I’m sort of new to robotics and it’s my first time not using a pre-made motor driver. So far I have 4 3-phase brushless drive motors rated for 36V and 7A with 20A peak, I also have 3 24V steppers with 2A peak current for a basic robotic arm stub. I have the thermal, acceleration, top speed and torque calculations done and even a 4 speed transmission. All the mechanical and concept problems are worked out. The only real problem now is back voltage.
The drive motors are brushless with no circuitry aside from the integrated hall sensors. The drive motors have 0.38 ohm per coil resistance, per coil 1 mH (0.001H) inductance and a maximum current of 13A. I’ll be using 3 MJL3281A transistors per motor (1-per phase) to drive them. The transistors have a output capacitance of <600 pF not including the giant 45V 25A schtottky diodes. The MJL3821A datasheet mentions nothing about transition times, only that capacitance. I don’t know how capacitance would play into the transition time, so assuming the transition from 13A to 0A to the motors takes place in 10 microseconds that will be 1,300V = 0.001 *(13/0.00001). Which would destroy pretty much the entire robot from the insulation of the wires to the last semiconductor.
A similar thing happens with my stepper motors, but on a smaller scale of course.
I have all the diodes and transistors even heatsinks already, it’s just a matter of what value capacitor I should use. So really, all I need to know is how capacitance would play into the transition time.
1 “line” per transistor, a line is 2 or 3 coils in series. The company didn’t mention the exact number in the data sheet, but I assume it’s at least 6 coils.
Circulating currents? What do you mean?
The actual motor controller will be a TI C2000 (with a bunch of MSP430s monitoring temperatures and minor tasks) or possibly 2 but aside from that, it’s from scratch. I’m designing it with the help of TI motor control guides. I figured it wouldn’t be incredibly difficult to make a driver circuit for BLDC motors. In one of Atmel’s guides it says commutation for a 3-phase motor can be done with only 6 switches, 3 darlington transistors and 3 hall effect sensors. It’s differential drive, so technically I could commutate the 2 motors on each side as if they were 1.
Sorry, but the schematic is on my other computer, I would make another copy but I don’t know of a circuit diagramming software for Ubuntu.
An equivalent circuit of on of the differential drive motors below. The inductors are 1mH and resistors are 0.38 ohms each. The transistors to the left are actually hall effect sensors. If you can’t tell, the motors are wound in a delta configuration making them slightly less efficient and more complicated to control than an equivalent Wye winding. But really all I need to know is what value and what voltage capacitor to use so I don’t destroy everything.
Here’s a pretty rough sketch (the driver transistors aren’t even connected to the driver yet!), the C2000 is much bigger and more complicated than the MSP430 so I’ll add it later. The 430 is just a stand in part to convey the idea. I’m still working on the component sketch for it. I still need to know how capacitance factors into the transition time from 13A to 0A.
Microchip has a few App Notes on using BLDC motors and the drive requirements. These are a good place to start reading. So look up, download and read the following, then many of your questions will be answered:
Well, thanks for note recommendations, Delta wound motors are just a little bit trickier than Wye wound motors. Which I’ve got plenty of experience with.
I’m using the guide provided by TI for the C2000 and a supplementary guide from Atmel. The C2000 was made to commutate anything inductive really, It has software libraries that are almost completely dedicated to 3-Phase BLDCs because they’re so common. From a control standpoint I have this under control. The thing is, they mentioned nothing about back EMF suppression aside from Schottky diodes. Not only that but the additional voltage from inductance caused powering down the coils.
Yeah that design wasn’t made to functional at all, but just to demonstrate a basic concept. It was really a copy-paste design I made for a Wye-wound motor a while ago. Once I get the C2000 defined in SPICE and the NPN+PNP drivers modeled accurately I’ll be able to show you the real design. I’m pretty new to the software so it’ll take a couple days.
Also, the MJLA3821 and it’s PNP version the MJLA1302A can handle up to 260V and up to 15A, so when powered by a 10 cell Li-Ion battery that’ll be 37V with 12A at peak output.
Alright, just so you guys know, I know what I’m doing with the drive motors. I just couldn’t draw it properly on that other computer. I’ve commutated Wye motors before. All the wire ends connect to either power or the TI C2000 as necessary. I just need to know the “magic capacitor” value and voltage. The main motors commutation is not at all the problem, it’s back EMF generated by commutation. The only motors I don’t know how to commutate are steppers. That’s what that strange little cluster of parts is at the top right.
Yep, I got one of the motors running! It’s not the most responsive but it does the job. If you run into a similar problem just look up “RC snubber” or “RC snubber circuits”
