← 研踪 PaperTrailarXiv:2611.23448 · v1
Compute Before You Digitize: ADC-less Mixed-Signal Processing for Scalable Cryogenic Qubit Readout
首次提交 2026-11-18 03:52
版本更新 2026-11-18 03:52
雷达观测 2026-11-19 19:08
摘要
We introduce DIRAC, an ADC-less in-cryostat qubit readout architecture that uses analog computing to redraw the analog-digital boundary. Directly operating on the front-end's analog output, passive complex mixers perform the equivalent of digital demodulation, ping-pong active-RC integrators average their outputs into down-sampled per-qubit traces, and a time-domain stage digitizes them, eliminating Nyquist-rate ADCs. A TX-RX co-design derives the mixers' local oscillators from the readout transmitter's baseband rather than per-tone PLLs. A per-line calibration flow trains the matched filter jointly with the neural network, improving fidelity and enabling low-bit-width, low-voltage digital processing. Designed in 22 nm FD-SOI and evaluated at 4 K on 1.12M experimental shots, DIRAC matches the fidelity of a floating-point (FP32) digital pipeline for readout windows down to 750 ns, at 31.6 and 238 lower energy than the fixed-point (FxP) and FP32 pipelines. It cuts the total 4 K power per physical qubit by 50.4% and 68.6% relative to FxP and room-temperature flex cables, respectively. At a fixed logical-qubit count and 4 K budget, this yields, against FxP, a code distance larger by 2 to 6 and a 1.6-7.3 (Willow) and 4.9-204 (SI1000, ) lower logical error rate.
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-18 03:52 · 最新观测版本
Compute Before You Digitize: ADC-less Mixed-Signal Processing for Scalable Cryogenic Qubit Readout
We introduce DIRAC, an ADC-less in-cryostat qubit readout architecture that uses analog computing to redraw the analog-digital boundary. Directly operating on the front-end's analog output, passive complex mixers perform the equivalent of digital demodulation, ping-pong active-RC integrators average their outputs into down-sampled per-qubit traces, and a time-domain stage digitizes them, eliminating Nyquist-rate ADCs. A TX-RX co-design derives the mixers' local oscillators from the readout transmitter's baseband rather than per-tone PLLs. A per-line calibration flow trains the matched filter jointly with the neural network, improving fidelity and enabling low-bit-width, low-voltage digital processing. Designed in 22 nm FD-SOI and evaluated at 4 K on 1.12M experimental shots, DIRAC matches the fidelity of a floating-point (FP32) digital pipeline for readout windows down to 750 ns, at 31.6 and 238 lower energy than the fixed-point (FxP) and FP32 pipelines. It cuts the total 4 K power per physical qubit by 50.4% and 68.6% relative to FxP and room-temperature flex cables, respectively. At a fixed logical-qubit count and 4 K budget, this yields, against FxP, a code distance larger by 2 to 6 and a 1.6-7.3 (Willow) and 4.9-204 (SI1000, ) lower logical error rate.
查看此版本解读 →查看此版本原文 ↗