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

[fechter]

That's pretty impressive. How does it do from a standing start on an uphill with the bike?

 

[Lebowski]

That's pretty impressive. How does it do from a standing start on an uphill with the bike?

The video was just a demo to show slow speed under load, the motor was started without load. I already figured out how
to determine the magnet positions sensorless (by injecting a high frequency tone), so sometime during the winter I
hope to have added sensorless start under load to my IC. The vid was made with the back-emf style sensorless FOC
I have now...

 

[crossbreak]

still impressive. as soon as you have implemented sensorless position sensoring i must get hands on one of those. Hope my dyno test bench works till then

It's kind of pointless to pick a lam thickness without the other parameters.

I stated it should be up to 2000rpm. With >0.5mm it has to have at least 32poles, otherwise it would rev faster. I should indeed add >85% efficiency and good way of air cooling. Anyway I will use it with a 80cc rotomax with inductance <10µH to see if it can drive it. Even a Greentime fails here (which is fine with small colossus) , so i'm keen to see how it does on the big Colossus which plays in the same league, i would gladly pay you a visit, arlo

This will be my first power stage so i'll stick to nils PCB design with some propper/tunable snubbers. I build a PWM Fet circuit for power dissipation of the dyno, guess i'll learn something in the meantime.

 

[Arlo1]

It's kind of pointless to pick a lam thickness without the other parameters.

I stated it should be up to 2000rpm. With >0.5mm it has to have at least 32poles, otherwise it would rev faster. I should indeed add >85% efficiency and good way of air cooling. Anyway I will use it with a 80cc rotomax with inductance <10µH to see if it can drive it. Even a Greentime fails here (which is fine with small colossus) , so i'm keen to see how it does on the big Colossus which plays in the same league, i would gladly pay you a visit, arlo

This will be my first power stage so i'll stick to nils PCB design with some propper/tunable snubbers. I build a PWM Fet circuit for power dissipation of the dyno, guess i'll learn something in the meantime.

What Im getting at is the rotor diameter and length and the magnet strenght and air gap as well as lamantaion material used, combinded with average rpm and how long it needs to be at the max rpm or close to it combined with other details will determin the lam thickness. A motor designer will take all of this and more into concideration. You want the thinest lams for less eddy losses and the thickest lams for lower cost.

 

[crossbreak]

right... as long as it fits between the 68mm cranks i'm fine with it outer dia should be smaller than 200mm...a BMC motor would be fine if it would not have this horribly lot of waste heat/ low 75% efficiency....if it would be >80% like the bafang BPM I would go for 2 of them two BPM would be also a good choice, it is <30mm wide so two of them would fit between the cranks easily... but enough of OT discussion, lets see what lebowski does

 

[madin88]

You want the thinest lams for less eddy losses and the thickest lams for lower cost.

a bit OT:

There is a german model plane motor tuner (powercroco) who tuned his outrunner motor with custom 0,1mm lams. The efficiency only increases marginal percent versus 0,2mm lams because between every lam there is a little gap with air and so there is in the end less of iron in the stator. His coclusion was that the best efficient motors have 0,2mm lams.

 

[Bluefang]

I am probably throwing something abit from left field and completely wrong but for those who are not smart enough to build their own controllers(myself) would it be worth using this with the basic controller that Arlo linked at the very beginning with the chip? http://www.futurlec.com/dsPIC30F4011_Controller.shtml

Then using that to power a standalone power stage pulled from a Honda Hybrid, the MDM is a IGBT based power stage which if you get one from a Civic its 600V200A IGBTs and 400V capacitors. Some guy talks about using it as a controller for AC motors using his basic controller and shows a link for a basic BLDC controller that someone else seems to have had working(can not find any info). http://www.myelectricboats.com/ControlBoxHonda%20Inverter_CharlesMathys.pdf

Extremely high voltage and some freaking awesome control. Hope i am not too far off the mark.

 

[Lebowski]

Interesting, using the power/driver stages of a car inverter...

About the PCB Arlo linked at the beginning, this is unusable. The pinout of my chip is different... On my chip certain pins are used for
certain functions, and these are different from what the PCB in the link has. You would need to build your own PCB, but this can
be done even with universal hole-PCB.

The car inverter... if it just takes 5V digital inputs for the 6 power transistors, then it should be OK to use it. Problem though is that
my chip needs the information from the 3 current sensors in the wires going to the motor. The sensors are already in the
Honda inverter linked in your post, but are the voltages of the correct amplitude and polarity ? Are the signals available on the
plug of the Honda part ? Does the Honda part generate all the voltages it needs for itself ? I'm thinking about the intermediate (15V ?)
supply for the IGBT drivers and the supply for the current sensors ?

Lastly, my chip only works with BLDC motors, not AC induction motors.

 

[Bluefang]

will only be running it as BLDC, just saying that the guy who i first found playing with the honda MDM was running AC motors some how from a BLDC power stage.

Pitty about the pcb, i did not notice anyone saying it would not work i was reading too much too fast.

http://voltswagen.blogspot.com.au/2010_09_01_archive.html Is a basic diagram of someone elses project for one of the output plugs. Pretty sure all the sensors feed the MCM thru another plug

Does this help with info on the current sensors http://endless-sphere.com/forums/viewtopic.php?f=7&t=22246 5th post down, took me a while before i remembered it

Derek
Fyi I have not touched one of these power stages, i was just looking into it recently and then remembered about your controller side. Thought they might make a nice match for some high power fun

 

[fechter]

The current sensors on a Honda run off 5v and are bidirectional with an output of 2.5v at zero current.
I'm not sure about the IGBT modules, but the datasheet is available. There is lots of information about this part on InsightCentral. The whole power stage is pretty cheap from a junked car.

 

[Arlo1]

Lebowski would this help?? http://www.ti.com/ww/en/analog/dataconverters/inductance-to-digital-converter/index.html?DCMP=sva-psp-ssp-ldc1000-en&HQS=sva-psp-ssp-ldc1000-em-lp-en&sp_rid_pod4=Mzk1MDk1MDMyMTAS1&sp_mid_pod4=42613634&spMailingID=42613634&spUserID=Mzk1MDk1MDMyMTAS1&spJobID=203378870&spReportId=MjAzMzc4ODcwS0

 

[Lebowski]

No, I think it is a chip made for inductive sensors. I think you could use it
to make an accurate speedy by inductively measuring spokes in a wheel as
they pass a coil. Kinda like an old fashioned metal detector. But I don't think they
are meant for measuring motor inductances.

Based on the internal signal levels etc, there's no problem measuring inductances
with the controller IC...

 

[Bluefang]

Lebowski, would a current sensor with a 2.7V base line work with your chip properly? You talk about one with a 2.5V i am hoping the 0.2V drift wont matter as then we already have a power stage build by the Honda team
0amps = 2.70v
12amps = 2.80v -12amps = 2.57v
60amps = 3.21v -60amps = 2.17v
90amps = 3.48v -90amps = 1.92v
120amps= 3.76v -120amps = 1.65v
150amps= 4.01v -150amps = 1.42v
180amps= 4.24v -180amps = 1.20v
Is a table of outputs from the Honda Hybrids MDM. The units will still need a 15V power supply aside from the signals from the controller but that should be easy to setup.

Also do you still have afew chips as i would like to get one to play with.

 

[Lebowski]

The 2.7V instead of 2.5V is not a problem. The sensor seems to be linear which is good. I think the main
point is the direction of the current, what do you call positive and negative current ? I have set it up such that
when the current flows from the controller to the motor the current is positive (and the output sensor
voltage higher than its 0 level, see also schematic in first post). And vica versa, when the current flows
from the motor to the controller it's negative and the output of the sensor should be lower than it's 0 level.

If the Honda module has this reversed please let me know as it's an easy change, but it is VITAL !!!!!!

I have some new chips just in from a Digikey order last week, so I can mail you some if you want.
You have to let me know about the current direction though. I ask 20 chf (swiss francs) per chip and 5
for shipping (paypal). Please PM me if interested.

Just wondering, do you know the correct PWM frequency for the Honda module ? Is it high or low active ?
Both these can be set in my controller IC so no problems there, but I'm just wondering...

 

[zombiess]

Lebowski, would a current sensor with a 2.7V base line work with your chip properly? You talk about one with a 2.5V i am hoping the 0.2V drift wont matter as then we already have a power stage build by the Honda team
0amps = 2.70v
12amps = 2.80v -12amps = 2.57v
60amps = 3.21v -60amps = 2.17v
90amps = 3.48v -90amps = 1.92v
120amps= 3.76v -120amps = 1.65v
150amps= 4.01v -150amps = 1.42v
180amps= 4.24v -180amps = 1.20v
Is a table of outputs from the Honda Hybrids MDM. The units will still need a 15V power supply aside from the signals from the controller but that should be easy to setup.

Also do you still have afew chips as i would like to get one to play with.

Even if his software won't work with it, it's easy to fix. Just build an op amp differential amplifier (voltage subtracter). This will let you bias the signal and set it to 2.5v. It's much easier to handle in software, I hate having to do analog circuits (not my strong suit) unless speed is critical. It's hard to beat many analog circuits for pure speed with digital, it's why they compliment each other so well.

If you still need a different over all range after that just run it through another op amp with the appropriate gain, you then have the ability to dial in voltage offset and voltage range of your sensor for around $2 in parts. I believe an LM358 would work OK since it has a slew rate of 0.3v/uS meaning it will add 0.3uS delay to the signal rising per volt. This would be an issue on a gate drive, but I would think it's OK for a current sensor.

Take what I say with a grain of salt as I'm no electronics wizard, I just dabble and have had to do this before, but only on slow stuff.

Lebowski, my driver stage construction is around 40% complete now.

 

[Bluefang]

Zombiess, what you just wrote was mostly mumbo jumbo to me

Lebowski, I am 99% sure that the current is in the right direction from reading afew things over on insightcentral forum so it should be a start at least as i have no chance of building my own power stage. Niels has been kind enough to pm me his Eagle diagram of a board with parts list so it should just be a matter of following directions and putting everything together. Yay for compulsory high school electronics where we learned how to build circuit boards Is it going to matter at all to the controller if i am running the power stage at ~300+V will i need to put a voltage divider on the voltage sense for the controller, change resisters?

 

[Lebowski]

Regarding the motor voltage sense resistors:

If you look at the manual page 8, it shows resistors Ra coming from the motor and resistors Rb connected to the middle
of the two 2200 ohm resistors and the 100nF cap.

Rb I would keep at about 33k.
Ra should be chosen such that you have a few mA flowing for full battery voltage

So with 400V I would make Ra 100 kOhm.
400V and 100k gives 1.6W losses (but it's only connected to Vbattery half the time)
so on average 0.8W losses. If you get 2W or 3W resistors you should be fine.

 

[Bluefang]

Cool, thanks for the info Between all the info you, Niels, Arlo and Zombies are putting into this thread its making it seem alot easier. I ll try and order the parts this weekend to make up my board so hopefully everything is ready for when the power stage and chip arrive.

On a serious note, Lebowski if you can get sensor less high torque start working you are a Legend Should make water cooling the motor feasible.

 

[HighHopes]

ya, i'll agree to that. not many in the industry can successfully do a zero start on sensorless in a reliable way. good luck!

 

[Bluefang]

The one part that had me worried was collecting the actual arts to build up the boards. Turned out to be super easy, took the bom produced from the eagle design file that Niels sent me to the local electronics shop.

Came back 1hr later and they had collected everything for me to make 2 boards including enough kit to make the boards themselves. Total cost $60 with quite a few spares

Lebowski will 2watt 110kOhm resisters work as they did not have 100k.

 

[Lebowski]

Yeah the value is not super critical, with the 400 volt you'll be using anything between 80 and 150 kOhm would be fine.

Maybe you can start your own build thread with pictures etc ? Would be interesting

 

[Lebowski]

ya, i'll agree to that. not many in the industry can successfully do a zero start on sensorless in a reliable way. good luck!

With the recumbent working as it should I've started looking at the sensorless startup under load.

I've run some preliminary tests which look good (meaning I get data about the motor position out).
First thing to solve though: in addition to different inductor values the code also sees the mismatch
between the gain of the current sensors/fixed motor winding differences. This 'gain' error is
indistinguishable from the motor inductances being not equal... but with the difference that it
generates a constant vector while the position dependent motor inductances generate a rotating vector.

First thing to do is to build models in Labview to see if the idea I have for the problem above works....

 

[Bluefang]

Is the recumbent running on sensorless atm? Really excited to hear you are making progress on it

I have got most of the parts together and just waiting for the chip to arrive so i can measure it and start printing/building the board up. I ll start a new thread about it when i have everything.

 

[Lebowski]

Yep, the recumbent is running sensorless...

 

[crossbreak]

the induction difference of the coil groups should not have much influence if you only measure the change of inductance... shouldn't it?