comprehensive guide to DIY BLDC controller?

[PhoenixOSU]

I have been writing a design log / app note as I design my controller in a modular fashion, updating as i go through each piece. Would anyone have a use for this? I've seen some interest in older posts about people wanting a controller design that is modular so that people could replace the 3-phase bridge or regulators to suit their needs.

Anyway, if anyone is interested, here is the draft-in-progress. I have a handful of schematics and a ton of reference PDFs on the topic of BLDC drive design which are also great sources of information. Is there a sort of library for this info or should I zip it all together when i'm done (in a month or so) and post it?

Also, I want to do this with a microcontroller and provide programming guides and source code and pin out info so people can use it as an ASIC, if anyone is interested and willing to help me design this, I'd be happy to collaborate.

 

[jsplifer]

Hello,

This was a great read. Thank you for writing and schematizing.

Scares the pants off me though as I'm just a layman. However it gives a solid direction to take with the first few steps.

I'm very interested in your project and am enjoying all your posts. I learn something new in each one.

I did have a bit of trouble viewing the Word document on my mac though, missing the images, perhaps you can export to pdf in the future? I am also interested in your scavenged pdf files relating to this project. If you have the time to post them somewhere it would be greatly appreciated.

So its off to the drawing board for me. I use a program called Dia for rough-out design, and I've also grabbed a spice simulator.

Look forward to more of your posts on this subject,
jsplifer

 

[PhoenixOSU]

Here is all the PDFs I’ve found that have helped explain a lot about motor controller theory. Hope they help others as much as they have helped me.

I’ve decided to buy an 18 FET Infineon, since the design is so similar to what I’m building anyway it’s a shortcut that will save me on cost in the long run. But I will continue my design over time. I also need to scale back a bit as summer classes and work are starting to demand more of my time.

 

[PhoenixOSU]

More files

 

[PhoenixOSU]

the last of them.... so far

 

[PhoenixOSU]

There are the app notes from Microchip about their motor control microcontrollers. If you want to use a uC, or just want to get to know how the controllers work, this will teach you a great deal about how controllers work.

 

[Anonymous]

.

 

[liveforphysics]

I'm very into your project!

Good stuff!

I would be interested to help support your efforts to make a sensorless control system, like an RC controller, only setup for switching banks with huge 100v FETs.

 

[PhoenixOSU]

I've found that its not hard, the difference is just in changing the configuration of the 3 phase bridge and adding back EMF sensing. I don't need BEMF sens on my own controller, but I am kind of interested in having them as a fail safe for the "oh shit a hall sensor failed" case when riding around. I'm getting the feeling that a microcontroller with a large program memory is going to be needed as well. I've started to toy with the idea of doing my own full design with 100V 190A 4mOhm D2Pack FETs from IRF (IRF 4010) this way it is easy to just add more FETs and redo the layout with a bigger (longer probably) bridge section, and possibly a different driver IC. I am swamped until tuesday, then I'll start redrawing schematics and posting them for review section by section.

 

[Drunkskunk]

The thing that gets me is why so many people are desiging these to continue using hall sensors.

Its possable to design a controller to use the back EMF in place of the halls, even for an instant start. RC planes, boats, and cars have been doing it for a while now.

 

[rkosiorek]

The thing that gets me is why so many people are desiging these to continue using hall sensors.

Its possable to design a controller to use the back EMF in place of the halls, even for an instant start. RC planes, boats, and cars have been doing it for a while now.

i spent the morning watching a couple of guys flying a big twin engine flying boat of of the Moira River. out of a less than a dozen launches both motors started up reliably every time. but a couple of times one motor or the other started in the wrong direction. no problem easy to fix, rapid shutdown followed by a quick restart and all was well. bad controller? maybe? but the reaction of the flyers was that this was not unusual behavior.

i'm just not sure if i want a backward running motor happening to me on my bike in traffic. just like with the sensors, sensorless is not a perfect solution either.

the other issue that i see is that the sensorless motor is limited to a minimum low speed. it has to turn fast enough to develope a large enough back emf to be detected. low speed control is sacrificed. of course you'll never notice it if you always start up with a wheelspin, burnout or wheelie. just be prepared that once in a while you will launch in the wrong direction.

rick

 

[johnrobholmes]

Sensorless instant start does not provide full torque at zero RPM like a sensored start does, and the minimum speed for sensorless commutation is a theoretical 1/12th KV. I like the ability to start up from zero RPM, but nothing says we cant have a controller that switches to sensorless mode or has it as a backup.

 

[rkosiorek]

Sensorless instant start does not provide full torque at zero RPM like a sensored start does, and the minimum speed for sensorless commutation is a theoretical 1/12th KV. I like the ability to start up from zero RPM, but nothing says we cant have a controller that switches to sensorless mode or has it as a backup.

i think that is what the Golden controller does.

according to the instructions for the controller you can also wire the phases in any order. you can then set the controller into a "learning" mode it will then go through a proceedure for it to "discover" the correct wiring sequence. all without having to play with the wires. i'm interested to hear how that works in practice.

rick

 

[PhoenixOSU]

If you flip any two hall wires you reverse the motor direction. So there are really just two options and no "wrong way" as long as the next state decode logic is correct, you can't blow it up or damage it by switching two hall wires, it just reverses direction. I think the same is not true of phase wires though, this could cause the field not to be correctly lined up to the rotor, and this could be very bad.... but I've never looked at what happens if I do, wouldn't be hard to figure out just draw some pictures and state tables and "what if" cases.

 

[Drunkskunk]

i spent the morning watching a couple of guys flying a big twin engine flying boat of of the Moira River. out of a less than a dozen launches both motors started up reliably every time. but a couple of times one motor or the other started in the wrong direction. no problem easy to fix, rapid shutdown followed by a quick restart and all was well. bad controller? maybe? but the reaction of the flyers was that this was not unusual behavior.
rick

That can happen when the prop is too small or too light. its not uncommon on planes, especialy if they were using higher speed, lower thrust props. it almost never happens on buggies or boats, as they have more resistance to spinning on start up.

Older controllers do it more, too. newer ones, and better quality ones rarely do it.


The system works by making a guess at firing order, and sending a pulse to all 3 wires. its a very low current pulse, and the back emf is compared on all 3 wires. that gives the controller the information it needs to determin the right firing order to get the motor started in the right direction. the better the controller, the lower power pulse it needs to send, but without some kind of resistance on the motor, it may move. A super light propeller has almost no mass or resistance to the motor and may let it move quite a bit. a 3 pound RC car has far more resistance and isn't likely to go anywhere untill the controller decides which way is right. (many of them are running 400 to 1000 watts. Ebike power range.) a 250 pound bike and rider is going to provide even more resistance, virtualy eliminating any chance of a backwards start.

the odd thing about brushless motors, once they start spinning backwards, it doesn't matter which way the pulses from the controller are pushing it, the wave form in the motor will continue to spin it backwards.

 

[PhoenixOSU]

Are you sure the motor would keep spinning backwards if the pulses were pushing it the opposite way? This doesn't quite mate sense to me, as if the motor field is aligned orthogonal to the rotor, it will experience a real torque in that direction, and i can't imagine any other forces acting on it that would keep resisting the torque. You could do an experiment and roll a BLDC scooter backwards with the key in and then gun the throttle, but this could also be dangerous so I recommend taking some precautions first. I'd be curious to see what would happen, I'd use mine but its a stepper motor which is not quite the same thing.

 

[Drunkskunk]

they don't run full speed backwards, but up to about 2/3 speed, with 1/3 thrust, yes. I said it wasn't common before, but with one particular plane I have, it happens 1 time in 4, and often will do a few rpm backwards at Very low RPM before kicking forwards.
that motor is a very low weight magnet ring GWS 2808 running 3s lipo and a very light prop. I would say feather weight, but I've actualy weighted the prop against an equily sized feather. the propeller was lighter. I've not been too carefull on the flightline with that plane, gone for instant full throttle from off, and scooted the plane backwards down the runway.

If you think about the magnetic field like a wave being washed around inside a cylinder, it might make more sense. the wave is washing through at higher RPM than the motor actualy is turning, and as such, you end up with pattern between the wave and the magnets in the wall of the cylinder that can be 2/3 repeated and 1/3 opposed if the wave is reversed.
A trick my teacher taught me to see the pattern was with ASCII, using 1s and 0s.
you can see the way the 0s line up in diagonals down and to the right, but there is another pattern of down and to the left as well, only you would need momentum to carry the motor through and less efficently.

11011011011011011011011011011011
01101101101101101101101101101101
10110110110110110110110110110110
11011011011011011011011011011011
01101101101101101101101101101101
10110110110110110110110110110110
11011011011011011011011011011011
01101101101101101101101101101101
10110110110110110110110110110110

 

[mu2b]

Hi Thanks PhoenixOSU for your excellent articles. I have been unable to see any of the graphics so I wonder whether you could help me.
I have a 12V/115A hydraulic power steering pump with a brushless PMAC motor and a proprietary TRW controller. For my application, I will need either full on or off. No need for throttle adjustments.
The TRW control board has for components:
1. TDK ZJY-2P02 : Common mode CAN filter
2. TLE 7469 : Dual rail 5V regulator
3. TLE 6250 : Highspeed CAN-transiever
4. Infineon SAK-XC164CS-16F40F : Microcontroller
5. MLX 15120DB : BLDC FET Driver
6. "EMS" : FET package

The MLX 15120DB is proprietary so it needs to be replaced and the same goes for the Microcontroller. I was thinking about the ML4425 chip (diagram attached). I would use the
1. irfb4110 MOFSETs to replace the IRFR120 (Q7-9)
2. IRF4905 or UCC27200 or keep the diagram's FQD8P10 (Q4-6. I am not sure which one would be the best)
3. The 24V-80V power would be replaced by the 12V truck's battery power.
4. R2-4 should be replaced by 5kΩ?

This is a one time project so any suggestions would be appreciated.
Thanks you
Regards
Luke