Servo - 201306

Discuss this article in the SERVO Magazine forums at http://forum.servomagazine.com

Figure 2

Robotis (makers of the Bioloid) used in their CM- 5 robot controller to talk to the servos. I am not much of an artist, so my cable socket drawing in the upper right of the figure isn’t all that pretty.

The Dynamixel bus is what is called half-duplex. This means that only one device can talk at a time. In this case, the Master (CM- 5) talks to a servo, and a servo Slave then responds. According to the documentation, the data is transmitted using a UART asynchronous serial format, which is good for us.

The default data rate is 1 Mbps (one million bits per second). That is really fast, which means we need to be careful in our hardware design — which is our first problem.

There is another problem we need to deal with: The Dynamixel bus is 5V. The UNO32 and Cerebot (MX3cK) are 3.3V. (Rats, another voltage level converter problem!)

SparkFun: www.sparkfun.com/ products/8745

My own design uses through hole components, since that is what I have in my junkbox. Like these two designs, 10K resistor pull-ups are on the 5V and 3.3V sides. That takes care of part 2, but when I tested the design on my scope (when it didn’t work), I saw that there were serious capacitive slews on the data line.

A slew occurs when a signal is changing so fast that the load getting the signal acts like an RC charging circuit, then that capacitor in the RC circuit can’t charge fast enough for the signal to be properly seen. Well, we can’t change the ‘C’ part of the circuit, but we sure can change the ‘R’ part!

My converter circuit uses 1K resistor pull-ups which reduces the resistance of the circuit 10X. Further testing with this change showed an acceptable slew rate and voila! My circuit started to work.

Figure 2 is the complete circuit design. I did not have any 74HC126s lying around, so I used a 74HC125 which is the same, but the enable on the buffers is active low instead of active high.