Transmon Phase Gates Controlled by Superconducting Soliton DAC
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Quantum Radio. Generated commentary on the latest quantum physics and condensed matter papers.
Kai: I'm Kai.
Mira: And I'm Mira, here with Lev.
Kai: We're looking at the paper: "Transmon Phase Gates Controlled by Superconducting Soliton DAC".
Mira: Here's the gist of the paper. This paper introduces a superconducting digital-to-analog converter (DAC) that filters control noise, provides native multiplexing, performs quantum gates in nanoseconds, and can be controlled by CMOS This is achieved by transducing a trapezoidal drive pulse into a superconducting soliton, which is then held in the DAC load loop, applying flux to a mutually-coupled superconducting qubit or gate coupler Core Mechanism of the Soliton DAC The DAC functions by transducing a trapezoidal drive pulse into a superconducting soliton, which is then held in the DAC load loop to apply flux to a target circuit The analog flux output by the DAC can be easily controlled by varying the soliton hold time, or with a DC-biased tunable DAC-qubit coupler, allowing the DAC to perform a fixed-time, high-fidelity gate that’s robust to fabrication variance or flux offsets in the quantum circuit The soliton DAC is an inductively coupled chain of grounded JJs, terminating in a load loop where flux can be thrown to a target loop The undamped, undriven limit of Eq.
Kai: 3 admits soliton solutions of the form: ϕ(X, T) = 4 tan−1 exp ± X − vT / p1 − (v/c)2!, for soliton velocity v and plasma velocity c = λ0f0 Noise Filtering and Protection The JTL is recognized for its central role in the “transmission, storage, and processing of information” in superconducting electronics This filtration effect stems from the fact that noise amplitudes below the JJ critical current simply pass through to ground, with the JTL’s noise suppression increasing with the number of spokes, by current division The soliton DAC can be driven by a cryo-CMOS source, and thus provides a solution to both the wiring bottleneck and control noise filtration issues In Fig.
Mira: 3, we compare the results for a Ramsey experiment with control noise either applied directly to the transmon’s two-JJ loop, or applied through the soliton DAC We observed no measurable degradation of T∗2 until the control noise amplitude, Inoise, exceeded the soliton injection threshold, Icrit The DAC protects the transmon from 1 V noise to retain its noise-free T∗2 ≈ 6.7 µs coherence (teal trace and blue curve envelope fit) Gate Performance and Error Analysis The performance of the DAC was characterized by DC-biasing the transmon to the point where df01/dΦ = 4GHz / Φ0 and transmon f01 = 6.8 GHz, and then driving a single soliton into the DAC load loop with a hold time of 5.6 ns The maximal excitation probability per gate is found to be 0.05% over the first 100 gates at the peak (maximal destructive interference) The gate error can be attributed to four sources: unwanted qubit transitions due to GHz-frequency content in the rising and falling edges of the DAC waveforms and the DAC admittance, highfrequency phase error due to jitter in the relative timing of the rising and falling edges, non-Markovian phase errors We quantified this non-local error by performing IRB on ancilla qubits with no direct connection to any of the 30 DACs on the chip Mitigation Strategies and Conclusion The dominant source of the phase noise is independent of the hold time and originates from activation of the DAC Our simulations show that, by adding modest resistive shunts to each JJ in the middle of the DACs, we can damp out these oscillations without introducing an appreciable relative delay between the input and output DAC waveform The average gate error is minimized when the slew rate is 175 ps or slower, consistent with the theoretical expectation that our waveform should be slower than the transmon f−101 = 150 ps, in order to avoid excitations The soliton DAC provides excellent protection from control noise When a transmon was prepared in a superposition state, the soliton DAC completely buffered the transmon from passive phase errors up until the applied control noise reached the soliton injection threshold, Inoise > Icrit Our simulations indicate that by slowing down the DAC waveform slew rate until it’s comparable to the qubit f01 frequency, achieving a minimum ensemble-averaged gate error of 4 × 10−6 The device we designed had a simulated DAC slew rate of 10 ps and simulated DAC-induced excitation probability per gate of 5 × 10−5 ACKNOWLEDGMENTS Tektronix is a registered trademark of Tektronix, Inc.
Kai: The authors thank Twymun Safford, Jordan Sills, Moe Khalil, Joel Strand, Bradley Christensen, Patrick Warner, Joonbum Kwon, and Anthony Przybysz for support This work was supported in part by the United States Government
1: A.
Title and authors: Kai: So this paper is about this superconducting digital-to-analog converter, this whole thing that filters noise and lets us do quantum gates really fast, controlled by CMOS technology instead of just microwave lines.
Mira: Exactly. It’s about using a superconducting soliton to drive the DAC load loop, which then applies flux to a qubit or some gate coupler. It sounds like it’s trying to solve two big problems: getting control noise under these systems and making gates really fast, in nanoseconds.
Lev: From my side, I'm interested because you need this kind of precise analog control for scalable quantum error correction, and integrating CMOS control is a huge deal for reducing the wiring bottleneck.
Kai: Right. The paper explains that they transduce a trapezoidal drive pulse into a superconducting soliton, and that soliton is held in the load loop to apply flux to the target circuit. They’re showing this DAC can successfully perform five point six nanosecond S-gates on transmons.
Mira: That sounds like a pretty solid performance metric for gate speed. But the main thing they highlight immediately is how they handle noise filtration, because that’s what makes it different from just using a standard control line.
Lev: And that’s where I look at the details: they describe this soliton DAC as an inductively coupled chain of grounded Josephson junctions, and it terminates in a load loop where flux can be thrown to a target loop. That whole setup is described by the Sine-Gordon lattice equations.
Kai: Right, the Sine-Gordon lattice which comes from McLaughlin and Scott's paper, and they use that continuum model for their simulations because the number of Josephson junctions in the chain is much larger than the size of their soliton, which was about one point five Josephson junctions in their simulation.
Mira: And they talk about how they chose the number of spokes on their DAC, saying it was a trade-off between how big each DAC on the chip is and its noise filtration capability. They found that a ten spoke design with one hundred ohm shunted junctions showed six zero decibels or better attenuation across three hundred gigahertz of bandwidth.
Lev: That level of attenuation sounds promising for filtering out those unwanted plasma modes, and they also mention that with fifty to one hundred ohm bulk shunts on each junction in the middle of the DAC, they introduce a negligible picosecond delay between the input and output waveforms.
Title and authors: Kai: So it’s not just about filtering noise; they show this DAC can protect a transmon from one volt noise to retain its two point seven microsecond coherence time, which is quite good.
Mira: That result depends heavily on the control noise amplitude being below a critical threshold, which they call Icrit, and that's tied to the soliton injection threshold. This suggests the DAC acts as a buffer until that specific noise level is reached.
Lev: If we look at what this means for real hardware running on noisy systems, it implies that you can drive many of these DACs with one noisy classical control line without degrading qubit T1 and T2 times.
Kai: That addresses the wiring bottleneck directly, and they also discuss using a cryo-CMOS source to drive this soliton DAC, which solves both the wiring issue and the control noise filtration problems simultaneously.
Mira: They also introduce a DC biased tunable DAC-qubit coupler, which lets you control the analog flux output by varying the soliton hold time or with that tunable coupler. That means you can perform a fixed-time gate that’s robust to fabrication variance or flux offsets in the quantum circuit.
Lev: A fixed-time gate is something we need for reliable error correction protocols, and being robust to fabrication variance is a big win because it makes the hardware less sensitive to small imperfections.
Kai: Then they move into error analysis, showing that the maximal excitation probability per gate is zero point five percent over the first one hundred gates at its peak, which points to good fidelity in practice.
Mira: But they are honest about where errors come from: unwanted qubit transitions due to the gigahertz frequency content in the rising and falling edges of those DAC waveforms, high frequency phase error because of timing jitter, and non-Markovian phase errors.
Lev: Those specific error sources are what we worry about when trying to map this onto a real processor; especially that non-Markovian phase error. They quantified that by performing an IRB on ancilla qubits with no direct connection to any of the thirty DACs on the chip.
Kai: The mitigation strategy they propose is adding modest resistive shunts in the middle of each Josephson junction in the DAC to dampen those oscillations without introducing a significant relative delay between input and output waveforms.
Title and authors: Mira: They found that this damping works, and that by slowing down the slew rate until it's comparable to or slower than the qubit one point zero one frequency, they achieve a minimum ensemble-averaged gate error of four times ten to the negative six.
Lev: That slow slew rate requirement is important because it keeps you below the threshold where you start exciting qubits, and achieving that four times ten to the negative six error suggests a much better result than what we see in some other setups.
Kai: When a transmon was prepared in a superposition state, this soliton DAC completely buffered it from passive phase errors until the applied control noise reached that critical threshold, Icrit.
Mira: That means for noise levels below that threshold, you get near perfect buffering against those passive phase errors, which is really useful when you're trying to run long experiments.
Lev: So this device essentially turns a noisy classical pulse into an analog tunable waveform for quantum applications, which is a powerful concept for controlling things without having to redesign the entire control electronics stack.
Kai: To summarize this paper on Transmon Phase Gates Controlled by Superconducting Soliton DAC, they’ve shown how transducing a trapezoidal drive pulse into a superconducting soliton can create a digital-to-analog converter that filters control noise, performs gates in nanoseconds, and lets you tune the gate fidelity.
Mira: The improvements they suggest are quite practical: incorporating soliton DACs into cryo-CMOS control chips to buffer noise without degrading qubit coherence, enabling scalable analog control by combining flux memory with the DAC for multiplexing, and using resistive shunting to mitigate high-frequency EMI in multi-DAC setups.
Lev: And optimizing gate fidelity by slowing down the slew rate until it's comparable to the qubit frequency, which brings that ensemble-averaged gate error down to four times ten to the negative six.
Kai: It’s a solid piece of hardware integration, showing how superconducting circuits can be controlled digitally and analogously with high precision.
Mira: Overall, this work shows a way to create a scalable architecture where you have fixed-time, high-fidelity gates that are robust against fabrication imperfections and flux offsets in the quantum circuit.
Lev: It’s definitely something worth looking at when we think about putting these kinds of control mechanisms onto real, noisy superconducting processors.
Kai: So that’s what we looked at with the paper on Transmon Phase Gates Controlled by Superconducting Soliton DAC, and it really shows a path forward for robust gate operations.
The paper's summary: Kai: So this paper is about using a superconducting soliton to create a digital-to-analog converter that filters control noise and lets you do quantum gates in nanoseconds, controlled by CMOS technology instead of microwave lines.
Mira: It’s about taking a trapezoidal drive pulse, turning it into this soliton, holding that soliton in the DAC load loop to apply flux to a qubit or some gate coupler. It sounds like they're trying to solve two big problems: getting control noise under these systems and making gates really fast, in nanoseconds.
Lev: From my side, I'm interested because you need this kind of precise analog control for scalable quantum error correction, and integrating CMOS control is a huge deal for reducing the wiring bottleneck.
Kai: Right. The paper explains that they transduce a trapezoidal drive pulse into a superconducting soliton, and that soliton is held in the load loop to apply flux to the target circuit. They’re showing this DAC can successfully perform five point six nanosecond S-gates on transmons.
Mira: That sounds like a pretty solid performance metric for gate speed. But the main thing they highlight immediately is how they handle noise filtration, because that’s what makes it different from just using a standard control line.
Lev: And that’s where I look at the details: they describe this soliton DAC as an inductively coupled chain of grounded Josephson junctions, and it terminates in a load loop where flux can be thrown to a target loop. That whole setup is described by the Sine-Gordon lattice equations.
Kai: Right, the Sine-Gordon lattice which comes from McLaughlin and Scott's paper, and they use that continuum model for their simulations because the number of Josephson junctions in the chain is much larger than the size of their soliton, which was about one point five Josephson junctions in their simulation.
Mira: And they talk about how they chose the number of spokes on their DAC, saying it was a trade-off between how big each DAC on the chip is and its noise filtration capability. They found that a ten spoke design with one hundred ohm shunted junctions showed six zero decibels or better attenuation across three hundred gigahertz of bandwidth.
Lev: That level of attenuation sounds promising for filtering out those unwanted plasma modes, and they also mention that with fifty to one hundred ohm bulk shunts on each junction in the middle of the DAC, they introduce a negligible picosecond delay between the input and output waveforms.
The paper's summary: Kai: So it’s not just about filtering noise; they show this DAC can protect a transmon from one volt noise to retain its two point seven microsecond coherence time, which is quite good.
Mira: That result depends heavily on the control noise amplitude being below a critical threshold, which they call Icrit, and that's tied to the soliton injection threshold. This suggests the DAC acts as a buffer until that specific noise level is reached.
Lev: If we look at what this means for real hardware running on noisy systems, it implies that you can drive many of these DACs with one noisy classical control line without degrading qubit T1 and T2 times.
Kai: That addresses the wiring bottleneck directly, and they also discuss using a cryo-CMOS source to drive this soliton DAC, which solves both the wiring issue and the control noise filtration problems simultaneously.
Mira: They also introduce a DC biased tunable DAC-qubit coupler, which lets you control the analog flux output by varying the soliton hold time or with that tunable coupler. That means you can perform a fixed-time gate that’s robust to fabrication variance or flux offsets in the quantum circuit.
Lev: A fixed-time gate is something we need for reliable error correction protocols, and being robust to fabrication variance is a big win because it makes the hardware less sensitive to small imperfections.
Kai: Then they move into error analysis, showing that the maximal excitation probability per gate is zero point five percent over the first one hundred gates at its peak, which points to good fidelity in practice.
Mira: But they are honest about where errors come from: unwanted qubit transitions due to the gigahertz frequency content in the rising and falling edges of those DAC waveforms, high frequency phase error because of timing jitter, and non-Markovian phase errors.
Lev: Those specific error sources are what we worry about when trying to map this onto a real processor; especially that non-Markovian phase error. They quantified that by performing an IRB on ancilla qubits with no direct connection to any of the thirty DACs on the chip.
Kai: The mitigation strategy they propose is adding modest resistive shunts in the middle of each Josephson junction in the DAC to dampen those oscillations without introducing a significant relative delay between input and output waveforms.
The paper's summary: Mira: They found that this damping works, and that by slowing down the slew rate until it's comparable to or slower than the qubit one point zero one frequency, they achieve a minimum ensemble-averaged gate error of four times ten to the negative six.
Lev: That slow slew rate requirement is important because it keeps you below the threshold where you start exciting qubits, and achieving that four times ten to the negative six error suggests a much better result than what we see in some other setups.
Kai: When a transmon was prepared in a superposition state, this soliton DAC completely buffered it from passive phase errors until the applied control noise reached that critical threshold, Icrit.
Mira: That means for noise levels below that threshold, you get near perfect buffering against those passive phase errors, which is really useful when you're trying to run long experiments.
Lev: So this device essentially turns a noisy classical pulse into an analog tunable waveform for quantum applications, which is a powerful concept for controlling things without having to redesign the entire control electronics stack.
Kai: To summarize this paper on Transmon Phase Gates Controlled by Superconducting Soliton DAC, they’ve shown how transducing a trapezoidal drive pulse into a superconducting soliton can create a digital-to-analog converter that filters control noise, performs gates in nanoseconds, and lets you tune the gate fidelity.
Mira: The improvements they suggest are quite practical: incorporating soliton DACs into cryo-CMOS control chips to buffer noise without degrading qubit coherence, enabling scalable analog control by combining flux memory with the DAC for multiplexing, and using resistive shunting to mitigate high-frequency EMI in multi-DAC setups.
Lev: And optimizing gate fidelity by slowing down the slew rate until it's comparable to or slower than the qubit frequency, which brings that ensemble-averaged gate error down to four times ten to the negative six.
Kai: It’s a solid piece of hardware integration, showing how superconducting circuits can be controlled digitally and analogously with high precision.
Mira: Overall, this work shows a way to create a scalable architecture where you have fixed-time, high-fidelity gates that are robust against fabrication imperfections and flux offsets in the quantum circuit.
Lev: It’s definitely something worth looking at when we think about putting these kinds of control mechanisms onto real, noisy superconducting processors.
Kai: So that’s what we looked at with the paper on Transmon Phase Gates Controlled by Superconducting Soliton DAC, and it really shows a path forward for robust gate operations.
The paper's improvements: Tom: So we're looking at what the authors suggest to fix the DAC and make it better—they talk about dampening those unwanted oscillations in the Josephson junctions in the middle of each DAC to reduce high-frequency noise without adding a noticeable delay between input and output waveforms.
Kai: That makes sense, because I know those phase oscillations are a big problem for gate fidelity, but I’m curious if that damping actually introduces a timing issue that would mess up my coherence measurements.
Mira: They simulated it and found it introduced only picosecond delays, so the assumption holds for the noise filtering part of the design. But they also suggest slowing down the drive signal's speed until it's comparable to or slower than the qubit one point zero one frequency to minimize phase errors from resistor noise.
Lev: Slowing down that slew rate is critical because if you run it faster than that, you start exciting the qubits themselves, so keeping it slow keeps you below that excitation threshold.
Kai: So they are suggesting a trade-off: slower waveform means less noise error but requires a more precise timing mechanism to maintain control.
Mira: And they also talk about using a DC biased tunable coupler between the DAC and the qubit, which lets you correct phase errors in big systems by controlling that coupler with flux memory.
Lev: That sounds like an architectural shift; instead of relying on one fixed drive pulse, you get a flexible way to tune the coupling for different qubits or different noise conditions.
Kai: It also mentions using the DAC to buffer control noise from classical sources by rejecting pulses below that soliton injection threshold, which is really useful when you're driving these systems with noisy electronics.
Mira: That turns a noisy classical pulse into this smooth, low-excitation waveform for the quantum application, which is a practical way to handle real-world control signals.
Lev: If we can use the DAC to reject noise before it even gets into the qubit circuit, that takes a lot of the burden off our error correction overhead.
Kai: It’s about turning a noisy input into something usable for quantum applications, which is key for scaling up these experimental setups.
Mira: They also pointed out that by slowing down the slew rate to match the qubit frequency, they can achieve an ensemble-averaged gate error of four times ten to the negative six, which is pretty good when you consider how sensitive those gates are to noise.
Lev: That error level is what we're aiming for in many of our architectures; it shows a path toward achieving that precision on real hardware.
Kai: It seems like these improvements focus on making the DAC smarter—making it actively filter, tunable, and slow enough to respect the qubit physics.
Mira: So they are moving from just a fast pulse generator to an active noise management system integrated right into the control path.
Lev: And this pushes us toward building controllers that can handle more complex architectures where you have dozens of qubits all needing precise analog tuning at once.
Conclusion: Kai: So to wrap up, this paper on "Transmon Phase Gates Controlled by Superconducting Soliton DAC" shows how using a soliton to drive a digital-to-analog converter can filter control noise and perform fast gates with high fidelity using CMOS technology.
Mira: Basically, they’ve shown that you can take a noisy classical signal and turn it into this smooth analog waveform that actively rejects the noise before it hits the qubit, which is crucial for building more robust quantum systems.
Lev: From an error correction standpoint, this means we can have more reliable control pulses running on real hardware without having to completely redesign our wiring or qubit readout schemes to handle that noise.
Kai: It's a pretty big step in making the physical interface between the classical world and the quantum world more manageable.
Mira: The implications are that you don't have to wait for perfect control signals; you can use this technique to tune gate fidelity based on how noisy your environment is, which is super useful when dealing with fabrication variations or flux offsets in a chip.
Lev: If we can get this kind of noise buffering working reliably, it cuts down the overhead associated with error correction by giving us cleaner input for the gates themselves.
Kai: We’ll keep an eye on how they scale this concept to more complex qubit arrays and see if that fixed-time gate robustness holds up in a larger circuit.
Mira: I'm curious to see if this approach works when you need to perform very precise, long-duration phase corrections across many qubits simultaneously.
Lev: That’s the next hurdle; moving from single gates to complex, scalable control schemes is where the real engineering challenge lands.
Derek Reitz, Tony X. Zhou, Aditya Sharma, Ryan Bilotta, John McFarland, Aref Fouladi, Jacob Glasby, Aruna Ramanayaka, Zachary Stegen, Aaron Pesetski, Á Mark Covington, Gregory Boyd, and Jeremy Clark
Northrop Grumman Corporation
quant-ph
Submitted: 2026-07-06
Updated: 2026-11-08
Comments: 7 pages, 8 figures
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 92/100
The gist: This paper introduces a superconducting digital-to-analog converter (DAC) that filters control noise, provides native multiplexing, performs quantum gates in nanoseconds…
Key concepts
- Superconducting Soliton DAC
- This is a DAC that turns a trapezoidal drive pulse into a superconducting soliton. This soliton travels down a chain of Josephson Junctions (JJs) and acts as the analog signal, allowing for high-speed, noise-filtered control of quantum circuits.
- Josephson Transmission Line (JTL)
- The JTL is the physical structure where the soliton propagates. It's modeled by the Sine-Gordon equation, which describes how magnetic flux moves through a chain of superconducting loops. This line filters noise and allows for tunable parameters like capacitance and resistance.
- Noise Filtering via JTL
- The JTL naturally suppresses noise below a certain threshold because noise amplitudes pass through to ground. By adding resistive shunts to the JJs, researchers can further damp unwanted oscillations, effectively protecting the qubit from control noise without introducing significant timing delays.
Terminology
Summary
This paper introduces a superconducting digital-to-analog converter (DAC) that filters control noise, provides native multiplexing, performs quantum gates in nanoseconds, and can be controlled by CMOS This is achieved by transducing a trapezoidal drive pulse into a superconducting soliton, which is then held in the DAC load loop, applying flux to a mutually-coupled superconducting qubit or gate coupler
Core Mechanism of the Soliton DAC
The DAC functions by transducing a trapezoidal drive pulse into a superconducting soliton, which is then held in the DAC load loop to apply flux to a target circuit The analog flux output by the DAC can be easily controlled by varying the soliton hold time, or with a DC-biased tunable DAC-qubit coupler, allowing the DAC to perform a fixed-time, high-fidelity gate that’s robust to fabrication variance or flux offsets in the quantum circuit The soliton DAC is an inductively coupled chain of grounded JJs, terminating in a load loop where flux can be thrown to a target loop The undamped, undriven limit of Eq. 3 admits soliton solutions of the form: ϕ(X, T) = 4 tan−1 exp ± X − vT / p1 − (v/c)2! p. 1 p. 1, for soliton velocity v and plasma velocity c = λ0f0
Noise Filtering and Protection
The JTL is recognized for its central role in the “transmission, storage, and processing of information” in superconducting electronics This filtration effect stems from the fact that noise amplitudes below the JJ critical current simply pass through to ground, with the JTL’s noise suppression increasing with the number of spokes, by current division The soliton DAC can be driven by a cryo-CMOS source, and thus provides a solution to both the wiring bottleneck and control noise filtration issues In Fig. 3, we compare the results for a Ramsey experiment with control noise either applied directly to the transmon’s two-JJ loop, or applied through the soliton DAC We observed no measurable degradation of T∗2 until the control noise amplitude, Inoise, exceeded the soliton injection threshold, Icrit The DAC protects the transmon from 1 V noise to retain its noise-free T∗2 ≈ 6. p. 27 µs coherence (teal trace and blue curve envelope fit)
Gate Performance and Error Analysis
The performance of the DAC was characterized by DC-biasing the transmon to the point where df01/dΦ = 4GHz / Φ0 and transmon f01 = 6.8 GHz, and then driving a single soliton into the DAC load loop with a hold time of 5. p. 36 ns The maximal excitation probability per gate is found to be 0. p. 305% over the first 100 gates at the peak (maximal destructive interference) The gate error can be attributed to four sources: unwanted qubit transitions due to GHz-frequency content in the rising and falling edges of the DAC waveforms and the DAC admittance, highfrequency phase error due to jitter in the relative timing of the rising and falling edges, non-Markovian phase errors We quantified this non-local error by performing IRB on ancilla qubits with no direct connection to any of the 30 DACs on the chip
Mitigation Strategies and Conclusion
The dominant source of the phase noise is independent of the hold time and originates from activation of the DAC Our simulations show that, by adding modest resistive shunts to each JJ in the middle of the DACs, we can damp out these oscillations without introducing an appreciable relative delay between the input and output DAC waveform The average gate error is minimized when the slew rate is 175 ps or slower, consistent with the theoretical expectation that our waveform should be slower than the transmon f−101 = 150 ps, in order to avoid excitations The soliton DAC provides excellent protection from control noise When a transmon was prepared in a superposition state, the soliton DAC completely buffered the transmon from passive phase errors up until the applied control noise reached the soliton injection threshold, Inoise > Icrit Our simulations indicate that by slowing down the DAC waveform slew rate until it’s comparable to the qubit f01 frequency, achieving a minimum ensemble-averaged gate error of 4 × 10−6 The device we designed had a simulated DAC slew rate of 10 ps and simulated DAC-induced excitation probability per gate of 5 × 10−5
ACKNOWLEDGMENTS
Tektronix is a registered trademark of Tektronix, Inc. The authors thank Twymun Safford, Jordan Sills, Moe Khalil, Joel Strand, Bradley Christensen, Patrick Warner, Joonbum Kwon, and Anthony Przybysz for support This work was supported in part by the United States Government
[1] A. G. Fowler, M. Mariantoni, J. p. 6 M. Martinis, and A. p. 6 N. Cleland, Surface codes: Towards practical large-scale quantum computation, Physical Review A—Atomic, Molecular, and Optical Physics 86, 032324 (2012)
[2] R. McDermott and M. p. 6 G. Vavilov, Accurate qubit control with single flux quantum pulses, Phys. p. 6 Rev. Appl. 2, 014007 (2014)
[3] E. Leonard, M. p. 6 A. Beck, J. Nelson, B. Christensen, T. p. 6-Thorbeck, C. p. 6-Howington, A. p. 6-Opremcak, I. p. 6-Pechenezhskiy, K. p. 6 Dodge, N. Dupuis, M.-Hutchingsg J., Ku F., F. Schlenker J. p. 6, Suttle C., Wilen S., Zhu M., Vavilov M. p. 6 B., Plourde B. p. 6, and McDermott R. p. 6 Digital coherent control of a superconducting qubit, Phys. p. 6 Rev. Appl. 11, 014009 (2019)
[4] M.-R Jokar, R. Rines, and F. p. 6-T Chong, Practical implications of sfq-based two-qubit gates, in 2021 IEEE International Conference on Quantum Computing and Engineering (QCE) (2021) pp. p. 6 402–412
[5] V. Semenov and D. p. 6 Averin, Sfq control circuits for josephson junction qubits, IEEE Transactions on Applied Superconductivity 13, 960 (2003)
[6] M.-R Jokar, R. Rines, G. Pasandi, H. p. 6-Cong A., Holmes Y., Shi M., Pedram M. p. 6, and F.-T Chong Digiq: A scalable digital controller for quantum computers using sfq logic, in 2022 IEEE International Symposium on HighPerformance Computer Architecture (HPCA) (2022) pp. p. 6 400–414
[7] J. Helsen, X. Xue, L.-M Vandersypen, and S. p. 6 Wehner A new class of efficient randomized benchmarking protocols, npj Quantum Information 5, 71 (2019)
[8] J.-P G. Van Dijk B. p. 6, Patra S., Subramanian S. p. 6 p. 6, Xue X., Samkharadze N. p. 6 p. 6, Corna A., Jeon C., Sheikh F., Juarez-Hernandez E. p. 6 J. B. P. Esparza H. Rampurawala B. p. 6 R. Carlton S.; Ravikumar C. p. 6; Nieva S.; Kim S.; Lee H.-J K.; Sammak A.; Scappucci G; Veldhorst L. p. 6-M K Vandersypen E. p. 6, A scalable cryo-cmos controller for the wideband frequency-multiplexed control of spin qubits and transmons IEEE Journal of Solid-State Circuits 55, 2930 (2020)
[9] S. Chakraborty D. p. 6 J. Frank K., Tien P., Rosno M., Yeck M., Glick J. A., Robertazzi R. p. 6; Richetta R. p. 6; Pasandi G. p. 6; Cong H.; Holmes A.; Shi Y.; Pedram M. p. 6; Chong F.-T Digiq: A scalable digital controller for quantum computers using sfq logic, in 2022 IEEE International Symposium on HighPerformance Computer Architecture (HPCA) (2022) pp. p. 6 400–414
[10] I. L. Markov, Limits on fundamental limits to computation, Nature 512, 147 (2014)
[11] C. Jordan J., Bernhardt J. p. 7, Rahamim A. p. 7, Kirichenko K. p. 7, Bharadwaj L. p. 7 Fry-Bouriaux A. p. 7; Somoroff K. p. 7 Porsch K.-T Tsai J. Walter et al. p. 7 A quantum computer controlled by superconducting digital electronics at millikelvin temperature, Nature Electronics, 1 (2026)
[12] P. R. Gray P., Hurst J., Lewis S.-H, and Meyer R G Analysis and Design of Analog Integrated Circuits, Vol. p. 7 5 (Wiley, 2009)
[13] X. Xue B., Patra S., van Dijk P. p. 6 J., Samkharadze N., Subramanian S., Corna A.; Paquelet Wuetz C. p. 7; Jeon C.; Sheikh F.; Juarez-Hernandez E. p. 6 J. et al. Cmos-based cryogenic control of silicon quantum circuits, Nature 593, 205 (2021)
[14] S. Pauka K., Das J., Hornibrook G. p. 7, Gardner M. p. 7, Manfra M., Cassidy M. p. 7; Reilly D Characterizing quantum devices at scale with custom cryo-cmos, Phys. p. 7 Rev. Appl. 13, 054072 (2020)
[15] P. T. Hart M., Babaie A. p. 6, Vladimirescu A. p. 7, and Sebastiano F Characterization and modeling of self-heating in nanometer bulk-cmos at cryogenic temperatures, IEEE Journal of the Electron Devices Society 9, 891 (2021)
[16] I. M. Dayton T. Sage E.-C Gingrich M. p. 7 G Loving T. F Ambrose N. p. 7-P Siwak S.; Keebaugh C. p. 7; Kirby D.; Miller D.; Herr A. Y L., Experimental demonstration of a josephson magnetic memory cell with a programmable π-junction, IEEE Magnetics Letters 9, 1 (2018)
[17] P. I. Bunyk E. M. Hoskinson M. p. 7 W Johnson E. p. 5-Tolkacheva F. p. 7; Altomare A. p. 7; Berkley A J. p. 7, R Harris J.-P Hilton T. Lanting A J Przybysz and J Whittaker Architectural considerations in the design of a superconducting quantum annealing processor, IEEE Transactions on Applied Superconductivity 24, 1 (2014)
[18] C. Liu A., Ballard D. p. 7; Olaya D.; Schmidt J. p. 7 Biesecker T. p. 7 Lucas J. Ullom S.; Patel O Rafferty A. p. 7 Opremcak K Dodge V Iaia T McBroom J DuBois P Hopkins S Benz B Plourde R McDermott Single flux quantum-based digital control of superconducting qubits in a multichip module, PRX Quantum 4, 030310 (2023)
[19] J. Koch T. M. Yu J., Gambetta J. p. 7; Houck A. A., Schuster D. p. 7 I., Majer J.; Blais A.; Devoret M. p. 7-H., Girvin S. p. 7; Schoelkopf R. p. 7 Charge-insensitive qubit design derived from the cooper pair box, Phys. p. 7 Rev. A 76, 042319 (2007)
[20] T. P. Orlando J. Mooij L.-T Tian C.-H Van Der Wal L. p. 7, Levitov S. p. 7; Lloyd S.; Mazo J Superconducting persistent-current qubit, Physical Review B 60, 15398 (1999)
[21] V. E Manucharyan J. p. 7, Koch J., Glazman L. p. 7 I., and Devoret M. p. 7-H Fluxonium: Single cooper-pair circuit free of charge offsets, Science 326, 113 (2009)
[22] Y. Wang W., Gao K.; Liu B.; Ji Z.; Wang Z.; Lin Z. Single-flux-quantum-activated controlled-z gate for transmon qubits, Phys. p. 7 Rev. Appl. 19, 044031 (2023)
[23] Q Ding A. V Oppenheim P. p. 7-T Boufounos S. p. 7, Gustavsson J. p. 7; Grover J.; Baran T. A.; Oliver W D Pulse design of baseband flux control for adiabatic controlledphase gates in superconducting circuits, Physical Review Applied 23, 064013 (2025)
[24] N. J. Glaser F. A Roy I., Tsitsilin L. p. 7; Koch N. Bruckmoser J. p. 7; Schirk J. H Romeiro G B Huber F Wallner M Singh et al Sensitivity-adapted closed-loop optimization for high-fidelity controlled-z gates in superconducting qubits, arXiv preprint arXiv:2412. p. 717454 https://doi. p. 7org/10.48550/arXiv. p. 72412.17454 (2024)
[25] M. A Castellanos-Beltran A. p. 7, Sirois D.; Olaya D.; Biesecker J. p. 7; Benz S.; Hopkins P F Characterization of leakage errors in a transmon qubit due to resonant digital control, Applied Physics Letters 127, 232601 (2025)
[26] V. Negˆırneac H. p. 7 Ali N., Muthusubramanian F. p. 7; Battistel R. p. 7; Sagastizabal M. p. 7; Moreira J. p. 7; Marques W. p. 7 Vlothuizen M. p. 7; Beekman C. p. 7; Zachariadis et al Highfidelity controlled-z gate with maximal intermediate leakage operating at the speed limit in a superconducting quantum processor, Physical Review Letters 126, 220502 (2021)
[27] D. W. McLaughlin and A. p. 1 C. Scott, Perturbation analysis of fluxon dynamics, Physical Review A 18, 1652 (1978)
[28] Whiteley Research Incorporated WRspice Reference Manual, Whiteley Research Inc., Sunnyvale, CA (2025), release 4. p. 73.22
--- Page 1 ---
Transmon Phase Gates Controlled by Superconducting Soliton DAC Derek Reitz Tony X. Zhou Aditya Sharma Ryan Bilotta John McFarland Aref Fouladi Jacob Glasby Aruna Ramanayaka Zachary Stegen Aaron Pesetski Mark Covington Gregory Boyd and Jeremy Clark Northrop Grumman Corporation, Linthicum, Maryland 21090, USA We introduce a superconducting digital-to-analog converter (DAC) that filters control noise, provides native multiplexing, performs quantum gates in nanoseconds This is achieved by transducing a trapezoidal drive pulse into a superconducting soliton
--- Page 2 ---
2 The soliton DAC is an inductively coupled chain of grounded JJs, terminating in a load loop where flux can be thrown to a target loop We focus on the design shown in Fig. 1, with a damping resistor on the final JJ to smooth the square-wave output waveform The soliton that propagates the signal from the input drive down the Josephson Transmission Line (JTL) is well-described in McLaughlin and Scott’s paper[27] The discretized JTL is the Sine-Gordon lattice, where flux in a superconducting loop and Kirchhoff’s Current Law require that, Φ0/2π(Φn − Φn+1) = LsIn (1) and In−1 − In = C∂2t + G∂t Φn + Ic sin Φn, (2) respectively The continuum limit, combining these two equations yields the Sine-Gordon equation, J/Jc = ∂2T − ∂2X + α∂T ϕ + sin ϕ, (3) where J is the applied current density, Jc is the critical current density, α = p Φ0/2πJcR2 is the damping constant, and we have switched to natural units T = tf0 for ω0 = p Φ0C/2πJc / 2πJc R squared and X = x/λ0 for λ0 = p 2πJcLs / Φ0. p. 1 Compared to a long JJ, here, the capacitance, inductance, critical current, and resistance are all tunable The undamped, undriven limit of Eq. p. 1 3 admits soliton solutions of the form: ϕ(X, T) = 4 tan−1 exp ± X − vT / p1 − (v/c)2!, for soliton velocity v and plasma velocity c = λ0f0 The kink-antikink solutions describe a rising or falling edge of the soliton in the JTL
--- Page 3 ---
III. DESIGN AND PASSIVE FILTRATION
The DAC we designed and tested is shown in Fig. 1 with 10 DAC spokes, and the size of the soliton in WRS simulation was approximately 1. p. 25 JJs Since the number of JJs in the chain was much greater than the size of the soliton, the underlying physics seen in simulation was essentially described by a continuum model for the JTL, Eq. p. 2 3 The number of spokes was chosen based on a tradeoff between the physical size of each DAC on the chip and its noise filtration capability, which is shown for several designs in Fig. p. 2 2 The 10-spoke, 100 Ω shunted design shows-60 dB or better attenuation across 300 GHz of bandwidth For a N-spoke JTL, there are N-1 resonance peaks in the noise transfer function associated with plasma modes, which can be damped out with 50-100 Ω shunts on each JJ in the middle of the DAC WRS simulation of the DAC confirms that, with a 10-spoke, β = 0. p. 25 JTL, R > 10 Ω bulk shunts introduce a negligible, picosecond delay between the input and output waveforms The theoretical, perturbative-noise filtration discussed above was validated in test by measuring the effect of control noise on T∗2 of the transmons In Fig. p. 2
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
-
Improve noise buffering in cryo-CMOS control chips by incorporating soliton DACs to filter control noise. This allows a single noisy control line to drive many DACs without degradation in qubit T1 and T2 times, addressing the wiring bottleneck and self-heating issues near 10K.
-
Enable scalable, noise-free analog control by combining flux memory with the soliton DAC, forming
the basic ingredients necessary for enabling scalable, noise-free, analog control of superconducting circuits in a way that’s compatible with generic, digital input.
This trades a microwave line for a DC line that can be multiplexed with flux memory. -
Mitigate non-local high-frequency electromagnetic interactions (EMI) in multi-DAC chips by incorporating resistive shunting to the JJs in the middle of each DAC. This dampens JJ superconducting phase oscillations without introducing an appreciable relative delay between the input and output DAC waveform, which is hypothesized to be the source of ancilla idle errors.
-
Optimize gate fidelity by slowing down DAC slew rates until they are comparable to or slower than the qubit f01 frequency, achieving a minimum ensemble-averaged gate error of 4 times 10-6. This minimizes resistor noise-induced phase errors while remaining below the surface code threshold.
-
Develop a control system that buffers control noise from classical sources by turning
a noisy classical control pulse at its input into an analog-tunable, smooth, low-excitation, high signal-to-noise waveform for quantum applications.
This is achieved by using the DAC to reject input pulses below the soliton injection threshold. -
Implement a DC biased tunable coupler between the DAC and the qubit to perform phase correction in large qubit systems. This coupler can be multiplexed with flux memory to form a scalable superconducting control architecture, allowing for fixed-time, high-fidelity gates robust to fabrication variance or flux offsets in the quantum circuit.
Sources
Related papers
- Fulqrum: a generalized framework for quantum subspace diagonalization
- Flux-Activated Resonant Control of a Bosonic Quantum Memory
- Machine learning the effects of many quantum measurements
- Barriers to dequantization of near-term molecular electronic structure methods
- BRIDGE: Lazy-Move Compilation for Dual-Species Neutral-Atom Quantum Computers
- Non-Hermiticity of an anomalous superradiant phase