Building an ESC

[rob77]

I know how CPUs can stop working for no reason at all. Using PWM just seems like a hack to me, and hacks are never good. I was looking for an elegant solution.

1. everything has a reason and actually i know what will be the reason of your project's fail ( no offense, just let's be realistic )
2. PWM is the elegant solution.
3. howgh.

[ziggyfish]

I know how CPUs can stop working for no reason at all.

 Do you seriously believe that?

Just kidding, although some software have mind of their own. I've had software crash and after a simple reboot everything works again, and never see the problem again. I know in theory its the programmers fault but sometimes you have to question whether the CPU is faulty.

[ziggyfish]

1. everything has a reason and actually i know what will be the reason of your project's fail ( no offense, just let's be realistic )

My solution works in a simulator, so I know I am on the right track with my design. I just need to buy the components now, and test everything in real life. I will need to rework some of the solution due to the real world not being perfect.

2. PWM is the elegant solution.
3. howgh.

Just think about it, all I want to do is convert a DC signal to a 3 phase output. The PWM solution requires a microcontroller, a 15557 line program, 3 ICs and 12 transistors. Not only that most ESCs also use PWM to communicate to the microcontroller, thus adding more complexity to the main CPU, and requires a specific programmer device to get configure correctly. To me that's not an elegant solution.

My solution only requires 4 ICs and output transistors which give me use of all 3 wires and everything is managed by the circuit, so the main CPU can just set and forget.

[David Hess]

Too little gate resistance with poor routing and youre overheating and blowing up FETs.

Or too much reverse transfer coupling for the gate drive.  Or too much source inductance for the dI/dT.

Do you have a scope though?  It would be rather difficult to diagnose and figure out exactly why youre blowing up FETs if you dont have one.

And some high voltage differential probes or maybe an oscilloscope which has isolated channels.  The probes could be more expensive than the oscilloscope.

When you find a limitation to that board that you cant fiddle with or tweak, then you move onto making your own since you know why you need, and what makes up a better one.

The layout problems become extreme when you have to simultaneously minimize inductance because of high dI/dT , minimize capacitance because of dV/dT,  prevent ground loops, and spread power dissipation out with multiple devices.

[David Hess]

My solution only requires 4 ICs and output transistors which give me use of all 3 wires and everything is managed by the circuit, so the main CPU can just set and forget.

I have this vision of Neo saying, "Transistors.  Lots of power transistors."

Pulse width modulation is trivial in the analog domain, requires fewer transistors to handle the power required, and minimizes obscure issues like secondary breakdown limits which are often nebulous.

[Precipice]

My solution only requires 4 ICs and output transistors which give me use of all 3 wires and everything is managed by the circuit, so the main CPU can just set and forget.

Yep. Everyone else who's ever built a BLDC controller is a simpleton. How could we have missed such an excellent solution?

More seriously, it sounds like time for you to start building stuff. Good luck! (Hint: I expect your solution to become more complex, and converge towards others, explodey lesson by lesson. Although there's the happy phase of 'it'won't start' first, followed by 'it won't stop speeding up'.)

A genuine hint, though, is to build yourself a test jig, with a DC motor and various resistors (or a proper load) to put loads on your motor. To start with, it's enough to couple the shafts with heatshrink, but eventually you'll want a proper coupler. (Unless you drive the DC motor, though, you'll not be able to simulate most realistic stuck starting modes, but it'll do for the first 80% of development)

Good luck, adventure awaits!

Just think about it, all I want to do is convert a DC signal to a 3 phase output.

You are aware that a stopped BLDC motor winding is just an inductor? The current will rise to destructive levels long before the motor starts to turn far enough to use the next phase.
And boom, you're down some transistors, or a winding.

[Kremmen]

[...]
My solution only requires 4 ICs and output transistors which give me use of all 3 wires and everything is managed by the circuit, so the main CPU can just set and forget.

Confidence is OK and desirable but then there is overconfidence and cocksureness. You figure out the rest.

[ziggyfish]

To all, thanks for all your help, and suggestions. Although the solution I have isn't based on PWM, you have given me more information than you think (for example calculating thermal heat, how a BLDC motor works).

I will start ordering the parts required next month and will post the final schematic when I have it tested.

Thanks for all of your help.

[Precipice]

I have this vision of Neo saying, "Transistors.  Lots of power transistors."

I'm waiting for 'my transistors blew up, where can I get bigger / better MOSFETs?'
or
'My breadboard has just evaporated'...
Still, building motor controllers from scratch is good and educational, it's hard to think of anything else so simple sounding that covers such a wide proportion of this fascinating game. From device physics through hard realtime software, just to do what some crappy brushes and copper segments do in a DC motor

[David Hess]

I have this vision of Neo saying, "Transistors.  Lots of power transistors."

I'm waiting for 'my transistors blew up, where can I get bigger / better MOSFETs?'
or
'My breadboard has just evaporated'...
Still, building motor controllers from scratch is good and educational, it's hard to think of anything else so simple sounding that covers such a wide proportion of this fascinating game. From device physics through hard realtime software, just to do what some crappy brushes and copper segments do in a DC motor

I have to add "long sleeve shirt" to my eye protection and face mask recommendation.

Also:

13) The use of sand bags in engine test facilities is encouraged. Note that this means bags filled with sand, not straw. While straw does make the bags easier to handle, the Borsk Test Facility has found them less effective in stopping flying objects.

http://www.srogers.com/comedy/aircraft_design.asp

[max_torque]

Anyone who thinks that the "clever" or "difficult" bit of 3phase BLDC motor control is generating the 120deg phase shifted fundamental PWM signals is in for quite a steep learning curve........   ;-)

[ziggyfish]

Anyone who thinks that the "clever" or "difficult" bit of 3phase BLDC motor control is generating the 120deg phase shifted fundamental PWM signals is in for quite a steep learning curve........   ;-)

 

[aqibi2000]

Genuinely loved this conversation. 

[Mechatrommer]

sensorless 60A ESC for 3 wires brushless motor... fwiw..
http://www.ebay.com.my/itm/Mystery-Fire-Dragon-60A-Brushless-ESC-RC-Speed-Controller-/261815005989?pt=LH_DefaultDomain_77&hash=item3cf563f325