Week 3, Day Three, 10/11/12

I realized earlier today that while I do most of the posting, I haven't acknowledged the most important part of this testing process. Thanks to Ron, Michael, Amanda and Alysa for putting in their time after school figuring out test plans, writing code, and taking down data. This wouldn't happen without them.

Talon / Jaguar / Victor Linearity Test
The first test we performed today was a comparison in linearity between the 3 motor controllers. The motor control outputs were not attached to anything, and we varied the PWM command in labview from -1 to 1 in increments of .1. Around the zero point, we changed to increments of .01. This was an effort to more closely observe the deadband response of the controllers.

Our First Talon was showing a forward command even when we were sending it a zero command. We concluded that it was not calibrated. The calibration went quickly and corrected that problem. Our second Talon had the correct center point, but when we went full reverse we didn't get full voltage. We decided to go through and calibrate all our speed controllers. In addition, we don't have a specific Talon class in LabView. We used the Victor class and it seemed to perform extremely well.

Several things stand out immediately. The Victors are quite non-linear. You can see that both the Talon and Jaguar have little to no difference and are almost completely linear. You can also see that the Victor has a comparitively large deadband around the zero point.

PWM vs. Voltage

In the second picture, we zoomed in on the graph to look at the zero point more closely. We removed the Victor from this graph to more clearly show the deadbands of the Talon and Jaguar.

The Talon spec sheet says that it's deadband is 4%. We measured exactly +/- 3.5% on our test. The Jaguar's deadband is +/- 5.5%. Additional resolution may have been provided by stepping through this zone in even finer increments, however we had already reached the point where we were seeing data variations of .1 volts between the two Talons, suggesting that experimental error was too great to warrant it.

PWM vs. Voltage, Zoomed on Zero Point

The raw data can be found here.

Window Motor Lockup
As the first part of the window motor lockup test, we created code that allowed us to vary the frequency and power of motor reverses. We started with the Jaguar in an effort to recreate the lock-ups that other teams have seen. Our first test drives a intake roller with very low friction. Each interval is a different direction, alternating back and forth. The hertz is the measure of time before we shift directions. While we didn't expect these motors to lock under low load and long interval, we wanted to be thorough with this test.

The 2010 Ball Magnet

Initial Battery voltage: 12.77 volts.

Jaguar results: 0.1 hz at .5 PWM command: No lockup
0.1 hz intervals at full speed: No lockup.
0.2 hz intervals at .5 PWM command: No lockup.
0.2 hz intervals at full power: No lockup.
0.5 hz intervals at .5 PWM command: No lockup.
0.5 hz intervals at full power: No lockup.
1 hz intervals at .5 PWM command: No lockup.
1 hz intervals at full power: No lockup.
5 hz intervals at .5 PWM command: No lockup.
5 hz intervals at full power: No lockup.
10 hz intervals at .5 PWM command: No lockup.
10 hz intervals at full power: No lockup.
100 hz intervals at .5 PWM command: No lockup.
100 hz intervals at full power: No lockup.
Maximum frequency of the mechanical system at full power (commanded 1000 Hz): No lockup.

Talon results: The same.

Victor results: The same.

We didn't expect to see a lockup under an almost no-load condition. Next week, we'll perform the same test with the ball roller clamped in place. A vacuum belt provides the connection between the motor and the roller, and will slip in a roller stall-condition. This will provide a fairly constant and fairly high torque against the window motor.

We may have a special edition Saturday. According to FIRST, we may have the LabView Beta materials delivered tomorrow. If that's the case, you'll see us post with initial results from the new software and a description of the required tests late Saturday night.