How much should I increase it for every step? I take it that I have to start from p=0, i=1 and increase the values until I get to a smooth rotation?
Lishui "Open Source Firmware" project / KingMeter 5S
- Thread starter AndZab
- Start date
stancecoke
1 MW
No. This are the extreme values. If it doesn't work with this settings, we would have to try completely other ways.
I am doing something wrong with debug. Have it set at 9600 asci with hterm data is flowing but i am only seeing squares so something is wrong in my setting. I hgave done this
github.com
07 The Diagnostics Mode
OSEC Open Source E-Bike Controller in a Windows toolchain for a STM8 - stancecoke/BMSBattery_S_controllers_firmware
github.com
Last edited:
stancecoke
1 MW
In debug mode, you have to use 56000 BAUD!
Yes tried three different cables only one worked 
Now I get with nothing hooked up to the controller
phase current offsets: 2040, 2044, 2056 <\r><\n>
internal temperature raw reading: 1143, <\r><\n>
Hall_Order: -1 <\r><\n>
Hall_45: 27 <\r><\n>
Hall_45: 334036992 <\r><\n>
Hall_51: 90 <\r><\n>
Hall_51: 1073741824 <\r><\n>
Hall_13: 152 <\r><\n>
Hall_13: 1825308672 <\r><\n>
Hall_32: -151 <\r><\n>
Hall_32: 2505375744 <\r><\n>
Hall_26: -90 <\r><\n>
Hall_26: 3221225472 <\r><\n>
Hall_64: -29 <\r><\n>
Hall_64: 3960864768 <\r><\n>
KV: 224 <\r><\n>
Lishui FOC v1.0 <\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
Now I get with nothing hooked up to the controllerphase current offsets: 2040, 2044, 2056 <\r><\n>
internal temperature raw reading: 1143, <\r><\n>
Hall_Order: -1 <\r><\n>
Hall_45: 27 <\r><\n>
Hall_45: 334036992 <\r><\n>
Hall_51: 90 <\r><\n>
Hall_51: 1073741824 <\r><\n>
Hall_13: 152 <\r><\n>
Hall_13: 1825308672 <\r><\n>
Hall_32: -151 <\r><\n>
Hall_32: 2505375744 <\r><\n>
Hall_26: -90 <\r><\n>
Hall_26: 3221225472 <\r><\n>
Hall_64: -29 <\r><\n>
Hall_64: 3960864768 <\r><\n>
KV: 224 <\r><\n>
Lishui FOC v1.0 <\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
0, 5, 5000, 32000, 0, 0, 0, 0, 3<\r><\n>
stancecoke
1 MW
That looks good, the Hall angles read from the virtual EEPROM are in about 60 deg steps.
Last edited:
This time it's faster at finding the combination and coming up to speed although is still vibrating at lower speeds. It also continues to do the "emergency stop" on the first autodetect and then just continues spinning and vibrating at a low speed after the second.
stancecoke
1 MW
Of course, as the hall sequence still is not proper.
I think it makes no sense to do further testing in this way. We could try constant ud without current control, but this would need some coding. But I think it's easier to try constant angles in the main.h
Taking one of the hall values and guesstimating the other with a 60 degrees difference then converting the guess to a 32 bit value and trying?
stancecoke
1 MW
Of course the hall states in the sketch are not representative. The sense of the autodetect is to find the angle of each switch from one hall state to the next....
Last edited:
stancecoke
1 MW
As said before, your motor seems to have an extreme cogging torque, the rotor bounces.
I never had this strange behaviour with a motor I tested...
Maybe the motor has a trapezoidal BEMF waveform and would better be driven by a simple square wave controller....
FAQ: What is trapezoidal back EMF?
Brushless DC motors typically have a trapezoidal back EMF. The back EMF shape determines what type of drive current and commutation method should be used.
www.motioncontroltips.com
Strangely a 6 FET Kunteng (with your firmware) can drive it without issues.
stancecoke
1 MW
The Kunteng uses the same logic as I've suggest for several times already. It uses a motor specific angle and uses 60° steps between the hall sectors.
If you look at your various log files, you will find a reasonable start angle for sure....
regards
stancecoke
edit:
I've analyzed your latest logfile and come to the conclusion, that the angles should be:
| Hall_45 | 90 | 1073741824 |
| Hall_51 | 30 | 357913941 |
| Hall_13 | -30 | 3937053354 |
| Hall_32 | -90 | 3221225471 |
| Hall_26 | -150 | 2505397588 |
| Hall_64 | 150 | 1789569705 |
So you can edit your main.h from line 95 to 104:
Code:
#define USE_FIX_POSITIONS 1
//Put values from the startup message after autodetect here, if you want to use fix positions. 32bit values for the hall angles!
#define KV 80
#define HALL_ORDER 1
#define HALL_45 1073741824
#define HALL_51 357913941
#define HALL_13 3937053354
#define HALL_32 3221225471
#define HALL_26 2505397588
#define HALL_64 1789569705The KV value is not right in this case, but that only matters, if you start rolling without the motor engaged and later asking power from the motor.
For the tricky signed 32bit representation here a little tool
GDB online Debugger | Code, Compile, Run, Debug online C, C++
Online GDB is online ide with compiler and debugger for C/C++. Code, Compiler, Run, Debug Share code nippets.
Last edited:
Thank you, will try.
Made my own quick script in python but yours is much better.For the tricky signed 32bit representation here a little tool
Just as a sanity check. I can feel the motor cogging when no throttle is applied (in standby) which does not happen on the Kunteng. Is that normal?
stancecoke
1 MW
The phases are floating, when the PWM is switched off. So there can't be any influence of the controller on the motor.
Maybe your throttle offset is too low, so you get a little power demand with throttle closed. You can disconnect the throttle to check if this has an effect on the cogging.
No effect. Can still feel cogging with a disconnected throttle. The low offset is 720 and max is 2950. Which I got from the adc values
I'm looking at building a trike w/two hub motors(+also likely a mid-drive), and would prefer using this over KT,
but i didn't come across any with reasonable pricing on aliexpress. Given current options at pswpower.com i'm
likely to buy 500w hubs at first, but would prefer controllers which could handle upgrading to 1000w hubs later.
Any ideas/tips on compatible controller models to search for? i mean one that would work with this fw, and have
the rated current around 20A or so for 48V.
but i didn't come across any with reasonable pricing on aliexpress. Given current options at pswpower.com i'm
likely to buy 500w hubs at first, but would prefer controllers which could handle upgrading to 1000w hubs later.
Any ideas/tips on compatible controller models to search for? i mean one that would work with this fw, and have
the rated current around 20A or so for 48V.
stancecoke
1 MW
How is the behaviour with the motor cable disconnected?
Or with the motor connected but the battery disconnected?
The motor spins fine while disconnected or connected to another controller. It's also fine with the battery disconnected.
Seems to work better than what autodetect spits out although the motor refuses to accelerate past a certain speed and continues to spin even if I let go of the throttle. It also sometimes just resets while in motion and then stops the motor.
stancecoke
1 MW
Strange, do you have a different motor to test, to make sure that there is no issue with your controller hardware?
Only a sensorless. If it is the controller, what should I look into first?
stancecoke
1 MW
Difficult. Do you have a scope?
Yes I have a simple Hantek 6022BL
stancecoke
1 MW
That's OK. So you can check the signal between GND and the three phase wires first. There should be no signal, if the motor has been in standstill for several seconds. If you turn the throttle slightly, there should be a square wave with an amplitude of the battery voltage with a duty cycle of about 50% on all three phases.
I expect, that the PWM doesn't turn off with closed throttle. In this case you should comment out the line 510. Then the default value from the config.h will not be overwritten.
The autozero function may fail, if the voltages are not stable fast enough after switching on.
regards
stancecoke
I expect, that the PWM doesn't turn off with closed throttle. In this case you should comment out the line 510. Then the default value from the config.h will not be overwritten.
The autozero function may fail, if the voltages are not stable fast enough after switching on.
regards
stancecoke
