Mounting Hall-sensors on Hyperion ZS4045-10

[Zapro]

Hi.

I have converted a Kickscooter to use an Hyperion ZS4045-10 motor http://www.hyperion-world.com/products/product/HP-ZS4045-10

It's being run by a Hobbyking 150A RC Car ESC http://www.hobbyking.com/hobbyking/..._150a_high_performance_brushless_car_esc.html

I have made my own clone of this sensor-board: http://e0designs.com/products/hall-effect-sensor-board/ and the sensors are working.

Running the motor without the sensors connected to the ESC, gives me this waveform on the sensors:


I see no problem with the signals.

When connecting the sensors to the ESC, i can make it turn in 2 permutations on the wires, but it's stuttering, sounding like something is grinding, and the phase current is quite high.

I have tried all combinations of phase wires ABC, CBA, CAB, BAC, ACB, BCA - nothing is "really good"™

Is there any reason for this to not work ?

TIA!

// Per.

 

[gwhy!1]

it just dont look right to me.. was you picture of all 3 channels realtime ?..

it should look like this.

 

[Zapro]

I obtained the readings by connecting the hall-sensors to my Logic Analyser, and just let the motor run open-loop.

The three sensors are spaced 22 degrees apart. The motor has 12 poles and 14 magnets. It should work...

Disregard the order of the three sensors - according to your diagram 2 and 3 should be swapped. I've tried all permutations without change.

As fas as i can see, The angle my sensors is at, is wrong as the phases are not 33,3% interleaved. Can you agree to that observation ?

I'll go back to the drawing board and see if i can figure out what i f*cked up with the PCB-layout - if i'm not mistaken, the sensors should be 17,14 degrees apart, and not 22

// Per.

 

[Zapro]

I've dug out an old sensored in-runner (Hyperion Swift) and yep, the phases are 33,3% overlapping.



Guess i will have to get back to the math, and figure out how to space the hall sensors when the bell on the motor is 49mm.

// Per.

 

[gwhy!1]

I obtained the readings by connecting the hall-sensors to my Logic Analyser, and just let the motor run open-loop.

The three sensors are spaced 22 degrees apart. The motor has 12 poles and 14 magnets. It should work...

Disregard the order of the three sensors - according to your diagram 2 and 3 should be swapped. I've tried all permutations without change.

As fas as i can see, The angle my sensors is at, is wrong as the phases are not 33,3% interleaved. Can you agree to that observation ?

I'll go back to the drawing board and see if i can figure out what i f*cked up with the PCB-layout - if i'm not mistaken, the sensors should be 17,14 degrees apart, and not 22

// Per.

Yes the sensors needs to be 17.14 apart for a 12 pole 14 mag motor , the order will depend on what your order of phase wires are connected, so yes, the hall outputs A,B and C maybe a completely different sequence but the relationship between the signals are the same.

 

[crossbreak]

i thought the 17.14° are only true if connected in Delta? Maybe it is WYE terminated?

Have a look here, maybe that helps http://mitrocketscience.blogspot.de/2011/08/hall-effect-sensor-placement-for.html

 

[gwhy!1]

i thought the 17.14° are only true if connected in Delta? Maybe it is WYE terminated?

Have a look here, maybe that helps http://mitrocketscience.blogspot.de/2011/08/hall-effect-sensor-placement-for.html

the 17.14 is to emulate 120d spacing , if the sensors are adjustable then they will work for either wye or delta

Edit:
I prefere 30 mdeg with the center hall flipped, its a little more forgiving for any errors in hall placement and makes it easier to tune. The down side is that it has a bigger footprint.

 

[Zapro]

Well, i've re-spun the PCB, now with 17.14 degrees spacing on the sensors.

Now the motor runs smoother (when it wants to run), but it's still very erratic :-( I guess the Hyperion motor is too well made, as there isn't enormous amounts of flux on the outside of the bell. Metallic stuff do stick to the outside of the bell, but not insanely much.

I've done two runs with the motor running un-sensored for sniffing the signals - one slow RPM, and one fast. It seems my sensors miss some steps, and that must be reason that the motor runs erratic with sensors connected.

I'm using some "N177 419" from my junk box. Maybe i need to get some more sensitive sensors? I could also go the way of optical pickup?

Attached is a picture of the test-setup and readings.

View attachment 2





// Per.

 

[gwhy!1]

seems odd that it appears to be only the channel C hall that is missing :? I found that if a hall sensor switches then it will always switch. I would try a different part number for the sensors initially or swap the channels around to see if channel C hall is faulty.

Edit:

When you took these readings was the the motor powered running? or was it spun by another means. The reason I ask is I did have some problems with one type of sensor I was using that it was to sensitive and it was picking up the magnetic field from the energised coils, but intermittently. optical will work but introduces another set of problems that needs to be dealt with .

 

[Zapro]

seems odd that it appears to be only the channel C hall that is missing :? I found that if a hall sensor switches then it will always switch. I would try a different part number for the sensors initially or swap the channels around to see if channel C hall is faulty.

Edit:

When you took these readings was the the motor powered running? or was it spun by another means. The reason I ask is I did have some problems with one type of sensor I was using that it was to sensitive and it was picking up the magnetic field from the energised coils, but intermittently. optical will work but introduces another set of problems that needs to be dealt with .

The motor was driven by the Hobbyking ESC, and the Hall-sensors mounted passively "looking" at the bell (like the pic). I could try to redo the test, while spinning the motor by another means (battery operated drill)

I have some other sensors at hand, some Honeywell SS443A and some EW-510, maybe i could try them...

EDIT: I've driven the motor with an electric drill, and yep, the third sensor is iffy, see pic.


// Per.

 

[gwhy!1]

just re-looking at you last set of signals... the drop outs are symmetrical... channel A, was the sort of thing i was getting with the sensors that were to sensitive , but channel C there is compete edges missing :?

 

[Zapro]

I've now tried with the other sets of sensors. The SS443A is no better than the first one i tried (117) but the EW-510 is very sensitive, and triggers very consistently.

Unfortunately, the legs on them prevent me from mounting them flush to the board, so i will have to do a little filing - having them stand in free air does make adjusting the angle quite tricky.

As you can see on the video, the motor now runs, and changes direction when the sensors are shifted side-wise. If i move the sensors too long from the bell, the ESC gives up, and reverts to Back-EMF commutation. Right now i am testing with a very puny 15V 3A lab power supply to avoid blowing stuff up - i have tried stopping the motor by hand, and there isn't very much torque available, which there should be - but i'm really afraid to put on the main battery, and maybe smoke motor+ESC.

View My Video

As you have experience with sensored motors, what does the video tell you - does it look "sane" with that kind of operation ?

I shot a pic of the EW-510 sensor reading too:


// Per.

 

[gwhy!1]

your motor in the video sounds fine and is behaving as i would expect, the signals also look very good, I would say that this was good to go. Do you have a e-bike controller ? as this will have current limiting. when setting up use a amp meter inline with the battery ( and maybe a 10A fuse ) and only use very small throttle movements , you need to tune the motor so it have the least amount of current draw. Run @ 1/4 throttle first and adjust for min current then try at halve throttle, keeping a eye on the current and try to adjust to see if you can get the current lower , do this in steps upto full throttle. From my experience it is possible to get the no load current running sensored a bit lower than running sensorless but with the 17.14 spacing it is very difficult to get it spot on but good enough.

 

[kfong]

You can go optical, there is a thread on that approach.

 

[gwhy!1]

You can go optical, there is a thread on that approach.

unless the optics are enclosed and there is no chance of contamination of the disk or optics then optical can work, but for 99% of ev use its just not practical. Halls will be a far more reliable solution