How Next-Generation Isolated Σ-Δ Modulators Improve System-Level Current Measurement

Isolated modulators are widely used in motors/inverters that require high-precision current measurement and current isolation. As motor/inverter systems move toward high integration and high efficiency, SiC and GaN FETs are beginning to replace MOSFETs and IGBTs due to their smaller size, higher switching frequency, and lower heat generation. However, isolation devices need to have high CMTI capabilities, and higher-precision current measurement is also required. The next generation of isolated modulators has greatly improved CMTI capabilities and improved its own accuracy.

Common mode transient immunity specifies the rate at which transient pulses, applied at the critical condition of the insulation, rise and fall. If exceeded, data or clock corruption may result. Both the rate of change of the pulse and the absolute common mode voltage are recorded.

The new isolated modulator was tested under both static and dynamic CMTI conditions. The static test detects single bit errors from the device. The dynamic test monitors the filtered data output to observe the change in noise performance in the random application of CMTI pulses. The detailed test block diagram is shown in Figure 1.

CMTI is important because high slew rate (high frequency) transients can corrupt data transmission across the isolation barrier. Understanding and measuring the effects of these transients on the device is critical. ADI’s test methodology is based on the IEC 60747-17 standard, which covers the common mode transient immunity (CMTI) measurement method for magnetic couplers.

The simplified CMTI test platform includes the following items, as shown in Figure 1:

• Battery power supply for VDD1/VDD2.

• High common voltage pulse generator.

• Oscilloscope for monitoring data.

• A data acquisition platform for analyzing the data and a 256x decimate sinc3 filter for isolating the modulator.

• Isolation module (usually using optical isolation).

• Isolation modulator.

Static and dynamic CMTI tests use the same platform, only the input signals are different. This platform can also be used to test the CMTI performance of other isolation products. For the isolated modulator, a single bit stream data is extracted and filtered before being transmitted to the control loop in the motor control system, making the dynamic CMTI test performance more comprehensive and useful. Figures 2 and 3 show the time domain and frequency domain CMTI dynamic test performance at different CMTI levels. As can be seen in Figure 2, for the same isolated modulator, the spurs become larger when a higher VCM transient signal is applied. When the VCM transient signal exceeds the isolation modulator specification, very large spurs appear in the time domain (as shown in Figure 2c). This can have serious consequences in the motor control system, resulting in large torque ripple.

Figure 3. Frequency domain dynamic CMTI performance

Figure 3 shows the FFT domain performance under different frequency transients (i.e., maintaining the VCM transient level by varying the transient period). The results in Figure 3 show that the harmonics are highly correlated with the transient frequency. Therefore, the higher the CMTI capability of the isolated modulator, the lower the noise level in the FFT analysis. Compared to the previous generation of isolated modulators, the next generation ADuM770x devices increase the CMTI capability from 25 kV/μs to 150 kV/μs, greatly improving the system transient immunity, as shown in the comparison data in Table 1.

Table 1. Comparison of main specifications

In motor control or inverter systems, the more accurate the current data is, the more stable and efficient the system will be. Offset and gain errors are common sources of DC errors in ADCs. Figure 4 shows how offset and gain errors affect the ADC transfer function. These errors can affect the system in the form of torque ripple or speed ripple. For most systems, these errors can be calibrated out at ambient temperature to limit the error impact.

Otherwise, offset drift and gain error over temperature become problematic because they are more difficult to compensate for. For converters with linear and predictable drift curves, it is possible (albeit costly and time consuming) to compensate for offset and gain error drift by adding compensation factors to the curves to make the offset drift curve as flat as possible, given the system temperature. Details of this compensation method are provided in Application Note AN-1377. This method can reduce the drift values ​​specified in the AD7403/AD7405 data sheet by up to 30% for offset drift and up to 90% for gain error drift. This method can be applied to any other conversion device when it is desired to improve offset and gain error drift at the system level.

There is another design called chopping technology, which is more efficient and convenient for system designers, and the chopping function can also be well integrated with the silicon itself to minimize offset and gain error drift. The chopping solution is shown in Figure 5. The solution implemented on the ADC is to chop the entire analog signal chain to eliminate all offsets and low-frequency errors.

The differential inputs to the modulator are alternately inverted (or chopped) at the input multiplexer, and an ADC conversion is performed for each phase of the chop (the multiplexer switches to either the 0 or 1 state). The modulator chop is inverted in the output multiplexer, and the output signal is then fed into a digital filter.

If the offset in the Σ-Δ modulator is represented as V _OS , then when chopping is 0, the output is (A _IN (+) − A _IN (−)) + V _OS ; when chopping is 1, the output is −[(A _IN (−) − A _IN (+)) + V _OS ]. The error voltage, V _{OS ,} is removed by averaging these two results in the digital filter, yielding (A _IN (+) − A _IN (−)), which is equal to the differential input voltage without any offset terms.

The latest isolated modulators improve the performance related to offset and gain errors by optimizing the internal analog design and using the latest chopping technology, which greatly simplifies the system design and reduces the calibration time. The latest ADuM770x devices have very high isolation and excellent ADC performance. An LDO version is also available, which simplifies the power supply design of the system.

A typical current measurement circuit for a motor system is shown in Figure 6. Although three phase current measurement circuits are required in the system, only one is shown in the block diagram. The other two phase current measurement circuits are similar and are represented by blue dashed lines. From the phase current measurement circuit, it can be seen that one side of the RSHUNT resistor is connected to the input of the ADuM770x-8. The other side is connected to the high voltage FET (which can be an IGBT or MOSFET) and the motor. When the high voltage FET changes state, there will always be overvoltage, undervoltage or other voltage instability. Accordingly, the voltage fluctuation of the _RSHUNT resistor will be transmitted to the ADuM770x-8, and the relevant data will be received on the DATA pin. The layout and system isolation design can improve or worsen the voltage instability, thus affecting the phase current measurement accuracy.

The recommended circuit setup is shown in Figure 6:

• VDD1/VDD2 decoupling requires 10 μF/100 nF capacitors which should be placed as close as possible to the corresponding pins.

• A 10 Ω/220 pF RC filter is required.

• An optional differential capacitor is recommended to reduce the noise impact of the shunt. Place this capacitor close to the IN+/IN– pins (0603 package is recommended).

• When the digital output lines are long, it is recommended to use an 82 Ω/33 pF RC filter. For good performance, consider using a shielded twisted pair cable.

• For higher performance requirements, consider using a 4-terminal shunt resistor.

Good layout is also essential to achieve the best performance. The recommended layout is shown in Figure 7. It is recommended to use differential pair routing between the shunt resistor and the IN+/IN– input pins to enhance common-mode rejection. The 10 Ω/220 pF filter should be placed as close to the IN+/IN– input pins as possible. The 10 μF/100 nF decoupling capacitor should be placed close to the VDD1/VDD2 power supply pins. It is recommended to place a portion of the ground plane GND1 under the input related circuits to improve signal stability. For the independent GND1 line (shown in purple and parallel to the differential pair trace), a star connection is required from the shunt resistor to the ADuM770x-8 GND pin to reduce the impact of power supply current fluctuations.

The latest ADuM770x isolated Σ-Δ modulators increase CMTI to 150 kV/μs and improve temperature drift performance, which is very beneficial for current measurement applications. Using the recommended circuit and layout during the design phase will be helpful.