Issue 98, September 1998
Smart Rockets - Data Acquisition in Model Rocketry

RESULTS

Before leaving the lab, we went through every step in the sequence of the setup, launch (with no engine), data recovery, and data visualization. We even shook the rocket up and down a few times after pressing the launch key, just to generate acceleration data. Once the procedure was perfected, we moved outside.

The raw data from a launch with an Estes C6-5 engine is shown in Figure 5. The acceleration appears to be negative, but that’s because the on-chip buffer of the ADXL50 is an inverting amplifier.

The initial kick and the sustained thrust are easily seen, as is the ejection charge (around 6000 ms). There is also a spurious data point (around 4500 ms). While the accelerometer output has not been converted into acceleration yet, this plot still tells us that the system worked.

Turning Figure 5 into the desired acceleration plot requires a few steps:

• removing the spurious data points
• smoothing the curve with a three-point sliding-window average
• converting the 0–255 A/D readings to 0–5 V
• converting the voltage to acceleration, exploiting the fact that the system saw 1 g just before ignition

The first three steps are straightforward. For the final step, we used the published response curve of the sensor as well as the calculated transfer function for the amplifier, given the values of resistors R4–R6.

We know that, just before ignition, the rocket was experiencing the 1-g gravitational force, so we defined the observed DC offset to be 1 g. Figure 6, shows the final acceleration curve for our system.

Compare Figure 6 to the solid line in Figure 1. The general shape of the curves agree, although the measured curve is compressed in time. The maximum measured acceleration was 17 g, while the maximum acceleration predicted from the mass of the rocket and the characteristics of the C6-5 engine was 14 g.

The correspondence between theory and experiment isn’t bad, and it shows our system works as designed. There are several ways to improve the quality of the data.

First is to perform a careful calibration of the accelerometer. Calibration procedures for the ADXL50 are discussed in the datasheet. We calculated gains using the nominal values of our 5% resistors. Using 1% resistors in the accelerometer circuit would give better results.

Also, a more stable voltage source can be used as the reference for the PIC’s ADC. Other questions can be addressed as well: What causes the spurious data? What is the source of the ripple seen even when the rocket is on the pad?