Adding hall sensors to outrunners

[modern_messiah]

For an AF motor the Halls need to be mounted so that their active faces are towards the magnet poles, like the outrunners, so this means aligning them with their sensitive axis in line with the motor major axis, rather than at 90 degrees to it as is the case for an outrunner or inrunner.

I think this has clarified more than I was even asking - which is a good thing! So the sensitive axis of the sensor must be mounted into the magnetic axis of the poles. The physical position of the sensor is calculated based on the number of stator teeth, so the sensor is working out the position of the rotor based on the assumption that the sensor itself is physically placed at 120 electrical degrees (is that the right terminology?).

So the sensor is mounted to a "point of reference" (stator) and uses this information to determine rotor position.

*click*

The only problem I see now is that for a Axial Flux motor the sensors cannot be placed externally (like in Burties first post). Or maybe they can be mounted to the back of the motor instead of on top? Then again I'd prefer internal for a final build (more robust) but external is handy for testing...

Thanks!

[Jeremy Harris]

Finding a neat place for the sensors is always a pain. The Mars axial flux motors get around the problem by having a separate set of magnets to trigger the Halls, they don't use the main magnets on the rotor. You could fit the Halls internally, between the windings. Where you put them depends on the pole-pair/stator tooth configuration you choose, but you have some flexibility in that virtually all controllers will accept either 120 deg or 60 deg sensor spacing, so you can often find a position that can be made to work without too much hassle. If you have enough flux leakage from the backing plate, you could fit the sensors on the outside. This is how the external Hall method works on outrunners, there is enough flux leakage through the can to operate the sensor. The SS41/SS411 sensors operate at around +/- 40 to 60 gauss, so you don't need much leakage to make them work OK, as your magnets are going to be around 12,000 gauss I expect - just 1% flux leakage is going to give you around +/- 120 gauss at the back face of the rotor disc.

Jeremy

[modern_messiah]

you have some flexibility in that virtually all controllers will accept either 120 deg or 60 deg sensor spacing

For my test motor (technical drawings done - just not a good time of year to be sending them off to be cut ) I am using a 9 tooth 8 pole motor (bit odd but it is for testing purposes only). Thats 4 pole-pairs.

The spacing of those positions is 360/4 = 90 degrees and with 3 equally spaced sensors, 90/3 = 30 degrees between the sensors. So I can place these sensors 30 degrees apart on the back of the rotor (not attached to, but as close as I can without interfering with it's rotation). I have a feeling flux leakage won't be a problem...

[gwhy!]

Can someone please tell me how to figure out which slots to put the sensors in.... how do i know where 0 degrees timing is.... or 0-30 degrees or what ever timing would produce the lowest currents

I got one motor to work with 120 deg spacing, but the first motor that i did with 60 deg spacing on it, didn't work right... even with the middle sensor flipped...

Someone who has done this before please help, thanks

Hi Hum,

I have no idea why your 60degree ( mechanical )setup dont work, but some things to look out for is the center hall must be flipped and the controller must be set to 120 degrees or compatible or it will not work. (12 slot stator 14 magnet rotor, with the stock windings). Any correct timing/spacing will produce the lowest current.

[Jeremy Harris]

Ysterday I had my first problem with fitting Hall sensors to an outrunner. So far I've done several low Kv wound, wye connected motors, with internal Hall sensors fitted in every fourth slot (so 120 degree spacing - these were 12 slot stator, 7 pole pair rotor, motors). All have worked very smoothly with masses of low speed torque.

My milling machine has been running on a Turnigy Aerodrive SK6474 - 170 motor for over a year now, being driven by a cheap RC ESC. It's run faultlessly, but I've often wished for a little more spindle speed, which means upping the voltage to beyond that which a cheap RC ESC will run at. This meant fitting Halls to the motor so I could run it from a spare XieChang controller and a higher voltage power supply. I thought this was going to be a doddle, so fitted Halls as I've done several times before, like this:

Internal Hall mod top view.JPG (168.76 KiB) Viewed 667 times

You can just see the three Hall sensors set into slightly widened stator slots and glued in with epoxy.

I hooked everything up, only to find that the motor ran as rough as a badgers arse. The thing sounded dreadful, would trip the controller if the throttle was turned too fast or any sort of load was applied. Not only that, but it sounded like it had a misfire at some speeds, with the controller apparently just cutting out for a fraction of a second, fairly randomly. I suspected a dodgy sensor connection, or a bit of interference, so hooked it up to the 'scope. All the Hall signals looked fine, no matter what I did with wiggling wires or whatever, so it didn't look as if the problem was a poor connection or interference.

The motor is stock, so still connected as delta. This made me wonder whether my good fortune with all the other motors might have been because they were all connected wye. Unfortunately, I didn't want to reconfigure this one to wye, as I needed the higher Kv from delta to get the spindle speed I was after. I decided to do an experiment with variable Hall timing to see if this made a difference, using the idea that gwhy! had for fitting Halls. It was an hours work to make up a simple PCB to hold three sensors at the right spacing:

Hall sensors 63mm motor.JPG (41.77 KiB) Viewed 667 times

plus another half hour this morning to mod the milling machine motor mounting plate to hold the PCB:

Hall sensor board.JPG (196.11 KiB) Viewed 667 times

After a bit of playing around, I found that the roughness and cutting out problem was indeed caused by the Hall timing being out. This motor doesn't like running on internal sensors at the neutral position, as I needed to shift the timing by around 10 degrees from neutral (physical, ~ 70 degrees electrical) to find a point where the motor would run smoothly and accelerate cleanly .

It seems that fitting the Halls externally may well be the best way to do it for bigger, delta connected, motors, so this might be a useful tip for those looking to add Halls to standard, unmodified, outrunners.

Jeremy

[Thud]

Jeremy,
I can confirm your findings 100%.
I have also witnessed some other interesting issues in testing external Hall set up's
Here are my observations:

1) using the 17.14 deg. spacing can be finikey at times. It seems to work better on smaller motors.

2) On the big turnigys, in some positions the the adjustment becomes very sensitive & showes inconsistant performance under load.

3) Another thing that i found interesing is I can trigger the hall sensorer as far as .75" off the can.
(are ther potential benifits to a thicker flux ring?)

4) Once I made a 60deg hall spacing bracket It elliminated all weirdness on the 80mm motors. timing adjustments react as expected. & I can load the motor & not get the glitches that appeard previously.

once satisfied I designed a small universal motor mount for the turnigy motors.
This will snap into the motor's exiting mounting plate & be retained by the motor when attached to its mount.
The Halls will be on an ajustable bracket that bolts to the base unit. The whole unit is to be cast in some high temp urathane. I am awaiting some more silicone tooling rubber atm but here is a pic of the "base" Pattern for the mount:

60deg halls 003.jpg

60deg halls 006.jpg

60deg halls 007.jpg

This is my solution for a turn key Hall system applicable to the turnigy motors.
the Arc section will be aluminum cast in place.
input for design tweeks allways accepted.

[TylerDurden]

That's sweet!

Perhaps a step file?... so the cnc crowd can machine various versions/materials. Might be faster/cheaper than castings?

[Thud]

Hi TD,

To clairfy. I will make rubber molds from the patterns & then cast the aluminum bits into a hi-temp plastic. De-mold time is in the 5-10 minute range & the mold itself shold last for thousands of parts.

Here is a look at the proccess.
http://www.youtube.com/watch?v=cL33dO8mQuQ

Since I need more than a couple of these I am investing the time to make a mold for each 1/2 of the assembly.

[Jeremy Harris]

That looks to be a brilliantly neat way of doing the job, better by far than the rather delicate PCB method. I've epoxied over the components and copper on the PCB now, and it'll probably do for the milling machine, but a fairly easy to fit system like yours looks to be a better bet for motors used on bikes.

Is the intention to also cast the Halls inside the movable urethane moulding? If so, then that would make for a pretty tough mounting method.

I'm not surprised at the sensors working so far away. The SS411A sensors change state at around +/- 60 or 70G, whereas the magnets are around 13,000G, so even a 1% leakage through the can is going to give around 130G, around double that needed to switch the Hall. I'd expect flux leakage through the can to be around a couple of percent or so, and doubt they'd be much real performance gain by trying to reduce it further.

Jeremy

[Thud]

Is the intention to also cast the Halls inside the movable urethane moulding?

That is correct. I don't have that assembly pictured yet.
Need to make the jig to hold hall alignment in that pc for"potting" them into position.

A pr. of 4-40 screws will adjust it on the base mount just like your PCB example.

easy-peasy.

[AussieJester]

Thanks for the information Jeremy ... theres a number of you fellas using these type
of motors on your mills now... ThudSTeR's new mounting setup is superb, hope too get my mits
on one or two in the near future :: wink :: Top job Thud

KiM

[Jeremy Harris]

Thanks for the information Jeremy ... theres a number of you fellas using these type
of motors on your mills now... ThudSTeR's new mounting setup is superb, hope too get my mits
on one or two in the near future :: wink :: Top job Thud

KiM

My milling machine has done a lot of work on that small Turnigy motor, and now looks like it'll do a lot more. I did a test cut at low speed on some fairly difficult stainless steel tonight and was mightily impressed by the low speed torque. Now it's got sensors it will cut at low speeds that were unthinkable when I was using the old sensored controller, although that's not the reason I made the switch.

I'm now even more convinced that fitting Halls is really the answer to getting these motors to work responsively.

Jeremy

[oatnet]

Hi TD,

To clairfy. I will make rubber molds from the patterns & then cast the aluminum bits into a hi-temp plastic. De-mold time is in the 5-10 minute range & the mold itself shold last for thousands of parts.

Here is a look at the proccess.
http://www.youtube.com/watch?v=cL33dO8mQuQ

Since I need more than a couple of these I am investing the time to make a mold for each 1/2 of the assembly.

!!!! If you decide make these for sale, please sign me up for 3.

-JD

[katou]

I'm just checking in. Is this idiot proof yet? If so, let me have a crack at it, maybe you just need a better idiot.

Katou

[gtadmin]

I'm just checking in. Is this idiot proof yet? If so, let me have a crack at it, maybe you just need a better idiot.

Katou

[olaf-lampe]

Ysterday I had my first problem with fitting Hall sensors to an outrunner. So far I've done several low Kv wound, wye connected motors, with internal Hall sensors fitted in every fourth slot (so 120 degree spacing - these were 12 slot stator, 7 pole pair rotor, motors). All have worked very smoothly with masses of low speed torque.


You can just see the three Hall sensors set into slightly widened stator slots and glued in with epoxy.

Where exactly did you place the halls? Between the two Aa Bb Cc teeth? Would it help to place them 2 teeth further back ( or forth?) Would be 60 dgree difference


After a bit of playing around, I found that the roughness and cutting out problem was indeed caused by the Hall timing being out. This motor doesn't like running on internal sensors at the neutral position, as I needed to shift the timing by around 10 degrees from neutral (physical, ~ 70 degrees electrical) to find a point where the motor would run smoothly and accelerate cleanly .

Jeremy

[Jeremy Harris]

Where exactly did you place the halls? Between the two Aa Bb Cc teeth? Would it help to place them 2 teeth further back ( or forth?) Would be 60 dgree difference

I can't see why it should make a difference as to which slot the Halls are in, as their placement is relative to the edges of the magnet passing any stator slot, isn't it?

In the past I've just picked any random set of three slots at 120 deg to each other and it's worked fine.

Jeremy

[olaf-lampe]

Where exactly did you place the halls? Between the two Aa Bb Cc teeth? Would it help to place them 2 teeth further back ( or forth?) Would be 60 dgree difference

I can't see why it should make a difference as to which slot the Halls are in, as their placement is relative to the edges of the magnet passing any stator slot, isn't it?

In the past I've just picked any random set of three slots at 120 deg to each other and it's worked fine.

Jeremy

You're right Jeremy,
it isn't that important, but doesn't it make it easier to determine which hall corresponds to which phase?
For you it's not the first motor to mount halls on, so I guess you have a certain wiring-pattern, that always works?

-Olaf

[Burtie]

I cant quite work this out

Assertions (please correct me if wrong), Assuming 14 magnet, 12 slot motor driven by a 6 step controller:

1) The hall sensors respond to the position of the 14 magnets on the can.
2) Consider the symmetry of the magnets around the can. If we rotate the hall group by 51.4 mech deg (360 electrical deg), we would expect everything to work the same.
3) going a bit further, If we rotate the hall group by any multiple of 60 electrical deg (then sort out the wiring), everything should still work the same. (Because we use a 6 state controller.)
4) However, if we rotate the hall group by 1 slot, that is 30 mech degrees = 210 electrical degrees

Problem is that 210 is not a multiple of 60, so the timing must have changed by + or - 30 electrical degrees ??

Burtie

[gtadmin]

One's moving 1/7 of 360, the other 1/12 (stuffed that twice ), so the second will almost always be out of step with what's required.

[Jeremy Harris]

I cant quite work this out

Assertions (please correct me if wrong), Assuming 14 magnet, 12 slot motor driven by a 6 step controller:

1) The hall sensors respond to the position of the 14 magnets on the can.
2) Consider the symmetry of the magnets around the can. If we rotate the hall group by 51.4 mech deg (360 electrical deg), we would expect everything to work the same.
3) going a bit further, If we rotate the hall group by any multiple of 60 electrical deg (then sort out the wiring), everything should still work the same. (Because we use a 6 state controller.)
4) However, if we rotate the hall group by 1 slot, that is 30 mech degrees = 210 electrical degrees

Problem is that 210 is not a multiple of 60, so the timing must have changed by + or - 30 electrical degrees ??

Burtie

Glad it's not just me that finds the concept of 7 magnet pole pairs and 12 slots mildly confusing at times!

The Hall angular position datum is relative to the stator, as the Halls trigger commutation. The timing is relative to the magnet position. This means that the Hall sensors need to provide either 120 electrical degree or 60 electrical degree (relative to the stator) outputs. Because all slots are the same, in effect, the three Halls can be placed (at the correct angular spacing) relative to any of the 12 slots.

Timing variation is relative to the magnet pole passing position, so is angularly related to 360/7 per cycle, rather than 360/12 per cycle.

Jeremy

[gtadmin]

Glad it's not just me that finds the concept of 7 magnet pole pairs and 12 slots mildly confusing at times!

The Hall angular position datum is relative to the stator, as the Halls trigger commutation. The timing is relative to the magnet position. This means that the Hall sensors need to provide either 120 electrical degree or 60 electrical degree (relative to the stator) outputs. Because all slots are the same, in effect, the three Halls can be placed (at the correct angular spacing) relative to any of the 12 slots.

Timing variation is relative to the magnet pole passing position, so is angularly related to 360/7 per cycle, rather than 360/12 per cycle.

Jeremy

Yes, that's what I meant, but much more eloquently expressed by Jeremy

[Jeremy Harris]

You're right Jeremy,
it isn't that important, but doesn't it make it easier to determine which hall corresponds to which phase?
For you it's not the first motor to mount halls on, so I guess you have a certain wiring-pattern, that always works?

-Olaf

I always use a 'scope and a sensorless controller to find which wire is which, after I've wired up the motor. I don't bother to try and trace the windings and find out which Hall corresponds to which phase wire by inspection, as it's too much like hard work!

The technique I use is to power the motor with a sensorless controllers at a reasonable rpm, then I connect one scope channel to one of the motor phase wires and the other to Hall wires in turn (with them connected to pull-up resistors). It's then usually easy to see which Hall has edges that align with the phase switching point. By going through each phase in turn I can quickly determine which Hall corresponds with each phase. If I also note which direction the motor is turning, I can correctly label them so that I can then hook the motor up to rotate in either direction. My hack sensorless controller already has the three phase wires labelled, which makes life easier.

Jeremy

[olaf-lampe]

OK ,
I just ordered a 2 channel scope. Always wanted one, but never had a real reason. I hope it'll arrive soon, time is running out...
-Olaf

[Burtie]

...The Hall angular position datum is relative to the stator, as the Halls trigger commutation. The timing is relative to the magnet position. This means that the Hall sensors need to provide either 120 electrical degree or 60 electrical degree (relative to the stator) outputs. Because all slots are the same, in effect, the three Halls can be placed (at the correct angular spacing) relative to any of the 12 slots.

Timing variation is relative to the magnet pole passing position, so is angularly related to 360/7 per cycle, rather than 360/12 per cycle.

Jeremy

Thanks for the explanation Jeremy and gtadmin.
I just had a feeling that the stator slots were not all exactly the same for a 14 mag DLRK motor, since half of them are between different phases and half are in the center of a phase.

For Aa we would have in-phase, but opposing stator poles either side of the slot.
For ab we would have stator poles which were 120 deg out of phase either side of the slot.

If this were the case, the rotational symmetry for the stator would be every two slots i.e. 360/6 =(60 mech deg).
So it might make a difference which slot you choose as your timing datum.
I wonder if this could explain why your most recent motor conversion did not run very well when you randomly chose the slots.

If this were the problem, it may have shown when you looked at the Phase / Hall signal correlation during the setting up
with the sensorless controller.

I probably need to go and read some books on electrical machines...