The Rev Robotics "Spark" Motor Controller

The Rev Robotics Spark Motor Controller is a brushed motor controller for the FIRST Robotics Competition. From the REV web page:

• Input Voltage (nominal): 12V
• Continuous Current: 60A
• Peak Current (2 second surge): 100A
• Input Pulse Width Range (nominal): 1ms – 2ms
• Input Resolution: 1 microsecond
• Input Deadband: 4 percent
• Output Frequency: 15.625kHz
• Output Voltage Range: 0V - 30V
• Output Voltage Resolution: 0.001xVin
• Dimensions: 2.860 x 1.875 x 0.868

Merry Christmas, and happy new year! Here's the Spark as it arrived at our door:

Spark Box

The Spark comes with a PWM cable, several spare terminal screws and backing plates, and of course the controller itself.

Spark Open Box

The lid of the Spark box has a nice layout that explains the different parts of the controller:

Spark Lid

The controller itself has the markings you would expect. It has V+ marking for input voltage, M+ for motor outputs, and a Mode button to change brake and coast modes and to allow the user to calibrate the controller themselves. The PWM receptacle has a small "B" that denotes the ground or black wire side of the plug. We did notice that the PWM housing is not held in place particularly well on the spark. We always recommend using hot glue between the housing and speed controllers, and the Spark is no exception.

There is a receptacle for limit switches to stop motion, and a status light. The status light LED is extremely bright when it is in its blue state.

Spark

For those EE's in the crowd we decided to dissect the spark. It comes apart quite easily. Removing 4 screws on the bottom of the case gives you access to the circuit board.

Spark Bottom Removed

The guts of the spark simply slide out of the plastic case.

Spark Case

A top view of the Spark shows the majority of the real estate is taken up by the heat sink.

Spark Guts

A side view shows just how thick the heat sink is, and shows the adhesive thermal tape that attaches the board and components to the heat sink.

Spark Guts Side View

The heat sink can be easily lifted off the tape.

Spark Guts With No Heatsink

After reassembling the Spark, we ran it through the same test we did several years ago. Our baseline speed controller in the past was the Victor 884, so we included it in the tests this time around. In addition - we were doing the testing at home and didn't have a Talon or a newer Victor at hand. Pesky Christmas break.

Most are familiar with just how bad the Victor 884 linearity was. These tests were run with an 8 port cRio controlling the speed controller. The controller is powering a single CIM through a toughbox mini that is driving 3 wheels via chain.

Victor 884, Speed Graph

The Spark reminds us of the Jaguar and Talon speed controller linearity. It is highly linear except for a tiny roll-off at the very ends of the graph.

Spark, Speed Graph

Next, we ran through the test with no load on the output of the speed controller. The 884 is again very non-linear.

884 Voltage Output No Load

The spark is quite linear with no load.

Spark Voltage Output No Load

The only odditty that we noticed with the spark during testing was that the deadband was offset slightly. In addition, when testing as we slowly incremented the LabVIEW drive command upwards by hundredths, we saw a point where the speed controller was flashing it's light back and forth between neutral and forward (or reverse). In addition, the controller appears to begin flashing forward or backward before it actually applies a voltage.

More specifically, the controller begain blinking when we applied -0.04 in LabVIEW, it flickered back and forth between blinking blue and neutral. However, it did not apply a voltage to the output terminals until we reached -0.09.

In the other direction, the controller began to blink red at 0.09, but did not apply a voltage until we reached 0.10.

Spark Deadband

We performed some very basic durability testing on the spark as well. These included rapid forward / backward iterations that used to give the old Jaguar fits, and then we stalled the drivetrain and increased power slowly until we were near popping the snap action breaker. The motor controller did not experience any problems or overheat, and we drained 3 batteries that way.

All in all, it appears that the Spark motor controller will meet the requirements of FRC Teams at a price that is very appealing. The small issues we saw in testing may very well have been related to the lack of a "Spark" LabVIEW class, and they only appear at extremely low power levels that FRC robots rarely operate at for extended periods.