How do I measure noise density in a gyroscope?
Noise Density provides a useful metric for understanding the trade-off between the total noise and bandwidth of a gyroscope. There are a number of techniques for measuring this parameter, but this example uses the EVAL-ADIS for data collection and a common signal processing technique: the Fast Fourier Transform. The following steps provide an example for producing a Rate Noise Density plot for the gyroscopes in the ADIS16334, a fully-calibrated, six-degree of freedom IMU.
STEP 1: ADIS16334 Installation on the EVAL-ADIS
Connect the ADIS16334/PCBZ to the EVAL-ADIS. Note that this attachment process only requires the ADIS16334BMLZ and flexible connector (green PCB not required in this case). Note that the ADIS16334BMLZ and flexible connector only have 20 pins, while the mating connector, J4, on the EVAL-ADIS, has 24 pins. Make sure that the flexible connector mates to pins 1-20 on the EVAL-ADIS and properly seats into the mating connector on the ADIS16334BMLZ.
DO NOT USE THE SILK SCREEN ON THE FLEXIBLE CONNECTOR TO DETERMINE PIN 1. This is a multi-purpose connector and the pin 1 indicators are not correct, with respect to the ADIS16334BMLZ or J4 on the EVAL-ADIS. Also, note that the ADIS16334BMLZ uses M2x0.4x10mm (or longer) machine screws to mate to the appropriate, pre-tapped screw holes on the EVAL-ADIS. The following picture provides a visual example of how this will look, with the following exceptions: the ADIS16334BMLZ does not have the 10-32 hole on the side wall and is 5mm shorter than the package in this picture (this product will be announced later!). Make sure that the JP1 jumper (EVAL-ADIS) is set to +5V and proceed to the next step before connecting the EVAL-ADIS to the USB cable.
STEP 2: Install IMU Evaluation Software
Download IMU Evaluation software from www.analog.com/EVAL-ADIS, under "SOFTWARE AND TOOLS." Use UG-287 (EVAL-ADIS User Guide) to guide software and driver installation. See www.analog.com/EVAL-ADIS, under "DOCUMENTATION," for the link to UG-287. Make sure that the software package is communicating with the ADIS16334BMLZ by using the "Read" button in the Main Window (IMU Evaluation software). See the following figure for an example of the waveform recorder output.
STEP 3: ADIS16334 Configuration in IMU Evaluation
In the IMU Evaluation Main Window, click on "Register Access" and make sure that SMPL_PRD = 0x0001 and SENS_AVG = 0x0400. If they do not match this, use this window to update them and then exit to the main window.
STEP 4: Data Collection
Place the ADIS16334BMLZ (and EVAL-ADIS) in a location where it will not experience vibration. For this example, I placed this assembly in a small vice for mechanical stability, and then placed the entire system on my floor, which has a thin layer of carpet, on a concrete slab. In the IMU Evaluation Main Window, click on "Data Capture" and configure the software to collect 262,144 (2^24) samples, as shown in the following figure. Then, select the sensor(s) for data collection. The example below will collect data for the z-axis gyroscope only but this function will support full-rate, synchronous data capture on all of the sensor options.
NOTE: The accelerometer options may be useful for characterizing vibration on the test platform.
STEP 5: Analysis
The four file attachments provide an example of how to use the "pwelch()" command, inside of Matlab's Signal Processing Toolbox, for this type of analysis. In order to use these files, use the following steps:
- Copy them into Matlab's user directory on a PC.
- Open "GyroNoiseDensity.m" and edit the path associated with the "csvread()" command, on the first line, to point to the correct location for the ADIS16334-GND-Test-03.csv file location. You can also use the "File/Open" drop-down menu in Matlab to import the data file. Delete the first line in the "GyroNoiseDensity.m" file when using this option.
- Type "GyroNoiseDensity" at the command prompt and hit "Enter."
- The script will automatically use the other two *.m files to process and smooth the noise data.
The following figure illustrates the correct response. Note that the noise density is ~0.04 deg/sec per sqrt(Hz), which is ~10% lower than the "typical" number in the datasheet. The latest characterization on the ADIS16334 suggests that the majority of the units will be lower than the "typical" number in the ADIS16334 datasheet specifications (Revision A, Table 1, located on page 3). We look forward to your feedback!