Cut emissions with integrated data, power isolation: Steps 5-7

Article By : Anand Reghunathan, Koteshwar Rao, Anant Kamath

In the final steps to minimise radiated emissions, the authors talk about using common-mode chokes and the need for careful measurement.

The controller or “cold” side of most isolated systems is connected to a system chassis that is in turn connected to protective earth (PE). This strong earth connection ties the controller side ground to a stable point and reduces the conversion of common-mode noise to electromagnetic radiation in the system. Therefore, where possible, the control-side of an isolated system should be strongly connected to protective earth.

The isolated ground of interfaces such as analog and digital I/O modules, and isolated RS-485 and CAN, can be AC-connected to protective earth with the use of high-voltage safety capacitors. This capacitive connection is used to provide a return path to protective earth for electrostatic discharge (ESD), surge and electrical fast transient (EFT) strikes on the bus and I/O cables. It also serves the purpose of reducing the common-mode noise on the isolated ground, thereby reducing emissions through the interface and I/O cables.

Figure 15 shows an example of an isolated RS-485 system using an integrated signal and power isolation device. The microcontroller ground is directly connected, and the isolated ground is capacitively connected to protective earth, minimising electromagnetic radiation from the input supply cables, as well as from the RS-485 bus.

PowerDataIsolation_TI_F15_201703_cr
Figure 15: Direct and capacitive connection to protective earth in an isolated RS-485 application.
 

Step 6: Using common-mode chokes

Long cables or wires connected to the system containing the DC/DC converter can potentially pick up high-frequency switching components from the converter and act as transmitter antennas. Due to this, longer cables result in higher radiation levels. The solution is to keep the common-mode current loop as small as possible and the cables as short as possible. In case the system cannot avoid the presence of long cables and provide a good connection between the ground and power earth/chassis, common-mode chokes (figure 16) can be used for attenuating the common-mode noise. For integrated isolated power solutions, a choke helps reduce the common current as well as return-path loop area, indirectly reducing radiated emissions. Common-mode chokes can be inserted on the input and output power supplies or on any long cables connected to the system.

PowerDataIsolation_TI_F16_201703_cr
Figure 16: Common-mode chokes on input and output power supply and I/O lines reduce emissions.
 

Step 7: Careful measurement & quasi-peak vs. peak detection

When performing emissions measurements, you need to make sure that only the emissions from the equipment under test (EUT)/device under test (DUT) are being measured, and not emissions from other parts of the setup. While the following practices do not lower radiated emissions, they are critical to determining the true performance of the system under test.

The primary reason for radiation is the formation of antennas on the board. Long cables used to power up the system or probes used to measure any parameter can act like antennas and cause a higher emissions reading. The solution is simple: the setup used for emissions should closely mimic the final system conditions in which it will be operating. Properly shield all supporting equipment other than the system being tested with a Faraday shield: any container lined with a conductive metal like copper or aluminium. Keep cables connected to and originating from the system as short as possible and keep the module to which the cables are connected within the Faraday shield.

As shown in Figure 17, the board containing the IC is exposed for measuring the radiated emissions. The rest of the equipment—a 12V battery and low-dropout regulator (LDO) in this case—are inside the black box and lined with aluminium foil to shield the components inside from the measurements. The cables connecting the LDO to the board are as short as possible and in a tight twisted-pair configuration. The cables extend out through a small hole at the top of the box. This hole is as small as possible to avoid compromising the effect of the Faraday shield.

PowerDataIsolation_TI_F17_201703_cr
Figure 17: Emissions setup (CISPR 22).

If the power supply must travel over long wires, then common-mode chokes are recommended near the DUT, which will help to not unnecessarily increase the emissions. By implementing common-mode chokes near the DUT, the emissions of the actual setup are measured and the effect of the long cables will be nullified.

In case you need to make some modifications to the board in the midst of testing, such as adding extra components, always solder the component directly onto the board rather than connecting it to the board using long wires.

According to the CISPR 22 standard, the radiated emissions specification is specified as a quasi-peak limit, although you can usually use the peak-detector measurement to get a quick result.

The reduction in measured radiated emissions when doing a quasi-peak scan is evident in cases when the device/system uses a kind of clock dithering that can change the switching frequencies across a small band instead of concentrating all of the power at one single peak. Techniques like this can bring down the peak scan levels and will also show significantly better results when subjected to quasi-peak scans. At lower load operations, the power of radiated emissions is less, since the DC/DC converter is on for a much shorter time compared to the full load condition. As expected, this lower power is reflected as a significant improvement in the quasi-peak results.

We advise taking a peak-detector measurement first to find out the frequencies for the worst-case measurements, and follow with a quasi-peak measurement at those frequencies to estimate the true margin from the CISPR 22B quasi-peak limit line.

Table 1 shows that an additional margin to CISPR 22B can show up after performing quasi-peak (QP) measurement vs. a peak-detector measurement.

PowerDataIsolation_TI_T1_201703_cr
Table 1: Quasi-peak results.
 

Conclusion

Integrated data and power isolation devices simplify system designs and reduce board area. The use of low-inductance chip-scale transformers in these devices necessitates the use of high-frequency switching, resulting in higher radiated emissions compared to solutions with discrete transformers. Techniques such as lower supply operation, the use of interlayer stitching capacitance, filters and common-mode chokes can reduce radiated emissions at the system level. Take care during emissions measurement to ensure that only the system under test is exposed, and that all other input and probe cables are completely shielded.

Anand Reghunathan is an applications engineer with Texas Instruments in India.
Koteshwar Rao has been lead applications engineer with Texas Instruments since Oct 2009.
Anant S Kamath is a systems engineering manager with Texas Instruments.

Leave a comment