Adding hall sensors to outrunners

[gwhy!]

From what I have been playing with it looks like the 15 degree spacing is the best way to go ( no flipping hall sensors/signals ) using mechanical 15 degrees its far less disruptive to where the switching points are than what I am using ( 60 degrees mech ). The only thing I would say is as long as the leading edge of the first mag to the leading edge of the next is greater than 15 degrees then all should be well as long as the flux leakage isn't to great to cause false triggering.

[bigmoose]

I took a bit of the morning to run the flux leakage tests. The first pix shows the Turnigy C80100-130 in it's high tech wooden test fixture. The wires come out what I call the fixed bell. Station zero is the end of the outrunner closest to the fixed bell. The stations move towards the "prop end" in 0.25 inch increments.

For the first graph I aligned the center line (C/L) of one magnet with a pole. I measured down the center line of the magnet on the exterior outrunner shell. I then measured down the line marked 10 degress CCW looking into the motor from the prop end, 20 degress CCW, and finally 25 degrees Counter ClockWise.

For the second graph I turned the fixed bell 10 degrees CCW, and repeated the three measurements above.

For the third graph I turned the fixed bell 20 degrees CCW, and repeated the three measurements above.

My Preliminary Conclusions... don't you love engineers that work in "high risk" areas? Everything is "preliminary", until the project is successful.

From the first graph you can see that things are funky near the fixed bell. You have to get at least to station 5, or 1 inch from the fixed bell to get a decent signal. (Yes, Excel added a 1 to my station numbers. Excel station number 1 = Turnigy Picture station 0. You also need to stay clear of the prop end, but not by so much, s minimum of two stations or 1/2 inch.

If it were me, I would put the hall sensors over station 7, or 1.5 inches from the fixed bell towards the prop end. That looks to me like the preliminary optimum at this time.

I could just sneak my probe into the assembled motor to measure the surface flux on the magnets when they were between pole pieces, I measured right around 4900 Gauss. I could almost jam the probe into the air gap between the magnets and the pole piece, but I couldn't justify the cost if I ruined the probe, so I didn't do it!

I'll cut the plots a bit different and post them in a while.

[bigmoose]

Here is the same data plotted a little differently. Each plot is for a particular scan relative to the magnet center line, and then it has curves for the stator aligned with the magnet center line, advanced 10 degrees CCW and advanced 20 degrees CCW. There is a total of 4 graphs that correlate with the curves that were on the previous.

[liveforphysics]

*Luke runs over to his crappy hall sensor brackets and starts traceing a template out to make a spacer and raise them to the #7 middle position on the rotor bell. *

[bigmoose]

Luke, there is always a twist and turn in the road. I went back and looked at your twin Turnigy bike build thread, and it looks like I have the new Generation 2 Turnigy C80100-130; and your build might have used the first Generation. My silver outrunner bell looks longer than yours and is 3.130 in. ish from memory. Also mine has a big, like 2 inch ish bearing on the stator bell, I also have the screws on each end. (ish is a technical term that means I'm too tired to go back downstairs and remeasure it... so I go from memory!)

That said, I may have much more dead space in my version of the C80100 near the fixed bell than you do.

Looking at my flux chart it implies my magnets are around 1.5 inches long. Do you know how long yours are?

Always a twist, isn't there?

[liveforphysics]

Luke, there is always a twist and turn in the road. I went back and looked at your twin Turnigy bike build thread, and it looks like I have the new Generation 2 Turnigy C80100-130; and your build might have used the first Generation. My silver outrunner bell looks longer than yours and is 3.130 in. ish from memory. Also mine has a big, like 2 inch ish bearing on the stator bell, I also have the screws on each end. (ish is a technical term that means I'm too tired to go back downstairs and remeasure it... so I go from memory!)

That said, I may have much more dead space in my version of the C80100 near the fixed bell than you do.

Looking at my flux chart it implies my magnets are around 1.5 inches long. Do you know how long yours are?

Always a twist, isn't there?

You're right that my motors are the old-style with no lower skirt bearing like the fancy new style you've got. We both have identical sized stators though, so I would ass-u-me our magnets were also roughly the same length, and that your magnets just sit up a bit higher than mine to make room for the skirt bearing. This might mean my halls are ok where they are at height-wise.

[bigmoose]

We could use the symmetry of the old/new design, and index the measurements from the long shaft/prop end. That would place them approximately 1.63 in from the long shaft end of the outrunner housing. At first I thought we found something more profound, but it appears it is an artifact of the addition of the skirt bearing.

[Miles]

Came across this today - thought I'd post it here, in case there's anything to be gleaned from it.

[gwhy!]

Nice find Miles.
I think you just save my sanity ,
That anwsers some nagging questions that has been going through my head for the last 2 days....

[Miles]

Glad it helped

[gwhy!]

Burtie,
Can you (or anyone please) give quick explanation of the math to arive at the 34.3 deg?

Hi Thud,

I justified it to myself like this:

Given that we accept 120 mechanical degree spacing works for a 3 phase motor like this one.
The can has 14 evenly spaced magnets around it.
So for one revolution, the magnetic pattern repeats 7 times.
We could put a hall sensor in any one of 7 positions and it would give the same signal.

The spacing of those positions is 360/7 = 51.42 degrees.
But we need 3 equally spaced sensors, 51.42/3 = 17.14 degrees between the sensors (I think .)

Burtie

Hi Burtie and Thud,

Burtie Your spot on with this spacing @ 17.14 degrees mechanical ... Its taken a while for me to get my head around because of your opening statement about the magnets, but thanks to Miles ( his pdf find ) the penny has finally dropped on a few things . Thud, possibly a better way of looking at it is 2 magnets = 360 elec degrees which = 51.42 degrees mechanical with these motors ( 14 magnets ) so working back we need 120 degree electrical spacing hence the 51.42/3 = 17.14 degrees mechanical. Sorry if everyone else has already worked this out but I am a bit slow sometimes.

Edit:
Miles, the information in that pdf have just solved another issue ( puzzle ) I'm having with a new controller, you sir are a Star

[liveforphysics]

Great .PDF find Miles

That defiantly needs to get filed in the technical area. Very helpful info in there.

[Miles]

Other useful papers where that came from.....: http://www.kth.se/ees/omskolan/organisa ... ts?l=en_UK

[Thud]

Thanks guy's,
I have a clear understanding of this method used to trick the controllers into thinking your halls are on 120 deg. The math will change if you use 10 magnets or 20 magnets. (once drawn in cad it is clearly understandable )

[JEB]

Maybe I missed it, is there a place that covered some final thoughts on adding hall sensors to a Astro 3210 motor? (inside motor, or outside magnets,and halls, 17.~ deg, 15, 120, ???)

Also, will any brushless sensored controller work with the 7500 rpm of the motor? Suggestions on the best one to use?

[TylerDurden]

Astro 3210 sensors left-off around here: http://endless-sphere.com/forums/viewtopic.php?f=2&t=14210&p=231050#p231050

[gwhy!]

FYI
I have re-spaced my sensors to 17.14 mech degrees ( same as Burtie ) and it all looks good, scope signals are now spot on (unlike my previous spacing )and the motor runs smoother than before. I dont know if Burtie have done any load testing yet but I plan on doing some over the next couple of days on the bench. I have placed my sensors half why along the mag length ( thanks to bigmoose ) the hall signals do seem to be much less jittery than before. The real test will be pulling big amps and hope it all signals remains happy. I was pulling upto 100A+ with my previous spacing with no ill effects so fingers crossed this will be ok to.

[Tiberius]

I used the same type of latching sensors as Jeremy did, the Honeywell SS411A .

I was confused about this term "latching", as it struck me that what is wanted is a digital output but not latching. it turns out that Honeywell make a range of Hall sensors, with digital outputs, and that some of them are latching.

The SS411A is a bipolar, non-latching version. It will try to switch around zero magnetic field, but there will be some hysteresis which could cause a small phase shift. Its the one I would choose out of the range.

I believe other manufacturers make Hall sensors in similar packages but with analogue outputs. It might be worth doing the experiments with analogue sensors to find the best position - they will act like linear gaussometers - and then change to digital.

Nick

[gwhy!]

I used the same type of latching sensors as Jeremy did, the Honeywell SS411A .

I was confused about this term "latching", as it struck me that what is wanted is a digital output but not latching. it turns out that Honeywell make a range of Hall sensors, with digital outputs, and that some of them are latching.

The SS411A is a bipolar, non-latching version.

Nick

Hi Nick,
I am using the same halls SS411A and they do indeed latch . Strange that the data sheet you have pointed to says they are non-latching..

[Tiberius]

Hi Nick,
I am using the same halls SS411A and they do indeed latch . Strange that the data sheet you have pointed to says they are non-latching..

Its confusing. "Latching" means the output stays there when the input is removed, and it needs another operation to change it. But most electronic logic elements have a level of "hysteresis" which means that once the input has caused the output to change, it takes a certain minimum change on the input in the other direction before the output will change back.

In certain circumstances the behaviour will be similar. In this case a latching sensor will look like a bipolar non latching one with very large hysteresis.

As I said, its confusing and the Honeywell datasheets don't explain it very well.

Nick

[gwhy!]

Hi Nick,
I know what you are saying, A lot of the sites that sell them call them a digital Bipolar latch hall sensor so I never really questioned it.

[lutach]

I've been following this forum for a while now, but never registered. I must say, some of the folks here are very smart so I finally registered. I'll be following this forum much closer now and one day I'll try to post something of my own.

[JEB]

I purchaced some halls from Jamco, they turned out to be SS466A, they require 180G to flip, where the SS411A only require 60G, the small outrunner I tried the 66 with would change at about 1/4" away from the case, or less. More magnetic flux required to change, less chance of changing at wrong time?

[Tiberius]

... More magnetic flux required to change, less chance of changing at wrong time?

Yes, but the further from the zero crossing point of the field it will change. That will appear as a phase shift.

Nick

[Jeremy Harris]

Nick,

These sensors do appear to latch when used with the sort of magnetic field strength we're using. In practice, you find that the sensor changes state when a certain magnetic flux threshold is passed, with a given magnetic polarity. The sensor output then remains "latched" in that state until exposed to a similar magnetic flux with the opposite magnetic polarity.

This is a good characteristic for use as a motor sensor, where the controller is relying on accurate edge-sensing, as it means that the sensors are relatively immune to noise.

My experience, having converted three motors now, is that the output from these sensors, when used in this application, is very reliable. I was surprised at just how accurate the zero-crossing point of the three sensor outputs were, when compared to the trapezoidal waveforms from the motor being driven by a sensorless controller. With the motor delta wired, these were spot on, as close as I could measure on the 'scope (probably better than 5 electrical degrees).

Jeremy