Academia.eduAcademia.edu

Figure 9 – uploaded by Li Ran

See full PDFdownloadDownload figure
Fig. 10 High-frequency equivalent circuit. Fig. 10 shows an equivalent circuit of Fig. 8 [37]-[38], which is used to analyze the impact of the high-frequency flux density.

Figure 10 High-frequency equivalent circuit. Fig. 10 shows an equivalent circuit of Fig. 8 [37]-[38], which is used to analyze the impact of the high-frequency flux density.

Related Figures (34)

Dual active bridge (DAB) DC-DC converters, as shown in Fig. 1, is becoming a popular topology in applications such as DC microgrids and distribution networks [1], battery storage systems [2], on-board EV chargers [3], and power electronic transformers [4], due to bidirectional power capability, high power density, electrical isolation, and zero- voltage-switching (ZVS) [5].
Dual active bridge (DAB) DC-DC converters, as shown in Fig. 1, is becoming a popular topology in applications such as DC microgrids and distribution networks [1], battery storage systems [2], on-board EV chargers [3], and power electronic transformers [4], due to bidirectional power capability, high power density, electrical isolation, and zero- voltage-switching (ZVS) [5].
SOME STATE-OF-THE-ART CURRENT SENSORS ON THE MARKET These are five types of current sensors: shunt, Hall effect, GMR (giant magnetoresistance), fluxgate and OCT (optical current transducer). The shunt, Hall, OCT and GMR types are currently not suitable for the intended application due to large errors. The accuracy of a DC fluxgate current sensor is sufficient, but it cannot be used to measure DC mixed in a large AC current, as shown later. To handle the AC, the range of the current sensor needs to be relatively large. As the DC is small, using a sensor of a large range will increase the measurement error. Because there was no suitable current sensor, refs [27]-[31] proposed methods of measuring the transformer magnetic field directly to eliminate the DC bias.
SOME STATE-OF-THE-ART CURRENT SENSORS ON THE MARKET These are five types of current sensors: shunt, Hall effect, GMR (giant magnetoresistance), fluxgate and OCT (optical current transducer). The shunt, Hall, OCT and GMR types are currently not suitable for the intended application due to large errors. The accuracy of a DC fluxgate current sensor is sufficient, but it cannot be used to measure DC mixed in a large AC current, as shown later. To handle the AC, the range of the current sensor needs to be relatively large. As the DC is small, using a sensor of a large range will increase the measurement error. Because there was no suitable current sensor, refs [27]-[31] proposed methods of measuring the transformer magnetic field directly to eliminate the DC bias.
Fig.2 Asingle-core fluxgate DC current sensor. A single-core fluxgate can accurately measure a pure DC current [36]. Its structure is shown in Fig. 2. Wo~W2 are three separate windings on a toroidal core. Ipc denotes the measured DC current. W; is the magnetizing winding. v(t) and i;(t) represent the input voltage and magnetizing current, which are generated from a triangle-wave generator. If needed, resistance R; can prevent the triangle-wave generator from being short-circuited as the core saturates. Induced winding voltage v2(t) becomes a 0-3.3V square wave, v’2(f), after signal conditioning, as the fluxgate sensor output.
Fig.2 Asingle-core fluxgate DC current sensor. A single-core fluxgate can accurately measure a pure DC current [36]. Its structure is shown in Fig. 2. Wo~W2 are three separate windings on a toroidal core. Ipc denotes the measured DC current. W; is the magnetizing winding. v(t) and i;(t) represent the input voltage and magnetizing current, which are generated from a triangle-wave generator. If needed, resistance R; can prevent the triangle-wave generator from being short-circuited as the core saturates. Induced winding voltage v2(t) becomes a 0-3.3V square wave, v’2(f), after signal conditioning, as the fluxgate sensor output.
Fig. 3 Waveforms of v;(f), B)(¢) and v2(t) when the core is not saturated.
Fig. 3 Waveforms of v;(f), B)(¢) and v2(t) when the core is not saturated.
Fig.5 Waveforms of Bj(t), v(t) and v’2(t) when the core is saturated and Ipc 0. When /pc +0, it produces a DC flux density Bpc that is superimposed to Bi(t), which shifts B,(t) up or down, as shown in Fig. 5.
Fig.5 Waveforms of Bj(t), v(t) and v’2(t) when the core is saturated and Ipc 0. When /pc +0, it produces a DC flux density Bpc that is superimposed to Bi(t), which shifts B,(t) up or down, as shown in Fig. 5.
Fig. 4 Waveforms when the core is saturated and [pc=0.
Fig. 4 Waveforms when the core is saturated and [pc=0.
Fig. 7 Impact of DAB winding current on fluxgate, with N|=87 and N= 397 (numbers of turns of W; and W3). (a) experiment arrangement; (b) experiment result when a DAB winding current flows through Wo.
Fig. 7 Impact of DAB winding current on fluxgate, with N|=87 and N= 397 (numbers of turns of W; and W3). (a) experiment arrangement; (b) experiment result when a DAB winding current flows through Wo.
where vip and Bip are the peak values of vi(f) and B,(0), respectively. Ac is cross-sectional area of the sensor core, Nj the number of turns of W;, and f; the frequency of vi(¢). m is the waveform coefficient. When vj(f) is triangle wave, m=4. The range of the fluxgate sensor is determined by the difference between Bip and Bs. This is because Bjp-B; is to ensure that DC flux density produced by the measured DC current within the set range will not make the core exit saturation (i.e. -Bip+Bpc>-Bs or Bip-Bpc<Bs will not happen). The range can be expressed as
where vip and Bip are the peak values of vi(f) and B,(0), respectively. Ac is cross-sectional area of the sensor core, Nj the number of turns of W;, and f; the frequency of vi(¢). m is the waveform coefficient. When vj(f) is triangle wave, m=4. The range of the fluxgate sensor is determined by the difference between Bip and Bs. This is because Bjp-B; is to ensure that DC flux density produced by the measured DC current within the set range will not make the core exit saturation (i.e. -Bip+Bpc>-Bs or Bip-Bpc<Bs will not happen). The range can be expressed as
Fig. 8 Schematic diagram with a resistor connected across W>. Fig.9 High-frequency voltage in fluxgate windings when W) is connected to a 120 Q resistor and the DAB is fully loaded with ino(t) of 15A peak.
Fig. 8 Schematic diagram with a resistor connected across W>. Fig.9 High-frequency voltage in fluxgate windings when W) is connected to a 120 Q resistor and the DAB is fully loaded with ino(t) of 15A peak.
To reduce vni(t) and vyo(t), the output impedance of the sensor transformer must be reduced. Therefore, a relatively small resistor R2 (to be selected as 120 Q in this study) is connected across W2, as shown in Fig. 8. The peak values decrease significantly, as shown in Fig. 9.
To reduce vni(t) and vyo(t), the output impedance of the sensor transformer must be reduced. Therefore, a relatively small resistor R2 (to be selected as 120 Q in this study) is connected across W2, as shown in Fig. 8. The peak values decrease significantly, as shown in Fig. 9.
Fig. 11 Equivalent circuit of proposed sensor focusing on triangle-wave frequency component.
Fig. 11 Equivalent circuit of proposed sensor focusing on triangle-wave frequency component.
According to (4), the leakage inductance accounts for part of vn2(t). Due to Ly being very small compared to the self-inductance of Wo, it has little effect on the fluxgate. According to KVL, Faraday's law and the transformer principle, the relationship between these variables on the three windings is
According to (4), the leakage inductance accounts for part of vn2(t). Due to Ly being very small compared to the self-inductance of Wo, it has little effect on the fluxgate. According to KVL, Faraday's law and the transformer principle, the relationship between these variables on the three windings is
The complete model of the proposed sensor is shown in Fig. 12, which combines the models shown in Fig. 10 and Fig. 11, and also includes Jpc. Here, vwo(t), vwi(t) and vwa(t) are the total voltages of Wo, Wi, Wo. iwo(t), iwi(t) and iyo(A) are the total currents. A. Sensor requirements It can be seen from Fig. 12 that Wo, Wi and W2 are coupled by flux density B(t) in the core, which satisfies
The complete model of the proposed sensor is shown in Fig. 12, which combines the models shown in Fig. 10 and Fig. 11, and also includes Jpc. Here, vwo(t), vwi(t) and vwa(t) are the total voltages of Wo, Wi, Wo. iwo(t), iwi(t) and iyo(A) are the total currents. A. Sensor requirements It can be seen from Fig. 12 that Wo, Wi and W2 are coupled by flux density B(t) in the core, which satisfies
To realize the fluxgate function, the CT (Wi, W2, the toroidal core and R2) core still needs to be saturated. In the CT, W; is regarded as a current source, hence, the peak value of v2(f), i.e. v2p, determines core saturation and the sensor range. The relationship satisfies
To realize the fluxgate function, the CT (Wi, W2, the toroidal core and R2) core still needs to be saturated. In the CT, W; is regarded as a current source, hence, the peak value of v2(f), i.e. v2p, determines core saturation and the sensor range. The relationship satisfies
Fig. 13 Configuration of the proposed sensor. Based on the above analysis, the frequency of v2(f) (i.e. triangle-wave frequency) must be considerably lower than the frequency of vn2(f) (i.e. the DAB switching frequency), so that it is convenient to filter out vn2(t) by a low-pass filter. In line with this, the signal processing circuit of fluxgate has been redesigned, as shown in Fig. 13. There are two low-pass filters and a differential amplifier before the signal conditioner. A common-mode inductor is added on Wj. This is because winding parasitic capacitance can cause coupling with the DAB operation giving rise to common-mode interference that can seriously affect the signal conditioning circuit. This is particularly important in DAB units with high DC bus voltages. III. PARAMETER DESIGN OF SENSOR SYSTEM
Fig. 13 Configuration of the proposed sensor. Based on the above analysis, the frequency of v2(f) (i.e. triangle-wave frequency) must be considerably lower than the frequency of vn2(f) (i.e. the DAB switching frequency), so that it is convenient to filter out vn2(t) by a low-pass filter. In line with this, the signal processing circuit of fluxgate has been redesigned, as shown in Fig. 13. There are two low-pass filters and a differential amplifier before the signal conditioner. A common-mode inductor is added on Wj. This is because winding parasitic capacitance can cause coupling with the DAB operation giving rise to common-mode interference that can seriously affect the signal conditioning circuit. This is particularly important in DAB units with high DC bus voltages. III. PARAMETER DESIGN OF SENSOR SYSTEM
PARAMETERS OF 1KW EXPERIMENTAL DAB With the data in TABLE I, Jpc is about 0.2 A on the primary and 0.4 A on the secondary side. The sensor range is set at three times of 0.4 A, i.e. +1.2 A.
PARAMETERS OF 1KW EXPERIMENTAL DAB With the data in TABLE I, Jpc is about 0.2 A on the primary and 0.4 A on the secondary side. The sensor range is set at three times of 0.4 A, i.e. +1.2 A.
Fig. 14 Way for DSP to measure duty cycle, by counting clock pulses in the range of the signal conditioner output which represents DC current. In this study, the DSP used to control the DAB is also used to measure the DC current, by counting its own pulses in the duty cycle of the sensor’s output square wave, as shown in Fig. 14. Two observations are made here: if the resolution is higher for the triangle-wave or the DSP clock frequency, the resolution of the sensor out will be higher; if the triangle-wave frequency is lower or the target DC range is smaller, the resolution of the sensor output will also be higher, i.e. finer. Fig. 15 is the flow chart of the resolution design.
Fig. 14 Way for DSP to measure duty cycle, by counting clock pulses in the range of the signal conditioner output which represents DC current. In this study, the DSP used to control the DAB is also used to measure the DC current, by counting its own pulses in the duty cycle of the sensor’s output square wave, as shown in Fig. 14. Two observations are made here: if the resolution is higher for the triangle-wave or the DSP clock frequency, the resolution of the sensor out will be higher; if the triangle-wave frequency is lower or the target DC range is smaller, the resolution of the sensor output will also be higher, i.e. finer. Fig. 15 is the flow chart of the resolution design.
Fig. 15 Flow chart of resolution design. ,, the resolution of f; and the DSP clock frequency are selected from what is available. Only f, can be flexibly adjusted. The sensor resolution is designed through it.
Fig. 15 Flow chart of resolution design. ,, the resolution of f; and the DSP clock frequency are selected from what is available. Only f, can be flexibly adjusted. The sensor resolution is designed through it.
Fig. 16 Flowchart of power loss calculation. (13) is to meet the range requirement. (14)-(17) are to make the core saturate. (15)-(21) and (12) are to calculate the power loss. itp, iL2p, in2p, Ro, Ni and N2 from (12) need to be calculated for meeting the range at first. Then the core must be saturated
Fig. 16 Flowchart of power loss calculation. (13) is to meet the range requirement. (14)-(17) are to make the core saturate. (15)-(21) and (12) are to calculate the power loss. itp, iL2p, in2p, Ro, Ni and N2 from (12) need to be calculated for meeting the range at first. Then the core must be saturated
Fig. 17 Power loss curve when N, changes from 10 to 300. Given a value of Ni, the power loss can be calculated according to (12)-(21). When N; changes from 10 to 300, the power loss is shown in Fig. 17. N\=87 is the point of minimum power loss. Therefore N2=397, Ro~120 Q, iip~70 MA, v2p~1.4 V and Pioss*0.2 W.
Fig. 17 Power loss curve when N, changes from 10 to 300. Given a value of Ni, the power loss can be calculated according to (12)-(21). When N; changes from 10 to 300, the power loss is shown in Fig. 17. N\=87 is the point of minimum power loss. Therefore N2=397, Ro~120 Q, iip~70 MA, v2p~1.4 V and Pioss*0.2 W.
Fig. 19 Schematic diagram of the sensor circuit. The sensor circuit consists of two parts: the triangle-wave generator and the signal conditioning circuit, with functional blocks and components, as shown in Fig. 19. Photos of the sensor are shown in Fig. 20, with a 3 cmx3 cm PCB.
Fig. 19 Schematic diagram of the sensor circuit. The sensor circuit consists of two parts: the triangle-wave generator and the signal conditioning circuit, with functional blocks and components, as shown in Fig. 19. Photos of the sensor are shown in Fig. 20, with a 3 cmx3 cm PCB.
Fig. 18 Waveforms in sensor circuit (a)DAB no-load and Ipc=0 A; (b) DAB no-load and Jpc=1.2 A; (c) DAB is fully loaded and Ipc=0 A.
Fig. 18 Waveforms in sensor circuit (a)DAB no-load and Ipc=0 A; (b) DAB no-load and Jpc=1.2 A; (c) DAB is fully loaded and Ipc=0 A.
Fig. 20 Pictures of the current sensor.
Fig. 20 Pictures of the current sensor.
Fig. 21 Verification of sensor accuracy and resolution. A high-precision mu timeter, Fluke 8846A with error and resolution down to 100 pA, is first used to calibrate the designed sensor under flows through Wo to ma ke the reading of the multimeter to 1 mA. Under this circumstance, the value of DSP counter w change. The change is u the measured current. T hen adjusting the DC current, in ste pure DC. Adjust the DC current that be ill sed as a standard of 1 mA to calculate ps of 1 mAin [-10mA, 10mA] and larger at higher current levels. Comparing the output of the sensor, measured from the du cycle, with the multimeter reading, the difference is t measurement error of the sensor. The result is plotted in Fig. 21. The sensor output is sensitive and almost proportional DC current measured. A straight line can be used to curve the results. The discrepancy of curve fitting shows that the error and resolution could be as small as 1 mA. ty he to Fit
Fig. 21 Verification of sensor accuracy and resolution. A high-precision mu timeter, Fluke 8846A with error and resolution down to 100 pA, is first used to calibrate the designed sensor under flows through Wo to ma ke the reading of the multimeter to 1 mA. Under this circumstance, the value of DSP counter w change. The change is u the measured current. T hen adjusting the DC current, in ste pure DC. Adjust the DC current that be ill sed as a standard of 1 mA to calculate ps of 1 mAin [-10mA, 10mA] and larger at higher current levels. Comparing the output of the sensor, measured from the du cycle, with the multimeter reading, the difference is t measurement error of the sensor. The result is plotted in Fig. 21. The sensor output is sensitive and almost proportional DC current measured. A straight line can be used to curve the results. The discrepancy of curve fitting shows that the error and resolution could be as small as 1 mA. ty he to Fit
Fig. 22 Test circuit diagram. designed sensor. The magnitudes of the AC and DC components are changed to see whether the output of the sensor is the same as that in pure DC calibration. Set the DC current at 0, +1 mA or +1.2 A. At each point, the AC current changes from 0 to 15 A peak with a step of 1 A. The measurement errors are shown in Fig. 23 with the maximum occurring at full DAB load current and sensor full scale. As shown in Fig. 23, when the DC current does not change, as the AC current increases from small to large, the error increases. When the AC is constant, the error increases with the increase of the DC current. But the resolution did not change during the process. The maximum error is 4 mA at AC is 15 A, therefore, the proposed method can be applied from low current to high current.
Fig. 22 Test circuit diagram. designed sensor. The magnitudes of the AC and DC components are changed to see whether the output of the sensor is the same as that in pure DC calibration. Set the DC current at 0, +1 mA or +1.2 A. At each point, the AC current changes from 0 to 15 A peak with a step of 1 A. The measurement errors are shown in Fig. 23 with the maximum occurring at full DAB load current and sensor full scale. As shown in Fig. 23, when the DC current does not change, as the AC current increases from small to large, the error increases. When the AC is constant, the error increases with the increase of the DC current. But the resolution did not change during the process. The maximum error is 4 mA at AC is 15 A, therefore, the proposed method can be applied from low current to high current.
Fig. 23 Sensor output under different peak values of 20 kHz current. When using LAH 12 to measure the primary winding current (about 7.5 A), the error is 22.5 mA. When using it to measure secondary winding current (about 15 A), the error is 45 mA. Both of them are more than the error of proposed sensor. Therefore, LAH 12 has no superiority in the elimination of DC bias. The step response of the sensor is shown in Fig. 24. The measured DC current changes from 0 to | A. The response time of the sensor is between 10 ms and 40 ms. Therefore, the step response time is defined as 40 ms which is fast enough for DC bias elimination control.
Fig. 23 Sensor output under different peak values of 20 kHz current. When using LAH 12 to measure the primary winding current (about 7.5 A), the error is 22.5 mA. When using it to measure secondary winding current (about 15 A), the error is 45 mA. Both of them are more than the error of proposed sensor. Therefore, LAH 12 has no superiority in the elimination of DC bias. The step response of the sensor is shown in Fig. 24. The measured DC current changes from 0 to | A. The response time of the sensor is between 10 ms and 40 ms. Therefore, the step response time is defined as 40 ms which is fast enough for DC bias elimination control.
Fig. 24 Step response of the sensor. error is less than | mA when the measured DC current is 100 mA. Fig. 25(c) shows that the maximum error is less than 1.5 mA when the measured current includes 100 mA DC and 15 A 20 kHz AC. The reason why sensor steadiness is not tested at 1.2 Ais that as DC bias elimination control is activated, DC current will decrease. To virtually eliminate the DC, it is necessary to measure accurately when DC current is small. 100 mA is chosen arbitrarily in this study.
Fig. 24 Step response of the sensor. error is less than | mA when the measured DC current is 100 mA. Fig. 25(c) shows that the maximum error is less than 1.5 mA when the measured current includes 100 mA DC and 15 A 20 kHz AC. The reason why sensor steadiness is not tested at 1.2 Ais that as DC bias elimination control is activated, DC current will decrease. To virtually eliminate the DC, it is necessary to measure accurately when DC current is small. 100 mA is chosen arbitrarily in this study.
Fig. 25 Sensor output in 30 minutes, using the experimental method of Fig. 22. (a) without measured current; (b) with 100 mA DC; (c) with 100 mA DC and 15 A (peak) 20 kHz AC.
Fig. 25 Sensor output in 30 minutes, using the experimental method of Fig. 22. (a) without measured current; (b) with 100 mA DC; (c) with 100 mA DC and 15 A (peak) 20 kHz AC.
DAB AND TRANSFORMER LOSS CORRESPONDING TO FIG. 27 AND FIG. 28
DAB AND TRANSFORMER LOSS CORRESPONDING TO FIG. 27 AND FIG. 28
Fig. 26 Control block diagram to eliminate DC bias. Because the DC component is a small part of the winding current, eliminating it in the two windings is not easy to illustrate. However, the DC component in the magnetizing current can be more clearly shown. Therefore, the DC components in the magnetizing (viewed from the secondary side) and primary winding currents are thus measured and eliminated to demonstrate the concept of control. The block diagram is shown in Fig. 26 (n=2). The output of the PI block, Ad, is added to the duty cycle of the H-bridge to adjust the small DC voltage applied to the winding.
Fig. 26 Control block diagram to eliminate DC bias. Because the DC component is a small part of the winding current, eliminating it in the two windings is not easy to illustrate. However, the DC component in the magnetizing current can be more clearly shown. Therefore, the DC components in the magnetizing (viewed from the secondary side) and primary winding currents are thus measured and eliminated to demonstrate the concept of control. The block diagram is shown in Fig. 26 (n=2). The output of the PI block, Ad, is added to the duty cycle of the H-bridge to adjust the small DC voltage applied to the winding.
The DAB transformer is at a higher level of DC bias in the case corresponding to Fig. 27(b). But in TABLE III the power loss is only modestly increased as compared to the case of Fig. 27(a). This is because a large margin has been reserved in the DAB transformer design. If the margin is reduced in the future, the effect of eliminating DC bias would be greater.
The DAB transformer is at a higher level of DC bias in the case corresponding to Fig. 27(b). But in TABLE III the power loss is only modestly increased as compared to the case of Fig. 27(a). This is because a large margin has been reserved in the DAB transformer design. If the margin is reduced in the future, the effect of eliminating DC bias would be greater.
Fig. 28 Demonstration of dynamic DC elimination control. Channels from top down: magnetizing current (i) captured by CP8050A; primary winding current (ip) captured CP8050A; DC in magnetizing current (im) measured by the output duty cycle of the proposed sensor; DC in the primary winding current (ip) measured by output duty cycle proposed sensor Approximately, the first 100 ms shown in Fig. 28 is the case when the controller is arbitrarily prohibited. The winding current contains a DC component of about 45 mA and the magnetizing current contains a DC component of about 668 mA (viewed from the secondary side). The loss of the DAB is high (TABLE III). This is because the DC components are additional and superimposed on the original AC currents and hence will increase the loss. The reason why the loss is less significant in the transformer is that the winding resistance is small. The on-resistance of devices in the DAB H-bridges is greater than the resistance of the transformer windings. As a result, the loss effect of the DC components on the DAB H-bridges is more noteworthy.
Fig. 28 Demonstration of dynamic DC elimination control. Channels from top down: magnetizing current (i) captured by CP8050A; primary winding current (ip) captured CP8050A; DC in magnetizing current (im) measured by the output duty cycle of the proposed sensor; DC in the primary winding current (ip) measured by output duty cycle proposed sensor Approximately, the first 100 ms shown in Fig. 28 is the case when the controller is arbitrarily prohibited. The winding current contains a DC component of about 45 mA and the magnetizing current contains a DC component of about 668 mA (viewed from the secondary side). The loss of the DAB is high (TABLE III). This is because the DC components are additional and superimposed on the original AC currents and hence will increase the loss. The reason why the loss is less significant in the transformer is that the winding resistance is small. The on-resistance of devices in the DAB H-bridges is greater than the resistance of the transformer windings. As a result, the loss effect of the DC components on the DAB H-bridges is more noteworthy.
To further analyze the relationship between the DC bias and transformer loss, the loss was measured with the power analyzer under different DC bias conditions. The results are shown in Fig. 29. It is found that when the DC component of the magnetizing current is less than 200 mA the loss is almost constant. But when the bias is over 200 mA the loss begins to increase quickly. This is consistent with the allowable DC current of windings calculated in the previous section. When the flux density reaches the non-linear region of the B-H curve, the loss increases. In this study, the margin of the magnetic core area is relatively large; the allowable DC component of the magnetizing current is about 200 mA. With the proposed sensor, this margin can be reduced, thereby reducing the size of the DAB transformer.
To further analyze the relationship between the DC bias and transformer loss, the loss was measured with the power analyzer under different DC bias conditions. The results are shown in Fig. 29. It is found that when the DC component of the magnetizing current is less than 200 mA the loss is almost constant. But when the bias is over 200 mA the loss begins to increase quickly. This is consistent with the allowable DC current of windings calculated in the previous section. When the flux density reaches the non-linear region of the B-H curve, the loss increases. In this study, the margin of the magnetic core area is relatively large; the allowable DC component of the magnetizing current is about 200 mA. With the proposed sensor, this margin can be reduced, thereby reducing the size of the DAB transformer.
Fig. 29 Effect of DC bias on (a) transformer loss and (b) efficiency.
Fig. 29 Effect of DC bias on (a) transformer loss and (b) efficiency.
Related topics:
Electrical EngineeringMagnetic separationSaturation MagnetizationCurrent TransformersDC BiasCurrent measurementElectrical and Electronic EngineeringTransformer CoresFlyback ConverterCurrent sensor
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved