Implementing isolated shunt sensors using isolated delta sigma modulators & digital filters - Part 1
Resources
[MUSIC PLAYING]
In this video within the Implementing Isolated Shunt Sensors Using Isolated Delta Sigma Modulators and Digital Filters section of the training series, we will cover different reference designs that use an isolated modulator with the MSP430F67641's internal digital filters for implementing a class 0.5 S accuracy polyphase meter.
The TIDA-00601 TIDesign puts the different building blocks of the SN6501 isolated DC-DC power supply with external transformer, AMC1304 isolated delta-sigma modulator, and the MSP430F67641 to implement a class 0.5 S three-phased energy measurement system with galvanically isolated shunt current sensors. In this design, there is one AMC1304 and high side power supply for each phase. The AMC1304 device is a reinforced isolated modulator with integrated LDO, which means an external LDO is not needed in the high side power supply of each phase.
Through the AMC1304, the design has shunt isolation with working isolation voltage up to one kilovolts AC, and peak isolation voltages of seven kilovolts. In addition to sensing the voltage and decimating the bitstream from the AMC1304, the MSP430F67641 drives the LCD, takes care of RS-232 communication to a PC GUI, as well as calculates standard metrology parameters, such as active energy; reactive energy; active, reactive, and apparent power; RMS voltage and current; power factor; as well as line frequency.
Using the same board, a software variant of the design that added more power quality features was created. This design calculated the same metrology parameters as TIDA-00601, and also added fundamental voltage, fundamental current, fundamental active power, fundamental reactive power, current THD, and voltage THD. In addition, it also calculated the phase to phase angle between voltages, which could be used to determine phase sequence to help prevent incorrect installation of the system. In this design, since the output sample rate of the AMC1304 is relatively large, there's less attenuation results due to the roll-off from the delta-sigma sync cube filter attenuating harmonics.
Another alternative to powering the AMC1304 high side is to use a cap drop power supply instead of the previous SSN6501 base supply. Looking at the data sheet of the AMC1304, we found that the high side of each AMC1304 takes only a maximum of 6.5 milliamps. This 6.5 milliamps is assuming the maximum modulation clock frequency of 20 megahertz.
If we lower the modulation clock frequency down to 5 megahertz, we could further reduce the current consumption of the AMC1304 high side, as shown in the figure on the right. This current consumption is small enough so that a cap drop power supply could be used to easily power the AMC1304's high side. In addition, since the AMC1304 already has an internal LDO, the cap drop power supply does not need an external LDO, as long as the voltage fed into the LDO in-pin of the AMC1304 is between 4 to 18 volts.
The TIDA-01094 design takes the original TIDA-00601 TIDesign and replaces the high side power supply with a cap drop power supply. With this new cap drop power supply, this design is completely transformerless. The high side for the AMC1304 is powered directly from mains.
In addition, the data isolation in the AMC1304 is based on TI's capacitor-based isolation technology. With the lack of transformers in the design, this design is completely immune to magnetic tamper attacks, instead of only being resistant to a certain extent. In addition, in this design, support was added for both the AMC1304M05, which has an integrated LDO, as well as an AMC1305 device, which doesn't have a LDO, which will couple with the TLV70450.
As mentioned before, the TIDA-00601 and TIDA-01088 TIDesigns use the same hardware and just have different software. Given the 19.8 megahertz modulation clock frequency in these designs and 1024 over-sample ratio, the delta-sigma output sample rate for these designs is 19.334 kilosamples per second. Since all of these samples cannot be processed, a portion of these samples are skipped. For the TIDA-00601 design, every three out of four samples are skipped, which leads to an effective sample rate of 4.8335 kilosamples per second. For the TIDA-01088 design, because more power quality calculations are done, it is required to further reduce the sample rate, so every four out of five samples are skipped, leading to a sample rate of 3.8668 kilosamples per second.
Comparing these two designs to the TIDA-01094 design, there are a few other differences besides only the high side power supply. First, the modulation clock frequency in TIDA-01094 is decreased from 19.8 megahertz to about 5 megahertz to reduce the current consumption of the high side. Given the OSR of 1024 used in this application, this results in the output sample rate of the delta-sigma being reduced from 19.334 kilosamples per second to 4.9152 kilosamples per second.
This sample rate is slow enough for the metrology software to process each sample without skipping. As a result, the software is simplified from the TIDA-00601 and TIDA-01088 designs. Due to the software simplification, hardware phase compensation via the MSP430 delta-sigma module can now be enabled in this design, compared to the other two designs, which require an FIR filter for phase compensation.
In addition, the lower modulation clock frequency enables more flexible clock divider settings, which allow the CPU clock frequency to be increased from 19.8 megahertz to 25 megahertz. There's a higher modulation clock frequency and sigma-delta ADC rate for TIDA-01088 than TDIA-01094. In this design, we skip samples to get an effective sample rate that is small enough so that enough bandwidth is available to form metrology calculations.
However, the data rate from the sigma-delta's perspective is still the same. With sync cube digital filters, there is degradation in amplitude accuracy at higher frequencies. However, if you decrease the ratio of input frequency to ADC data rate by increasing the sigma-delta ADC rate, there is less amplitude degradation from the sync cube filter. Therefore, in the TIDA-01088 design, by having a high sigma-delta ADC rate in TIDA-01088, we will have a small degradation from the sync cube filter, which means more accurate THD readings compared to other designs.
This slide shows the active energy error for the TIDA-01094 design, measured from a current range of 100 milliamps to 90 amps. These tests were conducted with a voltage to current phase shift of 0 degrees, minus 60 degrees, and plus-60 degrees. From these results, everything was within 0.5%.
A voltage variation test was also performed. In this test, the active energy error is measured across a range from 75 volts to 70 volts. This error was less than 0.2% across this range.
The test is important for two reasons. First, it shows that the 10-bit SAR ADC on the MSP430F67641A used to measure mains voltage has sufficient accuracy. Second, it shows the voltage range at which the AMC1304 would operate, which is more important for this design than the TIDA-00601 design, because the high side of each AMC1304 is powered from mains now instead of the controller side power supply.