Lebowski's motor controller IC, schematic & setup, new v2.A1
- Thread starter Lebowski
- Start date
lizardmech
10 W
- Joined
- May 19, 2017
- Messages
- 67
How do you find the performance of the allegro ACS sensors in general? I have only used resistive current sense in the past but allegro now have some nice 100V high current SMD models and their existing models were upgraded with more bandwidth. It's almost impossible to find any performance comparison between the different current sensing options around or information about if you should prioritize bandwidth, rise time or other characteristics when selecting them for motor control.Lebowski said:
Lebowski
10 MW
I prefer proper current (magnetic field based) over resistive ones for the simple reason that if you want to keep the power dissipation the same, the higher the current rating of the resistor the lower the effective voltage you'll get.
A 1W sense resistor for 1A can deliver 1V signal, a 1W for 1000A will only deliver 1mV signal, with only 1uV per A.
I would say phase delay through the sensor is most important, as this delay translates to an offset between motor voltage and current
A 1W sense resistor for 1A can deliver 1V signal, a 1W for 1000A will only deliver 1mV signal, with only 1uV per A.
I would say phase delay through the sensor is most important, as this delay translates to an offset between motor voltage and current
The specification on the Allergro ACS states a response time of 4us and bandwidth of 120 kHz. Seems just barely fast enough.
lizardmech
10 W
- Joined
- May 19, 2017
- Messages
- 67
Response time seems to be the biggest weakness on the ACS, at least on the high current versions. They have 1mhz versions with sub 1us propagation but they never release them in the high current versions for some reason. The INA240 I have been using recently is faster but they are limited to 80V and shunt resistor heat is an issue, you gain a few watts worth of heat at 100A+ and accuracy theoretically suffers as resistance changes. An isolated current shunt amplifier I tried is faster than high current ACS but still had the down sides of resistive sense. I'm not sure how fast is fast enough for motor control, considering a worst case scenario of a motor with 200,000 ERPM you still have around 300us per electrical cycle?
lizardmech
10 W
- Joined
- May 19, 2017
- Messages
- 67
While researching giant magnetoresistance I discovered these isolated current sensors today, a little more expensive than ACS but sub 1us response time and high bandwidth.
https://www.sensitec.com/products-solutions/current-measurement/cds4000-family
https://www.sensitec.com/products-solutions/current-measurement/cds4000-family
Lebowski
10 MW
Hm I don't know, 120kHz bandwitdth for the ACS parts seem good enough to me. 100k-erpm translates to 1.6 kHz, a pole at 120kHz would give a phase rotation of less than 1 degree. Cycle time of the algorithm is in the order of 30usec, so 4 usec delay in the current sensor doesn't seem relevant.
Lebowski
10 MW
Been very busy here with the saliency (?) based position detection, want to detect the rotor phase at standstill by measuring the differences between the motor inductances.
And I want it good, just dividing the rotor position in 60 degree portions (like what you get with hall sensors) will not be good enough, I want like 8 bits or so. Basically the goal is to get phase info as if there was a resolver, but then without the resolver but based on the motor inductance info.
The level of processing this needs is staggering. To get 8 bit info is actually quite simple (but needs a trick that took me 5 days to figure out), the MSB though (the bit saying we're between 0-179 or 180-359 degrees) is a real #@$%#^*$&*. Been playing with sending out a bit pattern in the Field weakening dimension (Id or Iq, don't know what it is called officially, I call it I_imag), using matched filters etc to detect it back in the saliency and then based on inversion or not decide the MSB. This kinda worked but was horribly unreliable and audible noisy to boot (oh yeah, another thing is I don't want a loud tone or anything like that).
Also had to start using the DSP as simple first order lowpass filtering is no longer good enough. Have now switched over to FIR filtering and doing stuff like sample rate conversion.
I do now have something that works for deciding the MSB, but it does not work always. I already figured out why this is and also how to fix it, but it will again put an extra layer of complexity on top. Luckely at the moment I am nowhere near the limit of the processor, the one I have now does 70 M instructions per second versus 30 for the 30F.
Good thing it is so complex though, with me being out of a job at the moment it is a fun project to keep me busy (though with the summer going on here there is also lots of cycling and motorbiking to be done
)
And I want it good, just dividing the rotor position in 60 degree portions (like what you get with hall sensors) will not be good enough, I want like 8 bits or so. Basically the goal is to get phase info as if there was a resolver, but then without the resolver but based on the motor inductance info.
The level of processing this needs is staggering. To get 8 bit info is actually quite simple (but needs a trick that took me 5 days to figure out), the MSB though (the bit saying we're between 0-179 or 180-359 degrees) is a real #@$%#^*$&*. Been playing with sending out a bit pattern in the Field weakening dimension (Id or Iq, don't know what it is called officially, I call it I_imag), using matched filters etc to detect it back in the saliency and then based on inversion or not decide the MSB. This kinda worked but was horribly unreliable and audible noisy to boot (oh yeah, another thing is I don't want a loud tone or anything like that).
Also had to start using the DSP as simple first order lowpass filtering is no longer good enough. Have now switched over to FIR filtering and doing stuff like sample rate conversion.
I do now have something that works for deciding the MSB, but it does not work always. I already figured out why this is and also how to fix it, but it will again put an extra layer of complexity on top. Luckely at the moment I am nowhere near the limit of the processor, the one I have now does 70 M instructions per second versus 30 for the 30F.
Good thing it is so complex though, with me being out of a job at the moment it is a fun project to keep me busy (though with the summer going on here there is also lots of cycling and motorbiking to be done
)Altair
1 kW
Well we can say that you, being out of a job, has been a godsend for the e-bike community. :lol:
That sounds very interesting.
So you calculate the rotor position without injecting any pwm pattern and without injecting any "I_imag" (Id) currents?
Do you measure the Ld and Lq inductance (anisotropy) while the deadtime and turn-on time of the mosfets?
So you calculate the rotor position without injecting any pwm pattern and without injecting any "I_imag" (Id) currents?
Do you measure the Ld and Lq inductance (anisotropy) while the deadtime and turn-on time of the mosfets?
Lebowski
10 MW
I think the deatime is too short (set for 500nsec) to get anything useful out... plus there is phase current flowing to make the motor turn, effectively forcing what happens during deadtime...
Putting a pattern on the I_imag was like the neutral light on my old Moto Guzzi, it was right about 80% of the time... the 20% when it is wrong can be really dangerous though...
Putting a pattern on the I_imag was like the neutral light on my old Moto Guzzi, it was right about 80% of the time... the 20% when it is wrong can be really dangerous though...
I have read a paper where they used an FPGA to measure the currents @20MHz, but I also think
that deadtime is too short and full of noise because of the switching FETs.
Than you inject some kind of short pwm pattern like INFORM method and calculate the current derivative, right?
Do you have two different algorithms for low and high speed?
that deadtime is too short and full of noise because of the switching FETs.
Than you inject some kind of short pwm pattern like INFORM method and calculate the current derivative, right?
Do you have two different algorithms for low and high speed?
Lebowski
10 MW
INFORM method ? I dont follow the literature and like to do everything myself, is more fun... at high speed I will switch over to backemf based sensorless of course.
In INFORM (indirect flux detection using online reactance measurement) some pwm patterns are applied "out of order" of field oriented control and measure the current derivative depending on voltage.
This method is much quieter than injecting current waves on imaginary axis (I_imag).
Requires your algorithm measurement of phase voltages or do you measure only the currents?
This method is much quieter than injecting current waves on imaginary axis (I_imag).
Requires your algorithm measurement of phase voltages or do you measure only the currents?
Lebowski
10 MW
Is this the method where they measure inductance using the rise time of the current ? Don't you get only a 1 out of 6 info from this (so only 60 degrees resolution) ?
tecnologic
1 W
- Joined
- Mar 24, 2015
- Messages
- 54
No u get only a position every 6th pwm cycle. But this depends on the implementation. Prof. Schrödl located in Wien proposed this method.
I would expect because of the analog circuitry u have to increase current resolution. u will sample the current at lets say 40% and 60% of pwm cycle with PWM switiching at around 50% und build the current deravative?
I would expect because of the analog circuitry u have to increase current resolution. u will sample the current at lets say 40% and 60% of pwm cycle with PWM switiching at around 50% und build the current deravative?
Lebowski
10 MW
Just spent 2 days programming, but ran into a problem with the 33EV I'm using... some obscure mistake in the silicon explained in the errata (but who reads those ?). Luckely there's other ways to skin a cat...
Lebowski
10 MW
my idea for getting full 360 info has worked, now I get reliable info about whether the phase is 0-179 or 180-359. So all combined I now get full 360 degree info (smooth, not just 6 levels) about the rotor position, sensorless at standstill. Processing time is around 10 to 20 msec for the first reliable output after a cold start, after that it should be able to keep up until it shifts to backemf based sensorless. The short processing time makes that the controller can switch off the output stage while waiting for the traffic lights, and switch on only at throttle up. And independent of whether you rolled backwards or forwards, and without needing to accurately sample motor voltages (doesn't need motor voltages...)There is still a big mistery in there somewhere that I need to look at... a case of a polarity being different from what I expected. If I just dumbly take this into its stride it will work, I'm just curious why it is different and I want to understand this.
Next step is to add phase current.... I have kept this in mind and it should be OK, but you never know...
)kiwifiat
100 W
Lebowski said:
There is still a big mistery in there somewhere that I need to look at... a case of a polarity being different from what I expected. If I just dumbly take this into its stride it will work, I'm just curious why it is different and I want to understand this.
Next step is to add phase current.... I have kept this in mind and it should be OK, but you never know...
Great work Lebowski, any chance this will be back ported to the 30F4011?
Lebowski
10 MW
No chance... the hardware is just not there
whereswally606
100 kW
Great Work Bas, looking forward to this. Maybe its time your new chip and software had a thread of its own. Stop people getting confused.
Hi Bas, good job!
Is there a possibility to make a schematic like for 30F? I am in the middle of multi inverter project where i am committed to your design, but pinout is very general. I could just make a new brain board with same pinout if i had new chip schematic. Would it be powered from 5V or is it 3V3? In that case i would have to adapt signalling....
Does your sensorles start require any outside elements?
tnx
Arber
Is there a possibility to make a schematic like for 30F? I am in the middle of multi inverter project where i am committed to your design, but pinout is very general. I could just make a new brain board with same pinout if i had new chip schematic. Would it be powered from 5V or is it 3V3? In that case i would have to adapt signalling...
. Does your sensorles start require any outside elements?
tnx
Arber
Lebowski
10 MW
Its a 5V chip (a 33EV from microchip), but for now it is all very early stages. Main thing to see is how it holds up under high throttle...
I have some opamp circuits, mainly to help the ADCs...
I have some opamp circuits, mainly to help the ADCs...
Tomdb
10 W
having been dusting off the controller.
However now it wont even spin the motor. All calibrations run without an issue.
However it sometimes will not go out of drive 1 or keep flashing the led and not power the stage up.
The hex is below.
However now it wont even spin the motor. All calibrations run without an issue.
However it sometimes will not go out of drive 1 or keep flashing the led and not power the stage up.
The hex is below.
Code:
version:293_
0x0133 0xFFFF 0xFFFF 0x0503 0x03C2 0x0001 0x0001 0x0000
0x0400 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0x0370 0x25DB 0x7FBC
0x7D2E 0x0030 0x070E 0x050E 0x2324 0x000B 0xFFFF 0xFFFF
0xFFFF 0xFFFF 0x795E 0x0400 0xB76C 0x000A 0x0000 0x07AE
0x0018 0x0000 0x01E0 0xFFFF 0xF852 0xFFE8 0x0000 0xFE20
0x0000 0x4CCD 0x0018 0x0000 0x01E0 0xFFFF 0xB333 0xFFE8
0x0000 0xFE20 0xF965 0xFF07 0xFFFF 0xFFFF 0x2141 0xFFFF
0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF
0xFFFF 0xFFFF 0x000F 0xF71A 0x0812 0x0060 0x0033 0x0000
0xFFFF 0x0000 0x01AB 0x00D5 0x03F7 0x03F0 0x03F0 0x0EA5
0x0581 0x002D 0x06E2 0x0005 0x0C99 0xFFF8 0x036F 0x1000
0x0000 0x0000 0x1000 0x0000 0x0000 0x0188 0x01D8 0x003B
0x0000 0x06AD 0x0055 0x0473 0x0590 0x0350 0x0258 0x0064
0xFFFF 0xFFFF 0xFFFF 0xFFFF 0x0003 0x0000 0x0078 0x0000
0x0000 0xFFFD 0x0000 0xFF88 0x0000 0x0000 0x0005 0x0000
0x0064 0xFFFB 0x0000 0xFF9C 0x000C 0x0000 0x00F0 0xFFF4
0x0000 0xFF10 0x05CC 0x014C 0x0183 0x0301 0x0356 0x0080
0x0226 0x0301 0x0013 0x0010 0x0E10 0x03E8 0x03B6 0x6000
0x0200 0x00C0 0x4000 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF
0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF
0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF
0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF
0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF
0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF 0xFFFF
0xFFFF 0xFFFF 0xFFFF 0x0000 0x0000 0x0000 0x0005 0x0000
0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0xFFFF
0xAAAA 0x6666 0x4924 0x38E3 0x2E8B 0x2762 0x2222 0x1E1E
0x1AF2 0x1861 0x1642 0x147A 0x12F6 0x11A7 0x1084 0x0F83
0x0EA0 0x0DD6 0x0D20 0x0C7C 0x0BE8 0x0B60 0x0AE4 0x0A72
0x0A0A 0x09A9 0x094F 0x08FB 0x08AD 0x0864 0x0820 *Similar threads
