mounting hall sensors to R/C brushless motors

liveforphysics said:
I just bought a pair of the 72v units
Click to expand...

Hopefully you were aware before hand that there is a larger Infineon, which uses 18 fets:
http://endless-sphere.com/forums/viewtopic.php?f=6&t=6475&start=195#p141307

There's a few threads about it on E-S too if you hit up the search box.
 
So, efficiency goes down the pan, in high speed mode......?
Click to expand...
well....i think efficiency should be the same or even better, since your winding resistance is only 1/2
the downside of split phase, i figured, might be
1. the utilization of expensive switches is only 50% of all time, not cost effective.
2. you're effectively using only half of the motor in high speed mode.

compared to a normal brushless motor that uses only 1/2 of the turn number, this split phase motor give 2x torque at low speed (assuming using the same amount of current), but same high speed performance
 
voicecoils said:
liveforphysics said:
I just bought a pair of the 72v units
Click to expand...

Hopefully you were aware before hand that there is a larger Infineon, which uses 18 fets:
viewtopic.php?f=6&t=6475&start=195#p141307

There's a few threads about it on E-S too if you hit up the search box.
Click to expand...


I didn't know they sold a bigger one! Do'h! I guess I will just buy a pair of the bigger ones too. I think with a little work, these smaller ones may be capable of doing a few hundred amps, and fitting in a smaller size/weight package though. At least it would be nice :) Thanks for the tip!
 
You dont need anything bigger.
Just put 4110's in the ones you have.

-methods
 
Is this anything more than a tapped winding?
Click to expand...
yea, i think so, but i think people would tap from integer slots so that it's more balanced and easier to access.

in Y connection, 1/3 of the windings are at rest
in delta connection, 2/3 of the winding are at 1/2 effort.
in split phase, 1/2 of the windings are at rest at high speed mode.

not one of them is perfect...
 
There is a perfect winding option. There are even handy compact DC controllers made for it in E-bike voltage and power levels.

It's a 6-phase motor and 6-phase drive. So far, I've only heard of it on certain servo motors, and this electric scooter company.

http://www.efun-ev.com/
 
liveforphysics said:
There is a perfect winding option. There are even handy compact DC controllers made for it in E-bike voltage and power levels.

It's a 6-phase motor and 6-phase drive. So far, I've only heard of it on certain servo motors, and this electric scooter company.

http://www.efun-ev.com/
Click to expand...

I assume that would be 60 degrees between phases. Sort of like comparing a V8 to a 4 cylinder.

They claim 95% efficiency, that would be impressive :D
 
What's wrong with just bringing out all six wires, and doing external delta/wye switching? Are you saying the hall sensors would have to be phased/offset differently?

-- Gary
 
What's wrong with just bringing out all six wires, and doing external delta/wye switching? Are you saying the hall sensors would have to be phased/offset differently?
Click to expand...
yes, the hall sensors have to be placed 30 electrical deg offset between the two connection methods. you can find that the synthesized magnetic fields of the two connections are offset by 30 degree.
this patent diagram also mentioned about this offset:
http://www.patentstorm.us/patents/6493924/description.html

nevertheless, i think we could still use only one set of hall sensors, so that in wye connection they're correct and in delta connection, they're placed 30 degree advanced.

one more thing is that, delta-wye switching is hard to implement on-the-fly using mechanical relays.
On-the-fly delta-wye switching could be implemented using thyristors/SCRs for faster response. similar circuit could be found from oak ridge national lab's DMIC (dual mode inverter control, http://www.ornl.gov/~webworks/cppr/y2001/pres/115367.pdf) but they conclude that using 6 SCRs on the controller is too expensive. SCRs are, as i found from online stores, expensive and bulky...
 
What are your thoughts on running all 6 wires out of the motor, and connecting all 6 wires to separate channels in the controller, perhaps linking a pair of controllers together to make a 6-phase controller?

In my limited experience, the 6-phase option seems to be the no compromises method to do a motor, and less stress on each channel of a controller as well. Best starting torque, best midrange, and best high speed, all while holding efficiency in the +90% range.

Am I going to have to buy a damn E-fun 7kw scooter and take it all apart just to get a controller we can reverse-engineer it? I bet CNCaddicts motors would LOVE to have a 6-phase controller driving them. :) :)
 
What are your thoughts on running all 6 wires out of the motor, and connecting all 6 wires to separate channels in the controller, perhaps linking a pair of controllers together to make a 6-phase controller?
Click to expand...

Wouldn't this still be a 3-phase motor, but with both ends of each winding available, i.e. not a 6-phase motor?

My guess at a 6-phase motor would be 6 separate windings, set up as two 3-phase sets offset by 60 degrees. So you would need two three-phase controllers, which would potentially work independently, although would be smoother if they were synchronised but phase-shifted by 60 degrees.
 
I think that wouldn't escape the compromise of a 3-phase motor, just smooth the pulses a bit at the cost of complexity.

I think they use a 3-phase motor with its 6 un-combined windings, and feed them to a controller that switches them in a sequence to always be optimized. But, that is just my guess, as I do think that would enable the motor to always have best optimization for efficiency and power, which would be solving/eliminating the 3-phase delta-wye dilemma. No proof though, I've never even seen one of there 6-phase hubs in person. Just my $0.02 guess.
 
I spoke with Bob from Astro Flight about dual controllers in a 6 phase motor. He said the windings would "Talk" to each other confusing the back EMF of each controller.

Does that sound right?

Matt
 
Aren't sensored controller blind to back EMF though? That would/may only effect sensor less on brush less 6-phase right?

I thought about it some more.

Our motors now just do 3 windings, and then to control the 3 windings from only 3 wires, we have to tie them together to make a common. There is no place to tie them together that doesn't have it's own inefficiencies. You just get to pick high speed or low speed to perform better.

With 6 wires, you could have complete timing control and polarity control over each coil. You could attract and repel at exactly the right times every time a magnet passes by, without having to energize half of some other coil somewhere that isn't helping.

If I'm not mistaken, with the right controller, this would mean you could effectively more than double the torque potential for a given motor, by attracting and repelling at the optimized points, and not wasting power on a non-contributing coil. Does that seem correct, or am I looking at things all wrong?
 
I think it would require a much higher resolution sensor than just 3 halls. I know the servo motors I've seen that used 6 motor drive wires also had an optical disk sensor on the back of them that had a pattern with maybe a hundred little slots in it.

Which just gave me a thought...

Maybe you could build a controller for a 6wire motor in a kinda caveman fashion. The work would all be in the optical disk design. Design a disk that would be physically connected to the rotor, with 12 optical sensors and 2 sets of slots in the outside of the disk. One set of slots for each polarity. Cut the slots in a manor so that the FET bank is switched on when the optical sensor gets signal that the slot passage is open, and turn off when not. Have a bank for positive switching and a bank of negative switching, and obviously make sure the disk doesn't have any overlap in the slots, as that would result in fireworks :) The sensor would BE the logic of the controller. To control power/RPM, you would need a PWM signal to be what gets passed through the optical sensor, and make the PW% be your desired throttle input.

Now that I've throughly rambled, you guys and get back to this great thread.
 
I am not an engineer. But, I thought about this a while back. Making the disc is the easy part from my end. The electronics would be the tough part.

Hmm, you (or Methods) can make the electronic control and I can do the machining on the motor. :mrgreen:

Matt
 
guys,
what you mentioned above is very common in servo drive industry called "optical encoders and vector motor drives". yes, they are important motor control advances, but they are also complicated and require extensive parameter matching b/t the controller and the motor. IMHO, not an winning choice.
 
George- That would be the best way to control a motor though right? I'm not an expert, but I don't see any drawbacks other than the amount of work to initially create the setup. Am I missing something? I think you could have massive and efficient torque, right from zero RPM. And then high speed operation would also be stronger than a delta motor, and more efficient.

I think you know where I'm going with this. Splitting the windings up in a hubmotor, and having Matt make an awesome encoder wheel solution, could enable a hub motor to be efficient and have monster torque at zero RPM, and have super top speed power as well. Obviously a lot of work, but it seems like that could definitely be a huge improvement for a hubmotor, or any motor.

Matt- I was just hoping you would volunteer to make the wheel part of the thing :) I would be spending days with a sharpy and a drill press :( Methy has his EE masters, and specializes in digital stuff, he could do the interface between the encoder wheel and the throttle/PWM to feed the FETs, and I could make the FET banks, power buses, sinks and cooling design (I'm a datacenter power generation engineer specializing in cooling). I could also help to bank-roll the project.
 
I think I just need to send Methy a HV110 to mess with. If it really is "Easy" to amplify the controller signal to drive more FET gates, that may be the hot ticket.

I should give him a call and see if he is up for that.

Matt
 
Going back a few posts - to run Garys setup we could instrument with 6 hall sensors in two sets 30 degrees out of phase and just switch those at the same time we switch the main phase wires with any number of off the shelf $0.10 parts. Just TTL signals. In microcontroller time, those mechanical parts are moving at a snails pace.

As far as doing something crazy... I will say this:
If someone knows what the timing should be there is no doubt that I could write the code to drive any motor at any speed in any combination.
Though like georgeycc pointed out, it will be implementation specific - one controller - one motor

The uControllers I am working with these days have a Meg of ram, floating point math, 60 I/0 lines, A/D's, D/A's, counters, comparitors, and the kitchen sink all built in.
They run at 24Mhz, are programmed in C++, and draw less than 1uA standby, generally drawing less 10uA average for a complex program.
They can wake up from sleep for an interupt in 1 microsecond
Running a motor is childs play for these things. Slow as dirt they are.

So if someone wants to wind an old X5 I have a blown X5 I could donate
If someone else wants to instrument that with hall sensors go ahead
Then someone can design the high power board as I think that ish is voodoo. I am a digital guy.
We can get a motor guy to spec out all the timing at lay out the high level requirements for the uC.

I can write code to make **ANYTHING** do whatever I like but....
but...

All that sounds like a whole lot of work :|

I usually favor hacking existing products
In the past I have found that investing a great deal of time in a quasi new idea always lands me dissapointed when some company in China starts selling them for $5 the day after I am done.

I sure like your enthusiasm though! :wink:

-methods
 
I'm confident on making the power board. I don't have the skills to do the fancy back EMF sniffing stuff, but I can throw together a very robust power switching setup.

If I make the power board, you won't have to worry about it failing, unless your programming causes it to fire the positive feed and negative feed to a coil at the same time... Then all bets are off on anything. I think we need some simple external logic gate to act as a filter to prevent that from happening.

You really down for doing the encoder disk and sensor mounts Matt? I think making them as small as possible would be the hot ticket here, so you can save the program/design and easily adapt it to any motor. Or at least any motor with the right number of poles.

If we go hubmotor, I think I could re-wind your cooked X5, but I don't know if your magnets would still be good or not. If I re-wind that thing, I'm going to add a pass of thin copper tubing making a lap around the stator, and then make a custom way over sized hollow axel and oversize bearings, then make a bracket off of that point that enables it to mount in a standard bike frame. That way we can circulate water through that sucker and keep it cool even if we figure out a way to pump 20kw into it.

BTW- I made a new 2kw/h LiPo pack for my bike, and it can sustain 50kw and burst 70kw. It's all encased in custom ghetto-thermoformed polycarbonate. I think it would be able to pretty much handle all the power we could pump into this ghetto 6-phase monster project.
 
sounds like a giant project is going on here :D

btw, anyone tried the hall sensor mod recently? i am interested to know how other motors perform with the sensors.
 
You must log in or register to reply here.

Similar threads

S
Detecting misaligned hall sensors with high-lvel benchmarks
Replies
1
Views
544
amberwolf
amberwolf
methods
Front hub mounted into rear drops
Replies
4
Views
1,419
methods
methods
S
Hall sensors, how to?
Replies
5
Views
745
sbin7zaim
S
ElectricGod
Testing Hall sensors
2
Replies
29
Views
9,519
ElectricGod
ElectricGod
adrian_sm
Worlds Worst Dyno - Version 2
3 4 5 6 7
Replies
154
Views
18,099
crossbreak
crossbreak
Back
Top