Replacement Hall sensors in BLDC hub motors

barbill

10 µW
Joined
Aug 4, 2009
Messages
5
This is my first post here and I would like to introduce myself. I am an electrical, but not electronic, engineer from the UK with a task of getting both myself and my not inconsiderable beer belly over the hill to the pub and back again at a power assisted speed of 15 mph by lawful transport without the risk of losing my driving licence. I have ideas for doing this with a cheap imported direct drive PM hub motor and if anyone is interested in my thoughts and calculations I will post them later but first I would like to share my research regarding my current problem:

I recently removed the off side bell housing from my BLDC hub motor in order to shorten the threaded spigot upon which the sprocket cluster is mounted. I faced 3.8 mm off it in the lathe, reassembled it and gave it a spin to ensure it was free. I now find that all 3 Hall sensors are faulty. I can only assume that one or more of the Hall sensor wires was touching the motor phase wires when I gave my hub its spin and the voltage developed was sufficient to blow all the sensors.

The sensors are marked 41F 713 and I read from other posts that these or similar are quite common. Does anyone have a specification sheet for these sensors?

I ask because I read that others have had similar problems or problems with motor shudder that they have rectified by fitting Honeywell SS40 or 41 devices and this has propmted my own research. It would appear that digital Hall effet sensors come in 3 main flavours: Unipolar, Bipolar and true latching. Of these only the bipolar or the latching types would be suitable for the 50/50 duty ratio required by a brushless DC motor. Latching sensors would be ideal for reasons I mention below but it appears that many ordinary bipolar devices are being used because they once were much cheaper to manufacture.

Latching devices specify both maximum and minimum flux density operating points and also maximum and minimum negative flux density release points. They also give some sort of guarantee of equality of the switching and release points around zero gauss. Ordinary bipolar devices specify maximum operating point but not minimum. Likewise they only specify maximum negative flux release points and not minimum. They do however specify a minimum differential in switching flux and this apparently is quite accurate. i.e. the manufacturing process can better guarantee switching flux differential than flux thresholds. What this means is that ordinary bipolar devices are not guarantted to switch in positive flux but random samples may switch in decreasing negative flux. Similarly they are not guaranteed to release in negative flux but may release in falling positive flux. The result of all this is that ordinary bipolar devices do not guarantee a 50/50 duty ratio.

A data sheet for the Honeywell SS400 series can be found at http://www.honeywell-sensor.com.cn/prodinfo/magnetic%5Fposition/installation/p88700_5.pdf Comparing the SS413A bipolar with the SS466A latching device, we find that the ordinary bipolar has a maximum operating flux of 140 gauss and a minimum differential of 20 gauss. This means it could release at as much as 120 gauss; well within positive territory and producing nothing like a 50/50 duty ratio. The true latching device, on the other hand has a maximum operating point at 180 gauss and a minimum switching differential of 200. This means it must experience positive flux to operate and negative flux to release and is guaranteed to have a nearly perfect 50/50 duty ratio.

As I cannot find any information on my failed 41F sensors I have decided to replace them with Honeywells. As far as I can see, none of the SS40 and SS41 types are described as latching and the SS466A latching device is now only 9p (about 15 US cents) more expensive than the SS413A. I will order the SS466A devices at £1.11 plus VAT and I may even be rewarded for my initial carelessness with a smoother running and therefore more efficient hub.

barbill
 
i just ordered some more SS41 from mouser and can mail you some when they arrive next thursday. why do you think yours are dead?
Two of my sensors are permanently floating and the third is permanently sinking no matter what direction the flux.

Thanks for your offer but why would I want SS41 sensors after all I have said above?

Perhaps it may be easier to understand if you Google SS466A and SS413A PDF data sheets. They each contain charts and the former describes distinct maximum and minimum switching points. The latter only charts maximum and nominal points so any particular sample of SS413A, like the SS41, could operate or release even under the wrong polarity of flux.

The SS41 PDF data sheet avoids providing a chart but the minimum operate flux can be determined by subtracting the differential from the maximum. The SS41 and the SS431A could produce motor phase duty ratios as low as 60/40 depending on the particular sample whereas the SS466A is guaranteed to be almost 50/50.

barbill
 
I never thought about the hall sensors failing to switch like that before, and now that you did the thinking for me, it makes sense that it would increase the motors efficiency if you could make it switch 100% of the time.
Very good observation, because I don't remember anyone else talking about it before.
 
sorry i offended you by offering to help. why did you even post up if you already determined an identical replacement part would not satisfy your needs. i still don't know what the hell you meant about permanently floating, etc. but if you do that's great.
 
sorry i offended you by offering to help. why did you even post up if you already determined an identical replacement part would not satisfy your needs. i still don't know what the hell you meant about permanently floating, etc. but if you do that's great.

You did not offend and no appology necessary. The discussion I have opened is not easy to get ones head around. I opened it because I thought it may be of interest to others, not because I was looking for help in procurement of sensors and if I gave that impression then It is me that is sorry.

By floating I simply mean that the output is at no potential at all with respect to the supply. It will neither sink nor source. All the devices we are discussing are NPN output devices which means that they sink when operated or in other words the output switches to the negative potential of the supply.

What is more important is that my research suggests that the ordinary bipolar Hall sensor like the SS41 was originally designed as a cheap alternative for use in computer fans and the likes where phase duty ratios are less critical. In the case of speed freaks attempting to get the best efficiency from their bike motors, latching sensors are infinately more preferable and in my opinion could make a very significant improvement to both efficiency and smoothness.

barbill
 
barbill said:
<SNIP>

The SS41 PDF data sheet avoids providing a chart but the minimum operate flux can be determined by subtracting the differential from the maximum. The SS41 and the SS431A could produce motor phase duty ratios as low as 60/40 depending on the particular sample whereas the SS466A is guaranteed to be almost 50/50.

barbill

dnum, the SS466 is also a bipolar latching sensor.

the SS41 and the SS40 are perhaps the most commonly used sensors in small brushless motors. they are cheap and work well. dnum was just trying to be helpfull.

the sensors are placed in very close proximity to the magnet ring. on virtually all motors the magnets are placed closely together. on many the magnets actually touch. the magnetic transitions are very sharp, abrupt and powerful. the sensor is always immersed in this magnetic field. only the transitions are significant. there is little need for sensitivity. it is akin to measuring an elephants weight with a common bathroom scale.

but part of the fun is to experiment. do you have a scientific way to measure the difference between the common SS41/SS40 and the SS466? they are almost identical in price, so price is not an issue. no reason why they would not work, but i don't expect to see an improvement in the performance. then again i love surprises. give them a try an do report the results.

rick
 
rkosiorek,

I have no way of measuring the magnetic flux at any single point in the motor but the data sheets detail the operating points of the sensors.

Regarding the changing magnetic flux; again I don't know but I would assume that the sensor would be influenced more and more by the approaching opposite pole as it gets nearer. I think the flux would be sinusoidal but the easiest way to test would be to view the electrically generated emf of an otherwise disconnected spinning motor, on a scope. As the generated emf would be proportional to the magnetic field, so we could determine its shape.

Does anyone interested in this subject have a scope?

barbill
 
typical BLDC controllers are looking for a square wave from the hall sensors. the trasition being when the sensed magnetic field flips from one pole to the other.

BEMF waveform depends mostly on how the motor is wound. some motors produce a sinusoidal shape and others a trapezoidal shape. general BLDC controllers that do not use hall sensors employ some form of zero crossing detection to mimic a square wave output. so whether the BEMF is a sine or a trapezoid does not matter as only the zero cross is significant.

common BLDC controllers only use 6 phase states. there is no proportional blending of one phase state to the next. it is merely a hard switch.

the sensors you specify are digital devices. when the field reaches a trigger value they toggle on or off. nothing proportional about them. ther would only be a very small difference between them and the SS41timing. even then it would only affect half of the transitions. practically speaking there would be a greater error resulting from the crude methods used to mechnically secure and position the sensors.

theoretically you may be right. this may be an area for improvement or tweaking. the only thing i know is that Cyclone, BMC-MAC, Kolmorgen, Crystalyte, Nine Continents and Golden motors all use either the SS40 or the SS41 sensors in their motors. the SS466 is just another sensor belonging to the same family.

rick
 
rkosiorek,

I would like to stick to position sensing by Hall effect devices and I resist the temptation to reply to your control remarks and remarks on sensorless control. There is something I am still not getting across and the reason is probably my own poor explanation of the problem.

I agree that the Hall sensors produce a square wave and I cannot see how I have led you to think I believed it was proportional. In particular the controller wants to see a symmetrical square wave in order to produce a 50/50 duty ratio. Is that where I am going wrong? by 50/50 duty ratio, I mean each coil carries forward and reverse current for an equal angle of rotation.

The accurate detection of the magnetic zero cross is the crux of the problem and data sheets show that tolerances of ordinary bipolar Hall effect devices are massive. The flux necessary to switch one SS41 sample can be as much as 3.75 times greater than the nominal and because the differential is small in comparison, not only will it operate too late but it will also release too early.

A latching bipolar device by comparison has not only got a much wider differential between switching states compared to its maximum switching flux but it also has a uniformity of operation around zero flux.

Perhaps this fellow from Allegro can explain it all better than I can: http://www.allegromicro.com/en/Products/Design/an/an27705.pdf

barbill
 
Dude, throw your degree in the trash, and just fix the damn thing. Solder in some 41's and your good to go.
 
Soooo did the SS466's get installed and tested? :pancake:
 
This post comes up a lot when searching for hall sensors. So bump this up to the top for shitsngiggles.

Question is, how much out of spec is doable when it comes to Hall Sensors?

I try to get as close as I can to the SS41, so even though the first one, on the left is cheapest, I dont think it would work because of the 7V, so its just wiser to buy the SS41 compared to anything else as they are cheap, so just stock up.
https://sensing.honeywell.com/SS41-bipolar

Maximum Output Current: 4 mA 20 mA 20 mA 20 mA
Operating Supply Current: 14 mA 10 A 15 mA 10 mA
Operating Supply Voltage: 7 V 24 V 12 V 24 V
Supply Voltage - Max: 24 V 24 V 30 V
Supply Voltage - Min: 4.5 V 4.5 V 3.8 V

Hall Sensor.jpg
 
markz said:
I try to get as close as I can to the SS41, so even though the first one, on the left is cheapest, I dont think it would work because of the 7V,
Why not? Are you powering it from more than that?
 
Back
Top