluetooth app via esp32, if possible) could be made to interface with it, and of course that depends on interest and ideally more people confident/familiar with coding, which I'm sadly not. I'm only one guy exploring a niche solution for what is really a self-created problem, but a good small low-ish-power basic controller with decent FOC, a torque throttle and a price tag this low makes for an interesting experiment in a segment of DIY ebike tech that feels a little stagnant or slow at times compared to mid-drive, e-scooter and e-board development.
Cheers!
To address the obvious 'why?' question, I have a pretty limited amount of space for a controller on this bike without significant redesign or a phaserunner/CA3-sized budget increase, and little need to exceed the 20-25A the motor is rated for without oil or a better connector for phase wires. The M365 controller is a bit more compact where it counts for me, inexpensive, also 52V-compatible with only a reflash, has a 32-bit processor to the KT's 8-bit and can do better FOC calculations the slower KT can't manage. Just looking to achieve smooth, maybe slightly quieter operation and a little efficiency bump as a bonus. The Lishui LSW-675 from BMSBattery, from my measurement/estimation would still be a fairly snug fit, and costs roughly double the shipped price of this controller.
As the plan sits now, I'll start with a v2.1/v3.0 board (slightly reinforced traces with extra solder vs earlier models), add wire to traces and improve fet-heatsink contact over the rubber-pad-and-case-lid factory method with individual screws and thermal silicone pads. The XT30 connector will come out in favor of 12awg wire leading to an XT90 to match the 52V DIY pack, and the spade terminal phase connectors likewise will get 12awg wire soldered on (after drilling the small center hole out on each) going to 4mm bullets. A bit of a list, but easy tasks as far as I'm concerned. Without this extra reinforcement the scooter guys are taking these controllers to 35A or so battery current before they seem to start giving out, and even then usually under regen (not applicable for me), we'll see what it can do in a slightly different application with a larger motor when the parts arrive and update here accordingly.
Following up from the questions in the KT thread to not muddy its waters..
stancecoke said:It works on the Lishui project, but it should be no problem to make the configurator run on the M365 project also, if there is a bigger user-community for the M365 firmware...
I misread a little there for the status, sorry - it'd be convenient though the config.h file isn't otherwise too difficult to parse to hardcode values into for basic use with my ultra-rusty C++ knowledge, EBiCS documentation referencing and a bit of basic deduction, at least outside of what I've already asked. The M365 scooter community is still kicking, though it seems the many scooter-specific firmware configurators out there cover most of their needs with stock batteries and motors, the hardcoded 15-pole count being a bit of a speedbump for some modders iirc. I don't know if any KT/lishui display options (or the bluetooth app via esp32, if possible) could be made to interface with it, and of course that depends on interest and ideally more people confident/familiar with coding, which I'm sadly not. I'm only one guy exploring a niche solution for what is really a self-created problem, but a good small low-ish-power basic controller with decent FOC, a torque throttle and a price tag this low makes for an interesting experiment in a segment of DIY ebike tech that feels a little stagnant or slow at times compared to mid-drive, e-scooter and e-board development.
Cheers!