Using and Driving a Stepper Motor

By Sivan Toledo
June 2007

Stepper motors are interesting actuators. Unlike common DC motors (the class of motors that include all the Lego motors), they usually have several windings (coils) and more than two contacts. But activating the windings in a specific order, a controller can cause the motor to move in concrete steps. The behavior of steppers in a system is quite different than that of DC motors; more on that below. 

In this project I interfaced a stepper from an old scanner to the NXT. By interfacing, I mean here both electrically and mechanically. In fact, the mechanical part is the more challenging here. You can watch the movie on the right to get a sense of how the stepperr behaves. The program running on the NXT tells the stepper to rotate 24 steps, which are 180 degrees, clockwise, wait 2 seconds, rotate back, and so on.

The stepper is controlled by an interesting pair of chips, an I2C digital I/O chip, the MCP23016, and an array of 8 Darlington pairs, the ULN2803. The MCP23016 (and its smaller sibling, the MCP23008) are better in this application than a PCF8574, because it can source up to 25mA per I/O pin, whereas the PCF8574 can only source 100uA, which means that it cannot drive the Darlingtons directly without strong pull-up resistors.

Warning: connecting your NXT to any home-made gizmo (like the one described here) can damage it. Beware.

The Motor

Three stepper motorsThe picture on the right shows three stepper motors that I tried to use. The ones in the middle and on the left are from 3.5" floppy-disk drives. For some reason I could not get them to work properly. The one on the right is the one I ended up using. It comes from an old flat-bed scanner.

There are different several types of steppers. You can usually distinguish between them using an Ohmmeter. The excellent stepper-motor tutorial by Douglas Jones describes the different types and how to drive them, so I won't repeat this here.

My stepper turned out to be a 4-phase unipolar stepper with 4 windings. One endpoint of each winding is connected to a common wire, so there are 5 wires altogether coming out of the motor. Each winding is actually used 12 times around the motor, which gives 48 steps of 7.5 degrees each. When current flows between the common connection and one of the 4 endpoints of the windings, this creates a magnetic field that causes the rotor to align with that winding. If current stops flowing in that winding but starts flowing in an adjacent one, the rotor will step to align with the energized winding. By energizing the windings in one cyclic order, say 1-2-3-4-1-2-3-4, we cause the motor to rotate in one direction, and by energizing them in the opposite cyclic order 1-4-3-2-1-4-3-2 we cause the motor to rotate in the other direction. The rotation is discrete, in steps of 7.5 degrees. When one winding is energized continuously, the motor is stopped and will hold its position up to some level of torque.

The motor with the Technic gear, the board, and a prototyping PCB for the controller Interfacing the motor to the Technic system is challenging. The hard part is to interface the shaft, on which a brass gear was mounted, to Technic axles. I followed advice by Philo to use heat-shring tubing and glue to secure some Technic part to the brass tubing. I used an 8t Technic gear and 8mm heat-shrink tubing. When shrunk, the tubing gripped the Technic gear very firmly, but it didn't grip the brass gear very firmly. I added another layer of heat-shrink tubing around the brass gear, which improved the situation, and secured the whole thing with a bit of hot glue. This structure now transfers torque to the Technic gear very well, but it still came off when I made the movie above. I can still put the tubing back on the brass gear and it will trasfer torque, but it's not a very solid construction. Perhaps epoxy, which Philo also suggested, would have worked better.

The mounting for the motor was easier. I cut two rectangles of 4mm plywood to size (the sides are multiple of 8mm) and glued them together. I then drilled a large hole for the motor shaft and the gears, two small holes for the mounting screws, and 5 Technic-compatible holes on each side. I only used hand tools. There was quite a bit of chipping when I drilled the Technic holes near the edges of the board, since they are so close. In hindsite, it would have been much better to first drill the holes and then to cut the plywood to size. In any case, the resulting assembly is very strong and easy to connect to Technic constructions. Even a single 4mm board would probably have been sufficient.

The motor assembly, top view The motor assembly, bottom view

The Controller

The controllerI initially intended to use discrete MOSFET's to drive the stepper, driven by either a PCF8574 or by a microcontroller. But Michael Gasperi suggested on a forum a better solution, the ULN2003. This chip contains 7 Darlington pairs that can each sink 500mA, and each output comes with a built-in protection diode. Because the Darlington pairs use bipolar transistors, you need to push more than 1mA into an input of the ULN2003 to saturate the corresponding output. The PCF8574 is not good at pushing current. It can sink 25mA (that is, 25mA can flow into an ouput that is held low), but it can only source 100uA, not enough to drive a bipolar Darlington directly. Michael circuit used pull-up resistors to provide the extra current. This solution works, but it consumes several mA of current per Darlington even when it is off.

The MCP23016 is an  I2C 16-port digital I/O chip; the MCP23008 is similar, but has only 8 I/O pins. ou can configure each pin in these chips as input as output. Unlike the PCF8574, it can both source and sink 25mA per output pin. They have several other functions, like generating interrupts on input change and capturing the state of the inputs when a change occurs, but these are not useful with the NXT because it does not have interrupt inputs. The two sibling chips are not identical, and the MCP23008 has more features, like programmable pull-up resistors on input-pins. Another difference is that the MCP23016 uses an internal oscillator that requires an external resistor and an external capcitor, increasing the part count somewhat. Both chips are about as power-efficient as the PCF8574, which uses 10uA at standby: the MCP23016 uses 25uA, and the MCP23008 only 1uA. Both come in large DIP packages, which I used, and in several smaller packages.

In my controller, I used a ULN2803, which is similar to the ULN2003 but has 8 Darlington pairs rather than 7, and an MCP23016. Therefore, the controller can drive two steppers of the same type, or it can provide twice as much power to a larger stepper (by connecting Darlingtons in parallel), or a stepper and solenoids, etc. You can drive all the Darlington pairs of the ULN2803 using the smaller MCP23008, but I wanted to leave some I/O pins available for sensor input, say a switch or a Hall-effecct sensor, to sense the position of the stepper. Power is provided to the stepper from a NXT socket that is connected either to a 9V batter pack (using a converter cable) or to a NXT motor port. The stepper is connected to the controller using a 6-pin polarized header (but only 5 pins are used). The I2C pullup resistors are connected to the 4.3V supply line through jumpers, to allow them to be disconnected if the controller is used in parallel with additional I2C circuits on the same NXT sensor port.

I used a Schottky diode between the external power supply and the ULN2803, to protect the chip from negative voltage. But since it is a bipolar device, I don't think that negative voltage will harm the chip, and with its collector at a lower voltage than the emitter, I think that the Darlington will simply shut down.

Steppers vs DC Motors

As I  wrote above, steppers behave quite differently from DC motors, even when the DC motor is coupled with an encoder, like the NXT motors.

Shutting Down Add-On Actuators

The NXT has a useful feature that you hardly notice: when a program shuts down, the NXT shuts down all the motor ports. So if your robot misbehaves, you can always stop the program using the gray button and all activities stop.

This is not the case with NXT-controlled actuator that have a separate power source, like a battery pack. They can continue to operate and drain battery energy even when the program stops. This is a  particular problem with steppers, since they can consume considerable amounts of power when just holding a fixed position. So you may not see anything moving, but the motor may be drawing current. So you have to be careful and shut them down separately.

The stepper, controller, NXT, and batteries

The stepperr in a Technic construction

© 2007, Sivan Toledo