AN-1042: ADIS16130 Quick Start Guide and Bias Optimization Tips

Physical Mounting and Handling

The ADIS16130 comes in a plastic housing that provides four mounting holes accommodating either M2 or 2-56 machine screws. This package enables both connector-up and connector-down mounting approaches. Figure 2 provides a mechanical diagram for a system using the connector-down approach. In comparison, the connector-up approach uses the same hole pattern for the four mounting screws but requires an additional cable/connector interface for electrical connection. The connector provides a standard, dual-row, 1 mm pitch that offers many options for mating connectors. Figure 3 provides the suggested pad layout and hole pattern for the mating connector, assuming use of a connector from the Samtec CLM-112-02 family. The holes inside of the mounting pads prevent the pins on the ADIS16130 connector from bottoming out, which can cause mechanical stress and bias shifts.

Electrical Hookup

Figure 4 provides a basic schematic for connecting the ADIS16130 to a system processor, which serves as the SPI master. Once configured, the ADIS16130 produces data autonomously and continuously updates the output data register(s). This hookup diagram accommodates the power, the ground, four serial signals, a data-ready signal, and the bandwidth reduction capacitor. The data-ready signal typically drives an interrupt service routine in the master processor, which helps optimize data coherency and processor resources.

SPI Interface

Table 2 provides a list of typical configuration settings that master processors require for SPI communication with the ADIS16130. These settings are normally in control registers. For example, the SPI_BAUD, SPI_CTL, and SPI_FLG registers serve this purpose in the ADSP-BF533 processor family. Refer to the ADIS16130 data sheet for timing information.

Placing a command on SDI involves writing to the transmit buffer register (SPI_TDBR in the ADSP-BF533), and acquiring output data from SDO involves reading the receive buffer register (SPI_RDBR in the ADSP-BF533). For many processors, these commands automatically facilitate the clocks and sequencing according to the settings in the control registers.

Initialization

Initialize the ADIS16130 by writing each command that is listed in the SDI column of Table 3 to the SDI pin.

Reading Output Data

The data ready (RDY) signal indicates that new, unread data resides in the output registers. To optimize data coherency and processor resources, use this signal to drive an interrupt service routine in the master processor. It will pulse low for approximately 26 μs at a rate of 11.4 kHz, which indicates a new update in either the RATE or TEMP registers. Use the high-to-low transition to trigger action on the interrupt service routine. Read data after this signal pulses low by lowering the chip select line and then shifting (write) 0x48 onto the SDI line. Then, shift (read) the next three bytes in from SDO, which represents the resulting output data. Raise chip select between each 8-bit sequence. The most significant byte is first in the SDO sequence, followed by the next significant, and then the least significant. When 16-bit resolution is in use, only two bytes are output from the SDO during the read sequence.

Data Format

The ADIS16130 uses the offset binary data format.

Optimizing Bias Accuracy and Stability

Two common sources of error in gyroscope systems are initial bias error and bias change with temperature. Figure 6 provides a simple process diagram for correcting these errors in a digital processor platform. Each ADIS16130 has unique behaviors that require characterization to optimize the correction factors in this process diagram.

This calibration process results in the following mathematical relationship for the corrected gyroscope rate output:

RBC = RATE + PBC + (TEMP − T3) × BTC

Where:
RBC is the corrected output data.
RATE is the ADIS16130 gyroscope output data.
TEMP is the ADIS16130 temperature output data.

Bias characterization involves measuring the ADIS16130 output while it is in a zero-rotation state. Noise adds random errors to the output measurement and can influence the accuracy of bias measurements. A common approach to addressing noise in the bias measurement is through averaging sequential samples. The Allan variance curves in Figure 7 display a relationship between averaging time and bias accuracy for the ADIS16130.

While each application has its own performance criteria, the following procedure provides a basic bias calibration process for getting started:

• Attach the ADIS16130 to a fixed platform to prevent any motion during the measurement steps.
• Apply a stable +5 V linear supply to the ADIS16130. Note that the output bias can shift by 0.0002°/sec for every 1 mV of change in the power supply.
• Write the initialization commands from Table 3 to the SDI pin.
• Read TEMP and RATE data at the maximum sample rate of 5.7 kSPS for 10 sec. Average each set of data to produce a temperature estimate (T1) and a bias estimate (B1) associated with this average temperature.
• Wait 10 minutes for the device to self-heat.
• Repeat Step 3 to produce new estimates for temperature (T2) and output bias (B2). Then, calculate the bias temperature coefficient (BTC) by using the following equation:
  

  

  Using BTC, develop a look-up table for loading the correction factor suitable for the temperature conditions. Table 6 shows an example of correction factors in a table that corrects for a 0.04°/sec/°C bias tempco, using temperature step sizes of 0.02°C.
• Wait 40 minutes for thermal stability.
• Read the RATE data at a sample rate of 5.7 kSPS for 150 sec. Average this data together for a precise bias estimate and multiply it by −1 to produce the precise bias correction factor (PBC).
• Read TEMP (T3).

Additional Tips

In conclusion, the simple calibration process offered in this application note enables customers to achieve substantial performance gains and helps lay the groundwork for developing more extensive calibration approaches, if necessary. Here are some additional tips that may add value:

• The 5.7 kSPS sample rate enables filtering of resonant behaviors that can show up around 14.3 kHz.
• Use a 0.01 μF capacitor across ROA1 and ROA2 to reduce the sensor bandwidth to 50 Hz, which reduces the noise by 50%. The 50 Hz bandwidth also enables the use of slower sample rates. For example, use the RDY signal to drive a 200× counter to reduce the read rate to 230 SPS.
• The calibration process in this application note takes advantage of the thermal warm-up period in these devices. For better accuracy, use a controlled temperature chamber to hold the device at fixed temperatures while using longer averaging times for each bias point. This provides the opportunity for further performance optimization.
• If the processor system cannot read both gyroscope and temperature sensor outputs at the rate of 5.7 kSPS, reduce the read rate on the temperature sensor before considering a decrease in the rate measurements. If the maximum rate of temperature change is 1°C/sec, then a sample rate of 100 SPS should be sufficient for detecting 0.02°C changes for updating thermal correction factors.

作者

Mark Looney

Mark Looney是ADI公司（美国北卡罗来纳州格林斯博罗）的MEMS和传感器产品线应用工程师。自1998年加入ADI公司以来，他在传感器信号处理、高速模数转换器和DC-DC电源转换领域积累了丰富的工作经验。他拥有内华达州大学雷诺分校电子工程专业学士和硕士学位，曾发表过数篇有关在工业应用中运用MEMS技术的文章。