Engineering Power and MEMS Signal Chain Simulation Using LTspice

Adding an LDO regulator or a DC-DC converter to the sensor node enables powering from the master node on standard industrial voltage rails such as 12 VDC and 24 VDC. The choice of an LDO regulator or a DC-DC switching regulator depends on the application requirements. If the application uses a 12 VDC rail, an LDO regulator may be suitable to achieve ultra-low noise performance and produce acceptable power consumption in the sensor sub-node. For a 24 VDC rail, a more efficient DC-DC switching regulator is recommended to reduce power consumption. ADI's low noise Silent Switcher® architecture ensures higher energy efficiency and low noise.

24 VDC is widely used in railway, industrial automation, aerospace, and defense applications. The EN 50155 standard for electronic devices for railway use5 specifies a nominal input voltage of 24 VDC, but the nominal input varies from 0.7 VIN to 1.25 VIN, with an extended range of 0.6 × VIN to 1.4 × VIN specified. Therefore, the DC-DC devices used in the application require a wider input range of 14.4 VDC to 33.6 VDC.

With a 6.25 mm × 6.25 mm BGA package and a wide 3.4 VDC to 40 VDC input range, the LTM8002 Silent Switcher µModule® regulator is ideal for space-constrained vibration sensors used in railway vehicle monitoring.

Figure 7 replicates the schematic of Figure 4 with the addition of the LTM8002, and the power delivered from the master node to the slave node sensor is 24 VDC. Simulations show that a 1ms ramp time is required to reach the required 5 VDC ±1% output voltage on the LTM8002. It is recommended that designers implement a 2ms to 3ms time delay at power-up before initiating communication between the master node and the slave node. This will ensure valid data is obtained at the sensor node output.

Figure 7. Using ADI’s low-noise Silent Switcher devices in the sensor subnode (LTM8002) provides greater flexibility in power rail design.

Figure 8. Ramp time to reach desired 5 VDC on VPOUT is 1 ms, valid data on VOUT 2 ms to 3 ms later

Complete MEMS signal chain simulation

Analog Devices provides many design notes to help designers design MEMS signal chains and simulate them using LTspice (see Figure 9). While many MEMS have digital outputs, there are many high-performance sensors with analog outputs. Simulating the op amp and ADC signal chain can provide valuable insights before the hardware design is built.

To analyze the effects of low-pass filtering, amplifiers, and ADC inputs on sensor data, designers can refer to LTspice benchmark circuits from Gabino Alonso and Kris Lokere. 6 Simulation models are available for the AD4002 and AD4003 18-bit SAR ADCs and the 16-bit LTC2311-16. Erick Cook provides a useful hands-on guide to developing custom analog-to-digital converter models using LTspice. 7

More than 200 op amp models are available, including the ADA4807 and ADA4805 families. Voltage reference macro models are available, such as the ADR4525 and LTC6655-5, as well as the ADA4807-1 voltage reference buffer.

In one of his articles on condition monitoring systems, Simon Bramble describes how to use LTspice to analyze the frequency spectrum of vibration data. 8 Simon’s article provides helpful tips on formatting and analyzing captured sensor data.

Figure 10 shows an example LTspice model of the frequency response of the ADXL1002 low noise, ±50 g MEMS accelerometer. Using a series LRC circuit in LTspice Laplace format closely matches the MEMS frequency response. The simulated model shows good agreement with the datasheet typical performance, with a resonant frequency of 21 kHz and 3 dB at 11 kHz. For ac analysis, it is best to use the Laplace circuit in LTspice, but for transient analysis, discrete RLC devices should be used for best simulated performance.

Figure 9. Complete sensor signal chain simulation using LTspice (simplified diagram—all connections and passive components not shown)

Figure 10. (a) Laplace model of the MEMS frequency response, (b) showing the resonant frequency at 21 kHz and 3 dB at 11 kHz.

For analog output accelerometers (e.g., the ADXL1002), bandwidth is defined as the signal frequency at which the response to dc (or low frequency) acceleration drops to –3 dB. Figure 11 replicates the MEMS frequency response model of Figure 10 but also includes the op amp’s filter circuit. Using this filter circuit, more of the MEMS frequency response can be measured within 3 dB. The graph shows that at 17 kHz the op amp’s VOUT is 3 dB, while the output of the unfiltered MEMS is 3 dB at 11 kHz.

Figure 12 includes the MEMS input model (discrete RLC in Figure 10), op amp filtering, and the 16-bit LTC2311-16 SAR ADC model. The complete signal chain can be built and simulated using a modular approach, adding the wired interface and engineering power as separate modules.

For transient simulations, the LTC2311-16 DIGITAL_OUT node can be probed to see the digital output corresponding to the MEMS voltage input (VIN). The LTC2311-16 LTspice model can be modified to reduce the serial clock and CNV interface timing, and the digital output reference OVDD can be changed to any value in the range of 1.71 V to 2.5 V. Some RS-485 transceivers (for example, the LTC2865) include a logic interface pin, VL, that can run at either 1.8 V or 2.5 V, providing a perfect match for wired streaming of the ADC digital output data. The LTC2865 VCC pin can then be used to power the RS-485 interface separately at either 3.3 V or 5.0 V to provide higher voltage cable drive.

Figure 11. (a) MEMS frequency response and filter model, and (b) 3 dB point pushed up to 17 kHz (compared to Figure 10b at 11 kHz)

Figure 12. MEMS input model (discrete RLC in Figure 10), op amp filtering, and 16-bit LTC2311-16 SAR ADC model

Figure 13. MEMS model input voltage (VIN) and filtered digitized output voltage (DIGITAL_OUT)

Reference MEMS and Engineering Power Evaluation Platform

ADI’s wired condition monitoring platform provides an industrial wired link solution for the ADcmXL3021 triaxial vibration sensor. The hardware signal chain consists of the ADcmXL3021 accelerometer with SPI and interrupt outputs connected to the interface PCB, which converts the SPI to RS-485 physical layer to a remote host controller board over several meters of cable. The SPI to RS-485 physical layer conversion can be implemented using an isolated or non-isolated interface PCB, which includes iCoupler® isolation (ADuM5401/ADuM110N0) and RS-485/RS-422 transceivers (ADM4168E/ADM3066E). This solution combines power and data over a standard cable (engineering power), reducing the cost of cables and connectors for remote MEMS sensor nodes. A dedicated software GUI allows for simple configuration of the ADcmXL3021 device and capture of vibration data over long cables. The GUI software visualizes the data as raw time domain or FFT waveforms.

Figure 14. Wired vibration monitoring powered by data lines.

in conclusion

Modern MEMS sensor solutions are small, highly integrated, and placed near the vibration source to measure the vibration frequency. Changes in frequency over time indicate a problem with the vibration source (motor, generator, etc.). Frequency measurement is critical for CbM. Using an engineered power solution can save connector count and cable costs for MEMS sensors. LTspice is a powerful free simulation tool that can be used to simulate engineered power designs. Thousands of power device models, including the LTM8002 Silent Switcher device, can be used to complete the remaining design. Using the provided ADC, op amp, and MEMS models, a complete MEMS signal chain simulation can be achieved.

References

1 Hiperface DSL®—The digital evolution. SICK Sensor Intelligence, October 2020.

2 Richard Anslow and Dara O’Sullivan. “Enabling Reliable Wired Condition-Based Monitoring for Industry 4.0—Part 2.” Analog Devices, Inc., November 2019.

3 “IEEE 802.3bu-2016—IEEE Standard for Ethernet—Amendment 8: Physical Layer and Management Parameters for Single Balanced Twisted-Pair Ethernet with Power over Data Lines (PoDL).” IEEE, February 2017.

4 Andy Gardner. “PoDL: A Decoupling Network Demonstration.” Linear Technology, May 2014.

5 “EN 50155:2017 Electronic equipment for railway applications for use on railway vehicles”.

6 Gabino Alonso and Kris Lokere. “LTspice: Simulating SAR ADC Analog Inputs.” Analog Devices, Inc., November 2017.

7 Erick Cook. “Simulating Hybrid Continuous Sampling Systems with LTspice.” EDN Asia, January 2020.

8 Simon Bramble. “Analyzing Vibration Data in Condition-Based Monitoring Systems Using LTspice.” Analog Dialogue, Vol. 54

Issue 2, June 2020.

About the Author

Richard Anslow is a systems applications engineer in the Connected Motion and Robotics team within the Automation and Energy Business Unit at Analog Devices. His areas of expertise are condition-based monitoring and industrial communications design. He holds a B.Eng. and M.Eng. from the University of Limerick, Ireland. Contact him at richard.anslow@analog.com.