mounting hall sensors to R/C brushless motors

georgeycc

1 W
Joined
Mar 9, 2009
Messages
58
Location
san jose, CA, USA
i think this experiment might help some people who are doing the same stuff. i did this conversion so that i can use a industrial motor controller that has armature current limiting. this also enables usage of Kelly controller to work with R/C motors.

i converted two motors, one is HTX style outrunner with 200rpm/V kv value and the other is hacker style slotless motor. interesting enough, the hall sensors could be mounter outside of the flux ring. the flux leakage is really obvious with r/c motors.

i am building a dyno to test if the controller really limit the armature current precisely to give near flat torque shape, hopefully it will work. the other good thing is that this HTX generate huge torque at ZERO speed :D


btw, the timing is not exactly accurate as you can see from the waveform, but that's adjustable tho.

file.php


file.php


file.php

timing not perfect
file.php


file.php


file.php


file.php


====================
 

Attachments

  • hacker2s.jpg
    hacker2s.jpg
    63.5 KB · Views: 10,848
  • hackers.jpg
    78.4 KB · Views: 10,188
  • htx1s.jpg
    htx1s.jpg
    90 KB · Views: 10,849
  • htx2s.jpg
    htx2s.jpg
    79.4 KB · Views: 10,848
  • htx3s.jpg
    htx3s.jpg
    83.6 KB · Views: 10,840
  • htx4s.jpg
    htx4s.jpg
    71.5 KB · Views: 10,824
  • htxcs.jpg
    90.3 KB · Views: 10,173
This is something we really need untill we get some decent ESCs for bike use (ESCs that are good for more than 3kw or so).

JohnRobHolmes did the same thing. We spoke about it last night on the phone.

I will be calling Astro today to ask about having some motors manufactured this way.

Matt
 
Nice work!

What is the toothed ring doing?

You can get hall sensors that are pretty sensitive, so the flux leakage will be enough to trigger them. On the inrunner, you could mount the halls on a movable ring and change the timing :wink:

Things look pretty tight on the outrunner. It might be possible to mill some small slots in the end bell to give them a bit more clearance. You could also probably mount them so they pick up the flux leakage outside the ring or aimed at the ends of the magnets.
 
Personally, I think this is a premature step backwards, and negates a big part of the one real advantage to using RC-based setups, which is smaller, lighter controllers. There are already RC controllers that will handle 100-200A. the biggest limitation I see is that for most models there is a 50V limit. This is starting to change as well. Most of the major ESC manufacturers have demonstrated new models with much higher voltage limits.

Putting in hall sensors, just so a $400 Kelly controller can be used, is not progress, in my mind.

-- Gary
 
hi fechter,
that ring is the retaining ring on the rotor that i had to remove to have better access to the magnets (it's aluminum so it's not part of the flux path). i tried several mounting positions before i came to this one. mostly the sensors work when the motor's static but got interference when the motor is running and it's not electrical interference b/c i ran the sensors on a independent circuit.
for the inrunner, the sensors position is indeed rotatable :D but adjusting the timing did not give significant rpm increase tho, not like using vector control and flux weakening by which i could usually get ~20% no-load rpm increase.
i would very much like to have the motor milled to look better but unfortunately i dont have access to the machine shops....will keep it as is until it breaks.

hi Gary,
unfortunately most of the r/c brushless controllers are not using decent control techniques to make correct use of current. the r/c controllers are not designed to work from zero speed to 10k rpm but rather for mid-high speed only. but for EV use, we really need the motor to work at very low speed and have armature current limiting so that you wont burn the coil/battery up the slope. that's the basic idea of this conversion.
 
Very clever!

I have a $700 120V 220A Kelly controller that is gathering dust.
I would not hesitate to implement this for a second.
I especially like the fact that you can pick up right off the case of the inrunner.

Make the call Matt!

I agree that sensorless controllers are not where they need to be yet:
Not in low rpm control, not in voltage handling, and not in power handling.
Their market drive is for lighter and cheaper. No demand for them to progress past 12S.

I dont think it is a step backward to allow us to use legacy equipment to drive new motors.
Not only are we already invested in the Sensored controllers, but there is currently no quality alternative.
The high power ebike market is not even 0.001% of the ESC market so I am not holding my breath for any manufacturers to come out with our dream controller.

Maybe 5 years down the road this will be a memory, but today.... Its time to do what we have to do.

-methods
 
I wouldn't call it "legacy equipment". I'd call it equipment proven to work. Even the experts are blowing these RC controllers, and that's exactly why I'm still sitting on the sidelines for that game. It's great that these ESC's can push thousands of watts and weigh less than a pound, but they're useless without durability. FWIW, Knuckles sensorless module for a regular controller lasted less than 5 minutes, and I was taking it easy, not pushing it, so don't count on that as a viable alternative either.

John
 
John in CR said:
FWIW, Knuckles sensorless module for a regular controller lasted less than 5 minutes, and I was taking it easy, not pushing it, so don't count on that as a viable alternative either.

John

That's a shame, I had high hopes that this system would work.

As the RC ESC market doesn't seem likely to come up with what we need (high voltage, high current, reliable and tough controllers), then it seems we have two choices. Either we modifiy an RC ESC to handle higher voltages reliably, or we look to add sensors to motors and use existing, proven, controllers.

Out of interest, just how much can one of the big Infineon controllers handle if fitted with decent mosfets (maybe 4110s)? An affordable sensored controller that's good for around 80 to 100V at around 80 to 100A is probably a good starting point for many people, as RC motors that can take more than about 6kW are pretty expensive. Can the Infineon be modded to do this, rather than something like the Kelly?

Jeremy
 
I have modified the teeny tiny 6 fet Infineon that is a credit card size.
http://cgi.ebay.com/48V-350W-brushl...ViewItemQQptZScooters?_trksid=p3286.m20.l1116
I ran 4110's
Soldered the shunt to 1.2mOhms
I have been running it at 48V 50A and pulling 2,400W all the way to the freewheel speed.
I can only imagine how much you could pull from a full sized infineon

Does not even get warm.
To bump up to higher voltage all you have to do is run the 100V caps and the any voltage mod that those characters came up with.

No doubt the Infineon controllers are a viable cheap option, especially with the programming capability
Only obnoxious thing about them is the slow start up.

Kelly still offers higher level functions that the Infineons dont:
over temperature control for the motor and controller
the option to treat the throttle as Speed control, Torque control, or a mix
Options to run 0-5V, 5V-0V, etc on inputs
Variable Regen and switch regen
Huge voltage range up to 150V and currents up to 400A
etc.

The Kelly is just another class of controller to my mind. 90% of people need an infineon but some of us like to have the extra goodies.

But that should not be the focus of this discussion...
It really does not matter what controller we use.
Any controller will work

I will be implementing this concept with a very large, very high power project that I have in the works.
I have been fretting over how to control the power and now I know :wink:

Thanks again Georgeycc

-methods
 
Very interesting about the small Infineon, thanks. From your "experiment" it looks like one of these cheap controllers may well be a good way to go to make a beefier version to drive a hall sensor modded RC motor. I've just ordered one of these small controllers from that eBay link to play about with. I'll dissect it and see just what can be done to get it up to around 5 to 8kW, which is the sort of power I'm looking for eventually (I want to drive a pair of the big HXT motors on the final application).

I have a couple of Towerpro 5330 motors (rated at around 2.8kW each) that I'm using as a small-scale experiment and might try this hall sensor mod over the weekend. My initial thoughts are to just mill some small slots in the base of the case to take the sensors.

Are those hall sensors just arranged at 60 degree intervals around the case? Are there any particularly critical alignment issues with them?

Jeremy
 
Are those hall sensors just arranged at 60 degree intervals around the case? Are there any particularly critical alignment issues with them?
nope, mine mod is 120 deg, you'll have to calculate the electrical degree of one slot, my example is 12 slots, 14 poles, so one slot is 360*7/12 = 210 deg, so if you put the 1st sensor at somewhere, the next one should be 4 slots aways (4*210=840, 840 modulate 360 = 120), and so on.
also you'll have to fine tune the orientations of the sensors. b/c these motors are not designed to accommodate sensors, there's no easy flux path to put the sensors. that's why my sensor positions look so weird.
the other thing, i am using latching hall sensors, to avoid glitches.

my controller was originally rated 50V/10A, and mod to 80V/80A with no problem.


-george
 
Well, while you are looking at controller mods, how about adding a capability to do wye/delta switching in the controller? It will be easy enough to get AF to add hall sensors and to bring out all six phase wires, and I think David said he could do this on his new motors, when they are ready.

-- Gary
 
hi Gary,
adding delta-wye switching sounding interesting...
what do you have in mind? i think using mosfets might not work since the motor phase potential switches the full DC range, it would violate mosfet's max gate-to-source voltage. 100V Vgs appear too huge for mosfets.
maybe SCR, i'll do some more research here.
also, for dumb controllers like the one i have in hand, i guess switching delta-wye does not have to sync up with other signals, since it's not using emf or other state variables...i guess it would be much easier.
 
Something to considder.

I originally bought a big Kelly to do my bike. I was going to mount sensors on 1 motor and run them in series with the rotors linked mechanically in time with each other. Then I found out that the kelly can only do 60,000 electric RPM, which in a 12 pole motor would cap me to 5,000rpm. I didn't want to sacrifice 1/3 of my RPM range, so I scrapped the idea and went with RC controllers that do 210,000 electric RPM, or 17,500 motor rpm.

Before you go too wild, make sure you have a sensored controller that is compatible with the speed range of the RC motor you are wanting to use.
 
For the AF
http://www.astroflight.com/index.php?main_page=product_info&cPath=9&products_id=122

60,000 / 8 poles = 7,500rpm
Kelly is out for that one
At a KV of 275 and 50V it would want to max out at 13,750

Kind of a moot point anyway since the motor is rated for 48V 100A peak for 1 minute
That would be like using the Kelly to drive one of those plastic Barbie cars :mrgreen:
I see the Kelly as the place to turn if you want to run High voltage, high current, or both.
I dont know what people are putting through these Asto Flight motors.... If they are rated anything like a NEU then a whole lot more.

A $100 Infineon controller would handle that motor just fine if it had hall sensors.

-methods
 
Does anyone know what the upper electrical rpm limit is for the Infineon controllers? I've read most of the posts on the various Infineon threads, but the closest I've seen to this info is a mention that it drove a 32 pole Puma motor to 3300rpm (I think, I didn't write the numbers down). This implies that it's got an electrical rpm limit of around 105,000.

If this is the case, then it would seem to be reasonably OK for a hall modded outrunner, like the 130kV HXT perhaps, up to around 67 volts or so (assuming that the motor would tolerate nearly 9000rpm mechanically).

It may be that the Infineon will go higher than this, it'd be good for those looking at fitting halls to RC motors if we could find out just how fast it will run.

Jeremy

PS: I'm going to have a go at fitting halls to one of my 215kV Towerpro 5330 test motors over the weekend, in the hope that my small Infineon controller will turn up next week. If I can rustle up enough batteries (and mod the controller for high voltage) I may be able to find out the real controller limit.
 
Methy- Don't pretend you wouldn't power a barbie car with a 150v 300a Kelly if it had the RPM capability. Then don't pretend you wouldn't drive it to work honking the pink barbie horn while doing wheelies on the freeway.


Jeremy- You're awesome. I'm excited to see your findings. Let me know if I can help with anything. If the infinion controller can be built to be a HV HC replacement for a kelly and do high RPMs, that would make me switch over to them. I'm going to buy a pair of them now, just incase it turns out they can be made to work. :)

Best Wishes,
-Luke
 
I just bought a pair of the 72v units:

http://cgi.ebay.com/72V-1500W-brushless-controller-for-E-bike-scooter_W0QQitemZ300276897869QQcmdZViewItemQQptZScooters?_trksid=p3286.m20.l1116

Both controllers shipped was only $217. Exactly the same price with shipping as the new turnigy version 4 sensorless controllers. LOL, and all 4 controllers are still a few hundred cheaper than my Kelly that can't do high RPMs cost me. If these things can be upgraded to perform at 200amps AND are capable of higher RPM, they should work out well.
 
hi liveforphysics,
i saw your pictures, your turnigy motor should be the same as mine, it's a 12-slot 14 magnet poles motor, right? then the rpm limit should be 60,000/7~=8500rpm.

-george
 
for the speed limit of micro-controller based controllers.
usually the micro controller interrupts at each PWM cycle. so if the pwm freq is only 8k Hz, then the theory max rpm limit is that it sees hall status change for every pwm cycle, i.e. 6 cycles for one electrical revolution. so 8kHz would translate to 8000/6*60=80,000 electrical rpm limit. however at this coarse switching, i suspect that the commutation control would inefficient so i think it would not be good to operate at near this speed.
i guess what i am going to say is that, you can look for the pwm freq and have a vague idea about the speed limit.
for my industrial controller, b/c it uses analog commutation circuit, so there's no this limit and i can adjust the pwm freq to any value.
 
Luke,
Call me up when you are ready to mod out those Infineons. I should have warned you that you have to ask for EMS shipping otherwise they will come on the slow boat. He has a minimum order at which he will give you the fast shipping (you were over that minimum). Prepare to wait a long time.

I will send you one of the programmers I made and share my "lessons learned" with you.
Let me know when you are ready to order fets. We will order together and buy a butt-load.
I think we should stick with the tried and true 4110's though there are some new options out there.
Watch out with the Infineon as the throttle input is fussy. It is tuned to Hall voltage range so it will cut out if you go to high
A couple of inline resistors before and after the pot will fix this.

Like I said before, only problem with the Infineon is the start delay. I need to trace the board and look for the delay cap.
I dont think I am going to find one though. . .

I am getting excited. I want to order some stuff too! I just need an excuse.....

-methods
 
I have both the magura pot throttle and a hall effect throttle, so I could use either one.

It's a date on the FET buying. We can buy a 100pack or something, or whatever makes them cheaper. I also need some Hall sensors.

But first, I'm going to give the new version 4 controllers a whirl. If they can sync and run smoothly being sensorless, that will kick-ass, and I wont need to mess with being sensored :)
 
liveforphysics said:
I have both the magura pot throttle and a hall effect throttle, so I could use either one.

It's a date on the FET buying. We can buy a 100pack or something, or whatever makes them cheaper. I also need some Hall sensors.

But first, I'm going to give the new version 4 controllers a whirl. If they can sync and run smoothly being sensorless, that will kick-ass, and I wont need to mess with being sensored :)

Check with Bob Mcree before you order. He can get 4110s for under $2.00, I think. Also, he was looking for people to do a bulk buy on some 150V FETs.

-- Gary
 
Back
Top