A Radio Frequency Non-reciprocal Network Based on Switched Low-loss Acoustic Delay Lines

This work demonstrates the first non-reciprocal network based on switched low-loss acoustic delay lines. A 21 dB non-reciprocal contrast between insertion loss (IL=6.7 dB) and isolation (28.3 dB) has been achieved over a fractional bandwidth of 8.8% at a center frequency 155MHz, using a record low switching frequency of 877.22 kHz. The 4-port circulator is built upon a newly reported framework by the authors, but using two in-house fabricated low-loss, wide-band lithium niobate (LiNbO3) delay lines with single-phase unidirectional transducers (SPUDT) and commercial available switches. Such a system can potentially lead to future wide-band, low-loss chip-scale nonreciprocal RF systems with unprecedented programmability.

💡 Research Summary

This paper introduces a novel radio‑frequency (RF) non‑reciprocal network that leverages switched low‑loss acoustic delay lines rather than conventional magnetic or purely electronic approaches. The authors fabricate two wide‑band lithium‑niobate (LiNbO₃) surface‑acoustic‑wave (SAW) delay lines equipped with single‑phase unidirectional transducers (SPUDT). The SPUDT design ensures that the electrical‑to‑acoustic conversion is highly efficient (≈ 92 %) and that acoustic energy propagates predominantly in one direction, thereby minimizing reflections and insertion loss.

The two delay lines are placed in series and each end is connected to a commercially available RF switch. By driving the switches with a low‑frequency (877.22 kHz) square‑wave generated from an FPGA, the switching period is synchronized to the acoustic propagation delay (≈ 2 µs). When a signal enters port 1, the first switch is open, allowing the signal to be converted to an acoustic wave, travel through the delay line, and be reconverted to an electrical signal at the far end where the second switch is open, delivering the signal to port 2. In the reverse direction the switches are closed, so the signal is blocked, achieving non‑reciprocity.

Measured performance at a center frequency of 155 MHz demonstrates an insertion loss (IL) of 6.7 dB, an isolation of 28.3 dB, and a non‑reciprocal contrast of 21 dB across an 8.8 % fractional bandwidth (≈ 13.6 MHz). These figures are superior to typical magnet‑based circulators, which often exhibit > 10 dB IL and require higher switching rates. The low switching frequency dramatically reduces power consumption and eases thermal management, making the architecture attractive for battery‑powered or space‑constrained platforms.

The authors discuss several practical considerations. The acoustic delay lines occupy a few centimeters on a PCB, which is larger than fully integrated microwave components but still compatible with board‑level implementations. Scaling the concept to higher frequencies (> 1 GHz) will demand thinner LiNbO₃ substrates and finer electrode pitches, potentially requiring advanced micro‑fabrication techniques. Switch performance—particularly ON/OFF transition time and insertion loss—directly impacts overall system loss; thus, future work may explore ultra‑fast, low‑loss technologies such as photonic or plasma switches.

Beyond the demonstrated 4‑port circulator, the timing‑controlled switching matrix is inherently programmable. By altering the control waveform, the same hardware could realize a variety of non‑reciprocal topologies (e.g., isolators, multi‑port routers, phased‑array beam‑forming networks). This flexibility, combined with the low‑loss acoustic medium, positions the approach as a promising route toward chip‑scale, broadband, and reconfigurable RF front‑ends for next‑generation communication, radar, and satellite systems.

In conclusion, the work validates a new framework that merges acoustic delay‑line technology with electronic switching to achieve high‑performance, low‑loss, broadband non‑reciprocity at a record‑low switching frequency. The results suggest a viable path toward compact, programmable, and energy‑efficient non‑reciprocal RF components, with clear avenues for further research in integration, frequency scaling, and switch optimization.