mounting hall sensors to R/C brushless motors

[lesdit]

I would be a little hesitant to mount hall sensors with glue direct to the slots, as it would be hard to tune the position if there were some unanticipated flux distortion. On a board would be better, since it could be rotated slightly to tune.
If it is not possible to get a clean signal from hall sensors, we can always use optical sensors ( IR led/photo transistor pairs ) looking at a simple aluminum encoder 'chopper' disk. Not sure how much room there is in there, from the photos posted, and I don't have an Astro motor to look at yet.

[liveforphysics]

Just like the clyte's, they will have trouble with delta-wye switching the moment they are running with sensors, as the sensors will be 30deg adv or retarded after the switch. Make sure he checks efficiency with the sensored controller after mounting sensors. My personal experience showed me just an 1/8th" of shifting my hall sensors would cause major motor heating.

I think the right way to do this is to make a board that is a nice snug fit into the back bearing cover of the motor, and on the board have 6 tiny surface mount hall sensors, along with a pair of 4-channel op-amps (only 3 would be used on each one) that output to trigger a set of 3 tiny output FETs. The shaft of the motor would have a loc-tite press fit thin little magnet ring made to pass over the board in the back. Then you would have the normal +-5v, sensA, sensB, sensC, along with a wire to activate the switching of sensor position for delta-wye switching. The board could normally be set to power the WYE mode, and when it that lead get's tied to gnd, it could pull down the voltage feeding the WYE op-amp, and switch a little PNP on to turn on the other op-amp reading off the WYE positioned hall sensors. All these parts, includeing the hall sensors could be surface mount, so the whole populated board thickness could be about 3-4mm, and a 6 lead JST connector could be mounted somewhere on the back, with a little slot milled in the end-plate of the motor to let it poke out.

That way, you could have correct sensor position for delta and wye, and only need 1 simple extra wire to trigger the sensor position swap. Total components cost would be <$10 Single sided board, dab the pads with solder paste from large bore needle syringe, place the components by tweezers, toss in the toaster oven, and poof! Sensor board. I bet they could be assembled by hand in 5-10mins each.

[dontsendbubbamail]

liveforphysics,

Do you still have your hall effects mounted on the outside using the pipe fittings?

Bubba

[liveforphysics]

Yup.

[fechter]

Bob also verified that the 32-series motors have no problem doing delta-wye switching. He's tested this a number of times, most recently on a blimp that had a big 32" prop.

-- Gary

That's good to know. Like I always say, one test.....

Stuffing them in the slots should be fine if they can fit. Dealing with the wires is a bit of a PITA, but if he's going to do it, no worries. A board might still be a nice thing for retrofitting existing motors.

I'm not sure if switching from delta to wye causes a timing change. It seems like the relative position of the magnets to the poles wouldn't be changing, so I'm not sure why the timing would change. It seems to work out fine on a 9C hub motor.

On a little Ametek inrunner motor, what they did was extend the rotor a bit so it's longer than the pole pieces so it can face directly to the hall sensors. I think a similar position would work with the AF motor, even with the stock rotor.

Ametek hall sensors 1.jpg (57.42 KiB) Viewed 976 times

Ametek hall sensors 2.jpg (92.27 KiB) Viewed 976 times

[GGoodrum]

Bob said that the timing will shift 7-1/2 degrees, because two adjacent slots will be connected in series, in delta mode, but that at lower speeds, this shouldn't be a problem. What needs to happen is to optimize the timing for delta mode, at full speed. That will keep the heat down and the efficiency up.

I'm with Richard, we just need to try it and see. I'm just happy as hell that Bob is finally going to do something.

-- Gary

[gogo]

OOPS

[methods]

I'm with Richard, we just need to try it and see. I'm just happy as hell that Bob is finally going to do something.
-- Gary

I am with Gary and Fechter - We are way past the time for pie in the sky ideas about the best possible solution.
The 20 of us have been sitting around here playing with our butts for months now - A few have made minor progress - but for the most part we are stagnant.
I am ready to do some real testing.
I dont need all that fancy Delta-Wye switching.... I just want to be able to run my 3220 7T off of my 10KW controller.

I will find the limits of the 3220.

After we prove it works, demand will go up, Bob will see that building for ebikes pans out, and then we can work on the most elegant solution.

-methods

[johnrobholmes]

I am still waiting on my 3220 with halls. It just limits the motor so much to be sensorless, there are so few controllers large enough to work properly.

[swbluto]

The funny thing about this effort is that most of the motivation is to adapt to limited existing products, rather than creating what is provably possible. It seems possible to get the sensorlessness excellence of the Castle Creations controller and the brawn of an Infineon controller, but yet that product doesn't yet exist. Why isn't it being created? It seems understandable that most of the "powerful controllers" are used to power powerful, relatively slow motors and so the sensorlessness seemed less practical as it doesn't function well at low RPM and thus it's not really been pursued, but it seems a demand for it is growing given the need to powerfully power smaller, more powerful much faster spinning motors and both smart sensorless algorithms and powerful controllers exist...

However, I think I might speak unfairly as I believe another thread is aimed towards mixing brawn with the brains, but it seems a purpose designed controller would be better performing and functional. I guess the question is, is the demand large enough to motivate someone to pursue it?

[johnrobholmes]

I guess the question is, is the demand large enough to motivate someone to pursue it?

Yes, I believe so. It would be worth it just to do short runs for personal use IMHO. I think we can get a solid performer in a volume of one cubic foot, over engineered for reliable long term light EV operation.

[recumpence]

This is the whole reason I am running twin motors in my builds now. I am sick of waiting for a single motor solution, wether a large sensorless controller, or a sensored motor.

Twin motors and controllers are expensive and compicated to implement, though.............

I, too, wish Bob (or someone) would get off their freakin butt and get us a high power sensored setup!

Matt

[johnrobholmes]

You have sold enough astros to make an influence on them! You are limiting your client base as well!

[liveforphysics]

On a little Ametek inrunner motor, what they did was extend the rotor a bit so it's longer than the pole pieces so it can face directly to the hall sensors. I think a similar position would work with the AF motor, even with the stock rotor.

Ametek hall sensors 1.jpg

Ametek hall sensors 2.jpg

Fetcher- I think the black ring you are seeing is a ferrite magnet ring, and not an extension of the rotor's own magnets. From a cost perspective, as well as a functional perspective, it would make more sense than extending the motor's rotor magnets.

This is essentially the same thing I suggested above, only with a single set of hall sensors rather than a set optimized for both delta and star.

The funny thing about this effort is that most of the motivation is to adapt to limited existing products, rather than creating what is provably possible. It seems possible to get the sensorlessness excellence of the Castle Creations controller and the brawn of an Infineon controller, but yet that product doesn't yet exist. Why isn't it being created? It seems understandable that most of the "powerful controllers" are used to power powerful, relatively slow motors and so the sensorlessness seemed less practical as it doesn't function well at low RPM and thus it's not really been pursued, but it seems a demand for it is growing given the need to powerfully power smaller, more powerful much faster spinning motors and both smart sensorless algorithms and powerful controllers exist...

However, I think I might speak unfairly as I believe another thread is aimed towards mixing brawn with the brains, but it seems a purpose designed controller would be better performing and functional. I guess the question is, is the demand large enough to motivate someone to pursue it?

Come over sometime and I will show you an 18fet IRF4110 sensorless ESC and a 6fet IRF 4468 (monster TO247) RC brain equipped controllers (build using HV100 Version 4 brains). Both function, but both struggle to keep sync much more than they did before modification. I have very little spare time, but once I get whatever bugs keep them from holding sync figured out, they should be pretty neat. They will still always have the limits of low speed torque, and violent starting from a stop. A sensored application would really be a lot better solution IMO.

[lesdit]

Bob of Astro sells these motors mostly to the government. Think UAV. Think shipping pallets.
When I talked to him, he hardly remembered Matt's name. He did seem to like e bikes, however.

He didn't seem too understanding when I mentioned that an industrial motor needs an industrial class controller.
Maybe China will drink his milkshake.

You have sold enough astros to make an influence on them! You are limiting your client base as well!

[liveforphysics]

Something that may be handy for those seeking to add hall sensors to RC motors. Every CD rom, and most every CD player and most every brushless PC fan have a compact 3-hall sensor printed board included right at the base of the motor. Using the ferite ring magnets and holders used by some of these devices, it would only require de-gausing the ring, and then re applying the correct magnetic field to act as a sensor wheel setup for whatever motor. That last step could be very difficult though. Glueing in readily available tiny magnets to mimic the needed flux patter would be pretty simple though. Not really a production solution, but it would be less labor intensive than the method I used.

[methods]

Bob of Astro sells these motors mostly to the government. Think UAV. Think shipping pallets.
When I talked to him, he hardly remembered Matt's name. He did seem to like e bikes, however.

He didn't seem too understanding when I mentioned that an industrial motor needs an industrial class controller.
Maybe China will drink his milkshake.

You have sold enough astros to make an influence on them! You are limiting your client base as well!

Yep - the money is in government contracts - not one-off quacks that want to do 70mph on a bicycle
We are lucky that he even talks to us...

There is a sweatshop here in town that I go to to have cables built for work when our fab shop gets into a rut.
These guys are all sorts of friendly when I show up with the government procurement card...

Would you like to guess how they will treat me when I show up with a box of 7 raggedy looking half built ebike controllers from china talking about how I want to build them up as cheap as possible and sell them to my geek friends on the internet at a loss ?


-methods

[Miles]

He didn't seem too understanding when I mentioned that an industrial motor needs an industrial class controller.

You could try being less patronising?

[kfong]

http://www.maxcim.com/MEGAMax.html

Has anyone looked into these motors? They are already sensored. Not as big but they won't have any syncing issues.

[Miles]

http://www.maxcim.com/MEGAMax.html

Has anyone looked into these motors? They are already sensored. Not a big but they won't have any syncing issues.

Yes. Do a forum search for "maxcim". They're very good, but quite expensive. I almost got one in place of my first Astro, 2 years ago...

The mounting isn't ideal, for us.

[GGoodrum]

Bob of Astro sells these motors mostly to the government. Think UAV. Think shipping pallets.
When I talked to him, he hardly remembered Matt's name. He did seem to like e bikes, however.

He didn't seem too understanding when I mentioned that an industrial motor needs an industrial class controller.
Maybe China will drink his milkshake.

Bob is a great guy, and is one of the early pioneers in doing brushless motors for RC planes, etc. I've known him for about 7 years, but he's only remembered my name for the last coupe, or so. If it weren't for the fact that he got killed by the influx of cheap Chinese outrunner imports, he'd probably be doing a lot less industrial/military work these days. He even got burned by a guy that was going to do ebikes and he ordered 25 3210s, just before last Christmas. Bob and his crew worked over the holidays to get the motors ready and then at the last minute the guy cancelled the order, because he decided to go with cheap Chinese imported motors. It took me awhile to get him to warm up to the idea of doing anything for ebikes again, especially if it involves any significant investment on his part. I believe this was the main reason why he has dragged his feet on doing the sensored version. He just didn't want to spend the money doing the hall board, etc., for something that would have an undifned market. I think this week we broke through that barrier, when he decided that maybe putting the halls in the slots would work. He can try this with just what he has in hand.

He's also getting the ebike bug, now, I think, because he told me he went to BestBuy, who just started selling a few models of grossly underpowered ebikes, and he went for a test ride. He told me the performance was pathetic and wouldn't get the bike up most driveways, much less a small hill. I told him that's why we want to do this right, with a good motor. I told him I'd come up this week, and let him take my Port Runner for a "spin". I think I'm going to have to lock it in 3rd gear, though, so he doesn't end up on his ass, like I've been doing. He's older than me and has a lot less butt-padding.

-- Gary

[Miles]

He even got burned by a guy that was going to do ebikes and he ordered 25 3210s, just before last Christmas. Bob and his crew worked over the holidays to get the motors ready and then at the last minute the guy cancelled the order, because he decided to go with cheap Chinese imported motors.

I hate it when people do that....

[lesdit]

My conversation with Bob was very friendly and on a sympathetic note.
Miles, you're confusing my comments here with other conversations that are *not* here

He mentioned the bestbuy bikes to me as well, and how they could not even go up the ramp to the parking structure.
He never said that the motors he sells are too powerful for ebikes, or anything, so that's a good thing.

He didn't seem too understanding when I mentioned that an industrial motor needs an industrial class controller.

You could try being less patronising?

[gwhy!]

for anyones information, I made myself a hall sensor block to retro fit on the outside of a outrunner ( very similar to Lukes approach ), and i finally got a myself a small Infineon controller ( a real baby one ) , So now came the time to find out what hall matches up with what phase winding. Time to dust off the good old oscilloscope ( havent used it for years ), Sh*t I could find a second scope probe , anyway in the end out of frustration I just used trial and error and a current limited psu ( just in case ) I didnt get it right first time the motor just refused to spin it did try though but it did try to pull a lot of Amps, I didnt get it right the second time either but it did run but very noisy and also was trying to pull a lot of amps, third time lucky and it ran so smooth but was still trying to pull a lot of amps. I will dig out another scope probe when im back in work to see what is going on with the high current draw but I think it just may be just the timing needs adjusting ( the motor spins at about a million miles a hour and draws about 6 amps at WOT unloaded ). The point of putting this info up was to hopefully show peeps that you dont need a scope to find what halls goes with what phases trial and error will be ok as long as you current limit you power source or at the very least have a small 3a inline fuse so things dont get out of hand should you get it wrong.

[methods]

I was under the impression that the standard method is trial and error. I am an EE but I could not tell you how to do it with a scope
There are several spradsheets around here to use... Keeping track is critical as it is easy to get turned around if you dont write down what has been tried.
There are also quite a few tips - little tricks that can get you to the correct configuration in only 2 steps if you can ... say.. get it to run smooth backwards.

Given that there are 36 possible permutations - 1 guaranteed to work, several likely to work - I usually get it within 6 tries.
I dont use current limiting but I do use a current monitor and a soft touch on the throttle.

I guess that is not saying much though since I dont use current limiting ever -> I figure the worst I can do is V/R

-methods