Analysis and Demonstration of a Current-Stiff, Isolated Capacitive Bridge Transformer

arXiv:2611.14416 · eess.SY, cs.SY · Submitted 2026-11-10 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Robotics Radio. Generated commentary on the latest robotics and control papers.

Rosa: We're looking at the paper: "Analysis and Demonstration of a Current-Stiff, Isolated Capacitive Bridge Transformer".

Dev: Here's the gist of the paper. The paper presents a new DAB architecture based on Current Source Inverters coupled through and isolated by an AC capacitive bridge, yielding a system called a Capacitive Bridge Transformer (CBT) Pruning redundant reasoning tokens The traditional Voltage Source Inverter (VSI) based Dual Active Bridge (DAB) often has the transformer as the limiting factor of the DAB, especially as semiconductor switching speeds improve, due to steel core losses and winding losses This work introduces a new DAB architecture called a Capacitive Bridge Transformer (CBT) that eliminates all magnetic components in the AC portion of the isolated DAB circuit, which reduces frequency scaling barriers normally present in DABs How the training signal is built The CBT concept is best introduced through duality with its magnetic DAB counterpart on a per-phase basis The classic “T” circuit of the transformer simplifies to a series circuit without Lm when magnetizing inductance is far bigger than the leakages, i.e., Lp >> Lp The power flow for the CBT topology can be defined in terms of the two source magnitudes, the phase shift between them, and the joining impedance exactly like a VSI-DAB topology or electric machine equivalent circuit Where the gains come from The power flow for a single-phase CBT can be defined using Thevenin equivalent, where the power sourced by the primary side is given in Equation (a) The real power flow expression for the CBT topology can be derived in a step-by-step manner exploiting the concept of symmetry, considering a 1/6 cycle The power flow expression for Case 2, when 1/6 < Φ < 1/3, is given by Equation (33) Pruning redundant reasoning tokens The analytical model was confirmed with an experimental prototype operating at 2 kW and 94.3% efficiency The experimental data closely follows the trends predicted by the analytical model The maximum efficiency achieved was 94.3%, at a phase shift of 0.20 (72 degrees) Where the gains come from The authors found that the power flow peaks at a phase shift of 0.25 (90 degrees), as expected The maximum power angle curve efficiency peaks at 94.2% with 5.3 A of primary DC link current and 800 W of output power Pruning redundant reasoning tokens The paper concludes that an isolated dual active bridge with no AC magnetic components was developed by means of current source inverters and an AC capacitive bridge network The CBT operational concepts and analytical model were confirmed with an experimental prototype operating at 2 kW and 94.3% efficiency A future prototype will make use of four-quadrant bidirectional switches with low on resistance to further increase efficiency by eliminating diode forward voltage drop Where the gains come from The experimental results verified the analytical model in its current form, but they also point to areas for development, including semiconductor losses and reactive power reduction The peak efficiency could be increased substantially by using bidirectional blocking monolithic devices to eliminate the series blocking diodes Pruning redundant reasoning tokens The power flow expression for Case 3, when 1/3 < Φ < 1/2, is given by Equation (40) The measured and modeled values correlate well in the region from 36 - 162 degrees Beyond that range, the experimental results diverge from the model given non-optimal parameters including saturation, ripple, and losses The maximum realistically achievable effective turns ratio of the CBT is somewhat limited Pruning redundant reasoning tokens The paper was supported in part by USA National Science Foundation Grant No.

Rosa: 2328137 This work has been submitted to the IEEE for possible publication The authors thank the Wisconsin Electric Machines and Power Electronics Consortium (WEMPEC) and its members for their continued support Where the gains come from The circuit diagram of the prototype Capacitive Bridge Transformer is shown in Figure 9 The measured DC-DC power of the prototype versus primary-to-secondary phase shift Φ for a constant primary DC link current of 8 A is plotted in Figure 12 The analytical model curve and experimental data points correlate well, supporting the fidelity of the power flow analysis The ripple seen in the red mutual capacitor current is due to the parasitic inductance of the mutual capacitors The shapes of the AC capacitor waveforms align with the underlying assumptions made for the analytical model in Section III Pruning redundant reasoning tokens The analysis and demonstration of a current-stiff, isolated capacitive bridge transformer is presented The analytical models are verified by steady state and detailed time domain measurements on a 2 kW experimental prototype This work was supported in part by USA National Science Foundation Grant No.

Dev: 2328137 The paper has been submitted to the IEEE for possible publication The authors thank the Wisconsin Electric Machines and Power Electronics Consortium (WEMPEC) and its members for their continued support Where the gains come from The analysis and demonstration of a current-stiff, isolated capacitive bridge transformer is presented The analytical models are verified by steady state and detailed time domain measurements on a 2 kW experimental prototype The authors thank the Wisconsin Electric Machines and Power Electronics Consortium (WEMPEC) and its members for their continued support The circuit diagram of the prototype Capacitive Bridge Transformer is shown in Figure 9 The measured DC-DC power of the prototype versus primary-to-secondary phase shift Φ for a constant primary DC link current of 8 A is plotted in Figure 12 The analytical model curve and experimental data points correlate well, supporting the fidelity of the power flow analysis The ripple seen in the red mutual capacitor current is due to the parasitic inductance of the mutual capacitors The shapes of the AC capacitor waveforms align with the underlying assumptions made for the analytical model in Section III The authors thank the Wisconsin Electric Machines and Power Electronics Consortium (WEMPEC) and its members for their continued support Page 3 --- P(θ) = Irms p / Irms s sin(θ) /

1/Cp + 1/Cm + 1/C'm + 1/Cs: (3) However, a true implementation of a CBT circuit utilizing CSIs will have chopped rectangular currents, significantly complicating the calculations but will retain the same physical operation The real power flow, internal apparent power, and the voltage/current transfer ratio expressions are derived in a step-by-step manner exploiting the concept of symmetry Page 5 --- Pp(0 < φ < 1) = -Tsw /

36CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs φ(4 − φ): (27) Pp(0 < Φ < 1/6) = -Tsw /

3CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs Φ(2 − 3Φ): (20) Vs DC / Vp DC = Ip DC / Is DC = -

3CpCsRs / Tsw 3/Cm + 1/Cp + 1/Cs Φ(2 − 3Φ): (21) Vs DC / Vp DC = Ip DC / Is DC = -

3CpCsRs / Tsw 3/Cm + 1/Cp + 1/Cs Φ(2 − 3Φ): (21) Page 7 --- Pp(0 < φ < 1) = -Tsw /

36CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs (3 + 2φ − 2φ 2): (39) Pp(0 < φ < 1) = -Tsw /

36CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs (4 - φ): (27) Page 9 --- Pp(0 < φ < 1) = -Tsw /

36CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs (4 - φ): (27) Pp(0 < φ < 1) = -Tsw /

36CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs (4 - φ): (27)

Paper summary: Rosa: So we're looking at this paper called "Analysis and Demonstration of a Current-Stiff, Isolated Capacitive Bridge Transformer." Essentially, they've taken the standard Dual Active Bridge or DAB and replaced the magnetic transformer with something completely different.

Dev: That sounds like they're trying to get rid of all the traditional magnetic parts in that isolated circuit. The main idea is using current source inverters coupled through and isolated by an AC capacitive bridge, which they call a Capacitive Bridge Transformer or CBT.

Rosa: Right, so what's the big problem they're solving with this shift? Because traditionally, when you use a standard DAB with a magnetic transformer, that transformer itself often becomes the bottleneck for how fast you can scale up the frequency of operation because of things like steel core losses and winding losses.

Dev: Exactly. This CBT architecture aims to eliminate all those magnetic components in the AC part of the circuit, which should help them break those frequency scaling barriers that are usually there with standard DABs.

Taro: From my side, I'm thinking about what this means for real-world operation when things aren't perfect. If you get rid of the magnetic transformer, you trade one set of losses for another, and we need to see how robust it is when the world throws us unexpected loads or weird switching behavior.

Rosa: That's a fair point about robustness. The paper introduces this concept by looking at the duality with a magnetic DAB on a per-phase basis, showing how things simplify if the magnetizing inductance is much bigger than its leakages.

Dev: They model the power flow in terms of two source magnitudes, the phase shift between them, and that joining impedance exactly like they would for a standard Voltage Source Inverter based DAB or an electric machine equivalent circuit.

Taro: So it's essentially showing that mathematically, you can define the power transfer using familiar concepts even though the physical hardware is totally different from what we usually see on the shop floor.

Rosa: And they actually derive the real power flow for a single phase CBT by using a Thevenin equivalent, and then they show specific expressions for different ranges of that phase shift, like when it's between one sixth and one third.

Paper summary: Dev: They give us equations, like equation twenty-seven and twenty," which define how the power flows across the system based on that phase shift. It looks like they're mapping out the transfer characteristics very clearly for different operating conditions.

Taro: I wonder how accurate those mathematical models are when you move from a clean simulation to something that has actual chopped rectangular currents in a real Current Source Inverter implementation. That's where the practical challenges usually show up.

Rosa: Well, they did test it experimentally with a prototype operating at two kilowatts and they got ninety-four point three percent efficiency, which is pretty solid and matches the trends predicted by their analytical model quite closely.

Dev: Ninety-four point three percent at a phase shift of zero point two, or seventy-two degrees, is a good starting point for efficiency in this setup. They also found that the power flow peaks around a phase shift of zero point two five, or ninety degrees.

Taro: The peak happening at ninety degrees makes sense if you're thinking about where the energy transfer is maximized under these specific constraints of the circuit topology they designed.

Rosa: They also found that their maximum power angle curve efficiency peaks at ninety-four point two percent when there are five point three amps of primary DC link current and eight hundred watts of output power.

Dev: That's a concrete operating point, but I’m curious about what they say about the limitations. They mentioned that beyond a phase shift range between one third and one half, the experimental results diverge from their model because things like saturation and ripple start messing up the predictions.

Taro: So they admit that while the math works well in certain ranges, when you push those parameters too far, you run into real physical issues like ripple or saturation affecting the performance. That's a crucial detail for any system design.

Rosa: They also noted that the maximum realistically achievable effective turns ratio of this CBT is somewhat limited, which hints at where they might need to focus their next design iteration.

Dev: So, to wrap up what they showed in "Analysis and Demonstration of a Current-Stiff, Isolated Capacitive Bridge Transformer," it's a system that uses current source inverters and an AC capacitive bridge instead of magnetic components for isolation.

Paper summary: Rosa: It's important to remember that even though they developed this concept, the authors mentioned that their prototype could be improved by using four-quadrant bidirectional switches with low on resistance to reduce those diode forward voltage drops and increase efficiency further.

Taro: So the path forward isn't just better math, but also better semiconductor choices for the physical implementation. That tells me it's not just a theoretical exercise for them; they’re thinking about how to actually build it.

Dev: Exactly. They verified their analytical models with steady state and detailed time domain measurements on that two kilowatt prototype, which gives confidence in the fidelity of their power flow analysis when parameters are within the predicted range.

Rosa: The circuit diagram for this Capacitive Bridge Transformer is shown in Figure nine, and they plotted the measured DC-DC power against the primary to secondary phase shift for an eight amp primary DC link current.

Taro: That correlation between the model curve and those experimental data points supports their analysis, but we have to keep an eye on that ripple seen in the red mutual capacitor current, which they attributed to parasitic inductance in those capacitors.

Dev: So they've laid out a solid foundation with this paper by showing how the CBT operates under steady state conditions and giving us some specific power flow expressions like equation twenty-one.

Rosa: They also pointed to the fact that a true implementation with current source inverters will have chopped rectangular currents, which makes the calculations significantly more complicated but keeps the physical operation consistent.

Taro: It’s a trade-off between mathematical simplicity and real-world current behavior, isn't it? The paper seems to be paving the way for designing isolated systems where you can potentially simplify the magnetic component structure.

Dev: Ultimately, this work is about demonstrating an isolated dual active bridge using only current source inverters and a capacitive bridge network, confirming its operational concepts with that experimental prototype operating at ninety-four point three percent efficiency.

Rosa: And they've put it out to IEEE for possible publication, so it’s definitely material for anyone working on power electronics or energy conversion systems who is looking at how to reduce those magnetic losses in high-speed applications.

Conclusion: Rosa: So we're wrapping up this look at the paper, "Analysis and Demonstration of a Current-Stiff, Isolated Capacitive Bridge Transformer."

Dev: That paper essentially shows how you can build an isolated dual active bridge without any magnetic transformer components in that AC part.

Rosa: Right. The main takeaway here is taking something that usually has magnetic parts—the traditional DAB—and replacing them with an AC capacitive bridge network instead.

Taro: I think what they're doing is removing a whole class of losses, like the steel core and winding losses you get in those traditional transformers when you push the switching speeds really high.

Dev: Exactly. They're aiming to solve those frequency scaling barriers that usually limit how fast you can operate a DAB system because of those magnetic components.

Rosa: And they did it by using current source inverters and this capacitor bridge setup for isolation, which is a pretty neat way to do it.

Taro: The experimental results showed ninety-four point three percent efficiency on their two kilowatt prototype, which is a solid number when you're talking about real-world power conversion.

Dev: It's good that they verified their analytical models with those measurements, showing the math matches what they actually measured in steady state.

Rosa: But they also flagged some things where the model breaks down, like when you go beyond a certain phase shift range because of things like saturation and ripple from the capacitors.

Taro: So it's not just a perfect theoretical curve; you get real physical effects kicking in when the conditions get extreme.

Dev: And they did point out that for their next prototype, they need to use better switching devices, like four-quadrant bidirectional switches, to cut down on those diode losses further.

Rosa: So the implication is a path forward where you can design these isolated systems with simpler magnetic structures and then just focus on improving the semiconductor choices for maximum efficiency.

Taro: It means the next challenge isn't just designing a better circuit topology, but finding better hardware components to make it actually work at its best.

Patrick Nowakowski, Cameron Little, Mahima Gupta, Daniel C. Ludois

Wisconsin Electric Machines and Power Electronics Consortium (WEMPEC) · University of Wisconsin-Madison

eess.SY, cs.SY

Submitted: 2026-11-10

Updated: 2026-11-10

Comments: 10 pages, 17 figures

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 81/100

The gist: The paper presents a new DAB architecture based on Current Source Inverters coupled through and isolated by an AC capacitive bridge…

Key concepts

Capacitive Bridge Transformer (CBT)
This is a novel isolated dual active bridge topology that uses current source inverters and an AC capacitive bridge instead of traditional magnetic transformers. It eliminates all AC magnetic components, which helps overcome limitations found in standard DAB designs related to frequency scaling.
Dual Active Bridge (DAB)
A DAB is a power converter that uses two active bridges to transfer power between two DC sources. The CBT is an isolated version of this concept where the isolation and energy transfer are managed by an AC capacitive bridge rather than a physical transformer.
Phase Shift ($\Phi$ or $\theta$)
The phase shift represents the difference in timing or angle between the currents supplied to the two DC sources. The paper analyzes how this phase shift directly influences the real power flow and efficiency within the CBT circuit, showing that optimal operation occurs at specific shifts.
Power Flow Equations
These are mathematical expressions derived from symmetry and equivalent circuit analysis that describe how power is transferred between the primary and secondary sides of the CBT. They allow engineers to predict performance metrics like real power flow based on the phase shift and component values.

Terminology

Summary

The paper presents a new DAB architecture based on Current Source Inverters coupled through and isolated by an AC capacitive bridge, yielding a system called a Capacitive Bridge Transformer (CBT)

Pruning redundant reasoning tokens

The traditional Voltage Source Inverter (VSI) based Dual Active Bridge (DAB) often has the transformer as the limiting factor of the DAB, especially as semiconductor switching speeds improve, due to steel core losses and winding losses This work introduces a new DAB architecture called a Capacitive Bridge Transformer (CBT) that eliminates all magnetic components in the AC portion of the isolated DAB circuit, which reduces frequency scaling barriers normally present in DABs

How the training signal is built

The CBT concept is best introduced through duality with its magnetic DAB counterpart on a per-phase basis The classic “T” circuit of the transformer simplifies to a series circuit without Lm when magnetizing inductance is far bigger than the leakages, i.e., Lp >> Lp The power flow for the CBT topology can be defined in terms of the two source magnitudes, the phase shift between them, and the joining impedance exactly like a VSI-DAB topology or electric machine equivalent circuit

Where the gains come from

The power flow for a single-phase CBT can be defined using Thevenin equivalent, where the power sourced by the primary side is given in Equation (a) The real power flow expression for the CBT topology can be derived in a step-by-step manner exploiting the concept of symmetry, considering a 1/6 cycle The power flow expression for Case 2, when 1/6 < Φ < 1/3, is given by Equation (33)

The analytical model was confirmed with an experimental prototype operating at 2 kW and 94.3% efficiency The experimental data closely follows the trends predicted by the analytical model The maximum efficiency achieved was 94. p. 93%, at a phase shift of 0. p. 120 (72 degrees)

The authors found that the power flow peaks at a phase shift of 0.25 (90 degrees), as expected The maximum power angle curve efficiency peaks at 94. p. 72% with 5.3 A of primary DC link current and 800 W of output power

The paper concludes that an isolated dual active bridge with no AC magnetic components was developed by means of current source inverters and an AC capacitive bridge network The CBT operational concepts and analytical model were confirmed with an experimental prototype operating at 2 kW and 94.3% efficiency A future prototype will make use of four-quadrant bidirectional switches with low on resistance to further increase efficiency by eliminating diode forward voltage drop

The experimental results verified the analytical model in its current form, but they also point to areas for development, including semiconductor losses and reactive power reduction The peak efficiency could be increased substantially by using bidirectional blocking monolithic devices to eliminate the series blocking diodes

The power flow expression for Case 3, when 1/3 < Φ < 1/2, is given by Equation (40) The measured and modeled values correlate well in the region from 36 - 162 degrees Beyond that range, the experimental results diverge from the model given non-optimal parameters including saturation, ripple, and losses The maximum realistically achievable effective turns ratio of the CBT is somewhat limited

The paper was supported in part by USA National Science Foundation Grant No. 2328137 This work has been submitted to the IEEE for possible publication The authors thank the Wisconsin Electric Machines and Power Electronics Consortium (WEMPEC) and its members for their continued support

The circuit diagram of the prototype Capacitive Bridge Transformer is shown in Figure 9 The measured DC-DC power of the prototype versus primary-to-secondary phase shift Φ for a constant primary DC link current of 8 A is plotted in Figure 12 The analytical model curve and experimental data points correlate well, supporting the fidelity of the power flow analysis The ripple seen in the red mutual capacitor current is due to the parasitic inductance of the mutual capacitors The shapes of the AC capacitor waveforms align with the underlying assumptions made for the analytical model in Section III

The analysis and demonstration of a current-stiff, isolated capacitive bridge transformer is presented The analytical models are verified by steady state and detailed time domain measurements on a 2 kW experimental prototype This work was supported in part by USA National Science Foundation Grant No. 2328137 The paper has been submitted to the IEEE for possible publication The authors thank the Wisconsin Electric Machines and Power Electronics Consortium (WEMPEC) and its members for their continued support

The analysis and demonstration of a current-stiff, isolated capacitive bridge transformer is presented The analytical models are verified by steady state and detailed time domain measurements on a 2 kW experimental prototype The authors thank the Wisconsin Electric Machines and Power Electronics Consortium (WEMPEC) and its members for their continued support The circuit diagram of the prototype Capacitive Bridge Transformer is shown in Figure 9 The measured DC-DC power of the prototype versus primary-to-secondary phase shift Φ for a constant primary DC link current of 8 A is plotted in Figure 12 The analytical model curve and experimental data points correlate well, supporting the fidelity of the power flow analysis The ripple seen in the red mutual capacitor current is due to the parasitic inductance of the mutual capacitors The shapes of the AC capacitor waveforms align with the underlying assumptions made for the analytical model in Section III The authors thank the Wisconsin Electric Machines and Power Electronics Consortium (WEMPEC) and its members for their continued support

--- Page 3 ---

P(θ) = Irms p / Irms s sin(θ) / [1/Cp + 1/Cm + 1/C'm + 1/Cs] (3) However, a true implementation of a CBT circuit utilizing CSIs will have chopped rectangular currents, significantly complicating the calculations but will retain the same physical operation The real power flow, internal apparent power, and the voltage/current transfer ratio expressions are derived in a step-by-step manner exploiting the concept of symmetry

--- Page 5 ---

Pp(0 < φ < 1) = -Tsw / [36CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs φ(4 − φ)] (27) Pp(0 < Φ < 1/6) = -Tsw / [3CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs Φ(2 − 3Φ)] (20) Vs DC / Vp DC = Ip DC / Is DC = - [3CpCsRs / Tsw 3/Cm + 1/Cp + 1/Cs Φ(2 − 3Φ)] (21) Vs DC / Vp DC = Ip DC / Is DC = - [3CpCsRs / Tsw 3/Cm + 1/Cp + 1/Cs Φ(2 − 3Φ)] (21)

--- Page 7 ---

Pp(0 < φ < 1) = -Tsw / [36CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs (3 + 2φ − 2φ 2)] (39) Pp(0 < φ < 1) = -Tsw / [36CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs (4 - φ)] (27)

--- Page 9 ---

Pp(0 < φ < 1) = -Tsw / [36CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs (4 - φ)] (27) Pp(0 < φ < 1) = -Tsw / [36CpCs Ip DC Is DC 3/Cm + 1/Cp + 1/Cs (4 - φ)] (27)

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Precise power flow calculation for Capacitive Bridge Transformer (CBT) operation across all phase shift ranges is enabled by providing detailed analytical models for each case (Case 1: 0 < Phi < 1/6, Case 2: 1/6 < Phi < 1/3, and Case 3: 1/3 < Phi < 1/2). This allows the system to accurately predict internal apparent power, reactive power, and effective turns ratio based on the specific operating phase shift.

  2. Enhanced design optimization for DC-DC converters is possible by using the derived formulas to select component values, such as load resistance Rs and capacitor sizes (Cp, Cs), to achieve desired performance metrics like a target effective turns ratio or maximum efficiency at a specific phase shift.

  3. Real-time monitoring and fault detection capabilities can be implemented by comparing measured operational data against the analytical model predictions, which will flag deviations outside the expected ranges for effective turns ratio and power flow, potentially indicating issues like saturation or ripple.

Related papers