Lebowski's motor controller IC, schematic & setup, new v2.A1

Hi Bas

I am trying to design multi inverter application. I have some questions maybe others will benifit from.

1. Can Lebowski controller be run without throttle 2 connected? Do i have to connect its input to GND?

2. How would you go by using Fault/desat output of drivers? It is a open collector output of the driver module. Can i just use one fast PNP transistor to connect Enable pin to GND (it is the Reset function)? Or would you use another chip like arduino? What happens in code after reset if Enable pin gets 5V back? Would it be better to just cutoff 5V from Enable side of resistor?

3. About CAN pins.. If i dont use them, do i also have to connect them to GND? Or for that matter every pin i dont use?

tnx

Arber

Hi Arber,

Just to answer your questions:

1) yes it can work with only 1 throttle, in the setup menu you can specify only to use throttle 1 and not throttle 2.

2) I would indeed pull the reset on a fault detection, pull the reset (MCLR) to ground. This puts the chip in reset, a consequence of which being that all PWM outputs go to the 'off' state (so the complete power stage switches off). It does this very fast without software intervention. You may want to add something (flipflop) that keeps the chip in reset till you push a button to release it. You can also just reset it very short and allow it to start back up. In that case it will go through the normal startup, so first drive_0 where it will wait on throttle closed and/or motor standing still (dependent on what you enabled in the recovery menu).

3) every pin you don't use (including unused throttle inputs), I just leave them open :? through that's not the way you're officially supposed to deal with open pins. What is always safe is to connect the pin with a 4700 Ohm resistor to ground. If the pin is an input it will just see a ground. If the pin is an output it can easily drive the load of 4700 Ohm so no harm is done....

Another question

I have connected brain with computer and tried to programm it.
There were some problems with brain falling out of sync. So Q...

1. Do i have to have very precise voltage for 5V? I have constant 4.95V source. Very stable. Is it enough. I could use 3A variable PSU and set it to 5.1V, but do i have to?

2. What is the procedure for entering setup? It seems so quantum. Some times i get there but sometimes i cant. There is no rule...
Can you tell procedure A to B to C...

3. I use button to signal setup. Does this pin stays connected while in setup? In that case i have to setup a switch.

4. I have Prolific USB to serial gadget connected to TX, RX, GND. Should i connect 5V wire also?

5. What about voltage measurement? Do i need that? I intend to use 380Vdc so it would be catastrophic to just connect it to brain whatever the Rinput. Do i also put those pins to gnd via 4K7? Also i tried to change DC voltage value in menu but it just shows 6V even though i dont have anything connected.

tnx

Arber

1) 4.95V should be enough i its very stable, the chip automatically resets when it drops to 4.5V or so...

2) When the processor starts the first thing it does is look at the setup pin and go either setup or motor mode. After that there is no way to change the mode anymore. The processor starts: this is at power on or after a reset. So have the setup pin the way you want it when you power on, or WHILE you perform a reset.

4) I also only connect RX, Tx and GND, 5V not necessary

5) you dont need voltage measurement, you can connect the pin to ground. FOC accuracy will suffer a bit but not much.

Lebowski said:

5) you dont need voltage measurement, you can connect the pin to ground. FOC accuracy will suffer a bit but not much.

Click to expand...

OK I have added 4K7 resistor to pins 5 through 8. That should also connect voltage measurement with GND. Now i tried to use 385V as battery voltage, but it wont accept HV and LV parameters. So i guess this is something i cant use?

On second thought, could i use Si8920BC isolated comparator with an opamp like MCP6001? What is the resolution of pin 6? Can i just connect this circuit that i already use in Johannes inverter?


This could be added easily if your measurement would allow. For now i just use TL431 with large value resistor divider + variable pot to triger DC and precharge relay through PC817 opto on my terms. It works but is a bit crude. No extra IC needed that way though.

tnx

A

I mean, it doesn't really matter what circuit you use, as long as the battery voltage is mapped to 2.5 to 5 V at the controller IC input. The mapping must be linear, and all possible battery voltages (so between completely full and empty) must fall in this range...

I like the isolated amp by the way, would be fun to try to replace the in the motor wires placed current sensors with this amp (as a cheaper alternative for lower power controllers). Use shunt resistors, measure and transfer the voltage to the 5V domain with this Si8920. Use bootstrapping to supply the motor wire side of the amp...

Ampera inverter interface works.
I can drive ACIM motor with another brain.
Now i will try PMSM with Lebowski brain. Signals work on bench. I can save settings from Arlo video. 50V Leaf motor test awaits.

Fault pin on drivers is causing confusion. It is open collector output type and it is ACTIVE when in normal PWM operation. What is confusing is that whenever inverter is starting PWM this signal goes float for 10us and then back to active. Or when voltage goes under 25Vdc or when there is true short on IGBT...
It seems like that is not actually Fault signal, but more likely change of status signal or something. Does someone have any idea what kind of output it could be?

EDIT: Yeah... It definitely is Fault signal. Since i used it to trigger NPN transistor base EMI from 5V line caused Fault line to trip whenever PWM was activated. I will have to add more capacitance to filter out EMI. And i will use a transistor with higher threshold for base trigger.

tnx

A

OK Bas you will probably laugh and Arlo will say "cool!" but hear me out anyway.

What if.... i used Volt inverter dual 6 bridge 500A modules in paralel with a single Lebowski chip control? Everything would be in paralel.
First I would connect phase outputs of each module to single 3phase Leaf motor. DC rail is connected together anyways.

Then i would use two SN74LS06 chips after DSPic to invert signal chip to provide correct signal timing and enough power for both power stages.
Now there comes a problem. Current sensors... How to provide accurate current control from both chips? Do you think i could use only single sensor set and double their sensitivity?
If i only used a single sensor set chip would think it operates under 1/2 power only. Could i just cut other sensors from the loop and hope that by derating power stage i would have enough safety margin? Inverter would provide 2x the power but i wouldnt care since i would use torqe control with throttle pedal.

I would of course derate each module to say 400A at 400V. But do you think i could do it? It would be a sweet package for the power.

A

Check this out!

https://tspace.library.utoronto.ca/bitstream/1807/18295/1/Fingas_William_Daniel_200911_MASc_Thesis.pdf

I found this while browsing for paralel controller design. According to this two inverters can run together on the same DC bus as well as AC bus. Well control must be tight, but it could be done. Want to try some day?

I will have a look after I am back from holiday... Am on Tenerife now, where it is nice and cool

Hi Bas

I found interesting chip that can measure outside thermistor and output digital signal. Is it possible to adapt this to your code so we could use existing (scavenged) inverter signalling?

https://datasheets.maximintegrated.com/en/ds/MAX6682.pdf

tnx

arber333 said:

Now there comes a problem. Current sensors... How to provide accurate current control from both chips? Do you think i could use only single sensor set and double their sensitivity?
If i only used a single sensor set chip would think it operates under 1/2 power only. Could i just cut other sensors from the loop and hope that by derating power stage i would have enough safety margin? Inverter would provide 2x the power but i wouldnt care since i would use torqe control with throttle pedal.

Click to expand...

If each section has it's own current sensors, you could take the signals from both and electronically add them. This could be done analog or digitally.
Another approach would be to route the wires from both modules to a single set of sensors. I don't know how it looks or whether that would be feasible.

Hm... yes! I remember i have 3x 1200A sensors that are powered from +/-15V. I would have to convert that to 5V then and amplify, but it would be feasible . Nice!
It would mean i would have to add a large box on Ampera inverter side to get cables through sensors. Did I mention i like mechanical problems more than electrical. :lol:

arber333 said:

Hm... yes! I remember i have 3x 1200A sensors that are powered from +/-15V. I would have to convert that to 5V then and amplify, but it would be feasible . Nice!
It would mean i would have to add a large box on Ampera inverter side to get cables through sensors. Did I mention i like mechanical problems more than electrical. :lol:

Click to expand...

Usually those sensors gives an output signal which is a small current (that is proportional to the measured current). So its just a matter of connecting some resistors of the right size on the output to get a 0-5V signal. But check the offset and noise specs, they might not be good for a sensor of that range (1200A).

Hi Anyone

Can you help me? I have V2.A1 Lebowski brain in Volt inverter, i setup my board with Arlo settings. I will use 360Vdc battery and Leaf motor.
So first when i sorted out precharge and reset functions to 280Vdc and performed setup with parameters i used 50V to try to setup hall sensors and motor inductance. I could get good offset measurement but inductance wouldnt work and it would report overflow error. So i prepared full 360Vdc battery.

Up to here all was understandably well. Now however brain is running all LEDs by twos whenever i start it. When i connect drivers and hall sensors and try to run hall sensor calibration and motor inductance brain goes into reset and again starts to show LEDs by two. What should i do? Can i save this brain or should i order a new one?
Bas help!

arber333 said:

Hi Anyone

Can you help me? I have V2.A1 Lebowski brain in Volt inverter, i setup my board with Arlo settings. I will use 360Vdc battery and Leaf motor.
So first when i sorted out precharge and reset functions to 280Vdc and performed setup with parameters i used 50V to try to setup hall sensors and motor inductance. I could get good offset measurement but inductance wouldnt work and it would report overflow error. So i prepared full 360Vdc battery.

Up to here all was understandably well. Now however brain is running all LEDs by twos whenever i start it. When i connect drivers and hall sensors and try to run hall sensor calibration and motor inductance brain goes into reset and again starts to show LEDs by two. What should i do? Can i save this brain or should i order a new one?
Bas help!

Click to expand...

Hi Arber, I always get that LED flashing pattern with a new chip until the inductance test has run. I'm not sure if that flashing pattern is documented, Lebowski can you eloborate?

You can try lowering the test amperage and also the erpm settings for the test. If you post your settings I will take a look.

I have been following your Volt inverter adventures and I am getting close to testing a drop in board using one of the AD2S1205 resolver chips salvaged off the original Volt brain board combined with Lebowski's brain chip. I decided to follow Arlo's lead and started testing at a low voltage. 18S works well, I will try 30S later this week. I plan on running 96S with the Volt power stage.

kiwifiat said:

Hi Arber, I always get that LED flashing pattern with a new chip until the inductance test has run. I'm not sure if that flashing pattern is documented, Lebowski can you eloborate?

You can try lowering the test amperage and also the erpm settings for the test. If you post your settings I will take a look.

I have been following your Volt inverter adventures and I am getting close to testing a drop in board using one of the AD2S1205 resolver chips salvaged off the original Volt brain board combined with Lebowski's brain chip. I decided to follow Arlo's lead and started testing at a low voltage. 18S works well, I will try 30S later this week. I plan on running 96S with the Volt power stage.

Click to expand...

Well thank you. Can you show the board or link to it maybe. I could use the input. Now i just use simple TL431 voltage trigger to keep the GND engaged for relays and i use the pair to keep brain in reset untill precharge is reached. That way if voltage is dropped DCswitch falls out but precharge is still active. Safety...
I may use Arduino nano to control this aspect with more accuracy in future. I also use fault pin from Volt driver section too.

I was successful in running Leaf motor in my friends Mazda RX8 using Johannes brain board and my interface. It was able to go up to 8K rpm but inverter software was made for ACIM and software needs some review. Principaly when you release throttle magnetic field does not rotate with rotor to give you 0 difference, but just stops. In ACIM that means rotor becomes spinning iron block. But here motor violently loads regen and that is most evident at low speeds. Driving is as of now most uncomfortable. I hope Johannes will adapt software soon since he also has Leaf motor in works.

That is why i wanted to try Lebowski FOC board. But now i am stuck with board resetting even before any test actually begins. Is this something that is known for this software version?

Before i had success on 50V in offset calibration, but i got owerflow on inductance test. But then i could see only first LED was lit. Now i get interleaved pair all the time. Oh yes i dont use hall sensors yet. For testing i didnt deem them neccessary. Should i use them?

Heh, if i try to run Leaf motor on 50V using some chineese bicycle inverter it spins right off the bat without sensors. Crazy!

Well i have done it! My motor turns beyond 8000rpm! I will have to mount hall sensors that i get soon to be able to drive comfortably. I will post som photos of that.
I was in contact with Lebowski and he provided guidence. In red is what he said in green what i told him. Maybe someone will find something good out of it.

L
The inductance measurement gives an overflow error when the voltage necessary to obtain the in the foc measurement menu set current is more than half the supply voltage. So with a high current and low voltage you can get this error.
The two by two led blinking indicates that you are trying to run in motor mode without having so the chip completely.
It sounds like the chip resets when you draw power from the main battery. This is typically caused by a dip in the 5V, causing a brown out reset. This can be caused by a 5V DC DC responding too slow to the input voltage drop... In the menu you start the measurement, the chip resets and enters motor mode where it then flashes the LEDs to show has not been setup completely.
Best is after every mayor change, like going from 50 to over 200 V battery, reset all settings. This is done by starting a chip update with '!' from the main menu, and then resetting it (this interrupts the update process but by then it has already reset all variables
A
Hm... i have lots of caps on 5V line. Should i increase the one directly next to the input pin. Oh btw i did not use 10ohm resistor for that one i use direct wire bridge. Should i use 10ohm for input?
Will this brounout be visible on scope? Where to connect probe? Hm... I have LM2596TV 5V regulatoron on line and it produces 5.05V where i measure near chip. Should i put another cap somewhere? Where is it most critical?
Hm... i did not use 100nF cap on current sensor supply. I can add this also.
L
the 10 ohm is there for filtering as that one input is used as a reference for the ADC
A
DOH! I will add 10R ASAP!
A
Dang! I inspected driver resistors and to my horror i found 150R!!! This is way over the chip drain limit. Do you think outputs could be damaged?
I just replace them by 330R and power section now works. I could run current sensor calibration and FOC impedance measurement.
However now i cant get the motor to drive 1. LEDs are still blinking.
L
If the LEDs are still blinking 2 by 2, not all menu's have been set up. For instance the onlije LR measurement menu, the least you need to do there is autocomplete it (and then turn off). In the throttle menu for indtace at least one of the throttle coefficients must not be 0. And in this way every menu has something that the chip checks put that it is filled in.
So, go through all the menu's and for a minimum auto complete all of them.

That is great news. I took a look at your DesignSpark files and it appears to me that you are not using any phase feedback to pins 5,6, & 8 on the 30F4011 brain, is that the case?

If so that begs a question for Lebowski, can the brain chip commutate the motor in sensorless mode without phase voltage feedback?

I have included a phase feedback network as per the original design with resistance values adjusted for my Volt inverter project, can I get away without it?

kiwifiat said:

If so that begs a question for Lebowski, can the brain chip commutate the motor in sensorless mode without phase voltage feedback?

I have included a phase feedback network as per the original design with resistance values adjusted for my Volt inverter project, can I get away without it?

Click to expand...

Yep it runs the motor only using the current sensor information. The phase voltage feedback is there so that in drive_0 it can see whether the motor is running or not (this is one of the checks at startup). It basically looks for near 2.5V voltage on the 3 pins to decide the motor is standing still. This check also serves to make sure all FETs are off, if one is permanently on (broken) it will not start up. You can turn on / off this check (I think it is in the recovery menu). If the check is off you don't need the voltage feedback network to the 3 pins.

This check is the only reason the voltage feedback is there. But it has saved my bacon once, due to rainwater in the controller I once had it fail this check. After drying up everything was fine again. But without this check it would have blown up.

It can run sensorless quite good, but i am concerned i will have trouble starting a car uphill. Leaf motor has resolver at the back. This is not much use to this controller as it needs comutation encoder. I can have direct access to rear motor windings. Motor is 48tooth, 8pole 3phase BLDC motor. First i gor 120deg placement at 16 teeth.
I was thinking if i could change my calculation of sensor placement from 120deg into 60deg would that be ok still
Instead of calculation for placement sensors: 360/48 = 7.5deg per tooth, 120deg/7.5deg = 16 teeth apart
I would use placement of sensors: 360/48 = 7.5deg per tooth, 60deg/7.5deg = 8 teeth apart
Heh, maybe i could get away with placement per 30deg? That would then be 4 teeth apart?
Can you estimate if that would work with your controler? It would make my cables quite a bit shorter inside motor.

Also I was thinking of making isolated phase feedback reading on my board. So from what i see controler brain needs only digital signal with less than 2.5V meaning 0 and higher up to 4.9V meaning 1? Would that work?
Or is there some rising edge function? I could make this work with some isolated circuit.

Also does anyone have an idea how to translate voltage value to this controler? Would my schematic from post https://endless-sphere.com/forums/viewtopic.php?f=30&t=36602&start=1700#p1397141 work, with some tuning of course. Again 2.5V is 0 and 5V is max reading? How precise is that for 360Vdc system. Would you recommend to use 4.5V for 390Vdc value then?

EDIT: @kiwifiat
! Caution, do not try to use phase sensing wires directly with Volt inverter! You may cause a short with GND plane somewhere. I think Volt inverter has its own isolation solved. I just bring it signals and it uses them. Better way to try would be to make isolated connections bridged to brain inputs. That way you can still connect them to gnd and you are good. I am doing it this way, but i have intent to adapt some isolation connection to phases

arber333 said:

Also I was thinking of making isolated phase feedback reading on my board. So from what i see controler brain needs only digital signal with less than 2.5V meaning 0 and higher up to 4.9V meaning 1? Would that work?
Or is there some rising edge function? I could make this work with some isolated circuit.

Also does anyone have an idea how to translate voltage value to this controler? Would my schematic from post https://endless-sphere.com/forums/viewtopic.php?f=30&t=36602&start=1700#p1397141 work, with some tuning of course. Again 2.5V is 0 and 5V is max reading? How precise is that for 360Vdc system. Would you recommend to use 4.5V for 390Vdc value then?

EDIT: @kiwifiat
! Caution, do not try to use phase sensing wires directly with Volt inverter! You may cause a short with GND plane somewhere. I think Volt inverter has its own isolation solved. I just bring it signals and it uses them. Better way to try would be to make isolated connections bridged to brain inputs. That way you can still connect them to gnd and you are good. I am doing it this way, but i have intent to adapt some isolation connection to phases

Click to expand...

You raise a legitimate point. If you look at the original Volt brain you can see that the voltage sense area of the PCB has no ground plane or internal layers. I am following the same principal and have a keepout zone around the battery voltage sense circuitry and the phase feedback and the clearances I am using exceed IPC2221 and UL standards by a wide margin. Still now that I know that phase feedback is not required for sensorless running I will reconsider if I want to use it or look at an isolated solution. It seems that a simple solution might be to use a micro on the powerstage side to measure the phase and battery voltages and provide isolation to the brain side via pwmed opto-isolaters driving an rc network for voltage feedback.

kiwifiat said:

arber333 said:

Also I was thinking of making isolated phase feedback reading on my board. So from what i see controler brain needs only digital signal with less than 2.5V meaning 0 and higher up to 4.9V meaning 1? Would that work?
Or is there some rising edge function? I could make this work with some isolated circuit.

Also does anyone have an idea how to translate voltage value to this controler? Would my schematic from post https://endless-sphere.com/forums/viewtopic.php?f=30&t=36602&start=1700#p1397141 work, with some tuning of course. Again 2.5V is 0 and 5V is max reading? How precise is that for 360Vdc system. Would you recommend to use 4.5V for 390Vdc value then?

EDIT: @kiwifiat
! Caution, do not try to use phase sensing wires directly with Volt inverter! You may cause a short with GND plane somewhere. I think Volt inverter has its own isolation solved. I just bring it signals and it uses them. Better way to try would be to make isolated connections bridged to brain inputs. That way you can still connect them to gnd and you are good. I am doing it this way, but i have intent to adapt some isolation connection to phases

Click to expand...

You raise a legitimate point. If you look at the original Volt brain you can see that the voltage sense area of the PCB has no ground plane or internal layers. I am following the same principal and have a keepout zone around the battery voltage sense circuitry and the phase feedback and the clearances I am using exceed IPC2221 and UL standards by a wide margin. Still now that I know that phase feedback is not required for sensorless running I will reconsider if I want to use it or look at an isolated solution. It seems that a simple solution might be to use a micro on the powerstage side to measure the phase and battery voltages and provide isolation to the brain side via pwmed opto-isolaters driving an rc network for voltage feedback.

Click to expand...

Seems like overkill to me considering it's only used to detect "moving" or "stationary" where moving is anything other than 2.5V (so say 0V or 5V) seems like a 2 state solution (single digital output) which is very easy to isolate would work on the brain side..
Andy

Animalector said:

kiwifiat said:

You raise a legitimate point. If you look at the original Volt brain you can see that the voltage sense area of the PCB has no ground plane or internal layers. I am following the same principal and have a keepout zone around the battery voltage sense circuitry and the phase feedback and the clearances I am using exceed IPC2221 and UL standards by a wide margin. Still now that I know that phase feedback is not required for sensorless running I will reconsider if I want to use it or look at an isolated solution. It seems that a simple solution might be to use a micro on the powerstage side to measure the phase and battery voltages and provide isolation to the brain side via pwmed opto-isolaters driving an rc network for voltage feedback.

Click to expand...

Seems like overkill to me considering it's only used to detect "moving" or "stationary" where moving is anything other than 2.5V (so say 0V or 5V) seems like a 2 state solution (single digital output) which is very easy to isolate would work on the brain side..
Andy

Click to expand...

For the phase voltage feedback I agree, Lebowski has commented that the battery voltage feedback has an affect on the FOC accuracy which might not be an issue on a 3kW ebike motor but on a 100-200kW car conversion could be worth the few dollars cost to implement. Maybe Arlo could tell us if he is using battery voltage feedback on his inverter?

Yes You need battery voltage input to brain just not phase voltage