Click here to go to our main page on phase shifters
Click here to go to our main page on time delay units
Click here to learn fundamentals of reflection phase shifters
New for May 2024. This concept was shared with us by Bal Govind, the lead author of a paper that appeared in Nature Magazine in March 2024, located here. Yes there is a paywall, but we obtained a review copy of the article for fair use here. You can download a free copy of the paper on ResearchGate if you don't mind giving up personal information (click here). Read about ResearchGate on Wikipedia before you click that link, never expect a free lunch from the internet and you won't be dissappointed.
The phase shifter circuit is realized in 45 nm CMOS technology and is a reflective structure: a compact 3D 3-dB hybrid coupler structure used to bounce signals off of two identical passive terminations. Varying the reflection phase of the terminations produces a signal that has characteristics of true time delay in coarse steps and as a phase shifter in smaller phase steps. The image below shows the concept. Figure (b) and (c) shows shunt segments along a series line that can be switched in and out using N-FETs to provide course delay steps. Opening up the segments closest to the input in sequence adds additional phase shift, with each segment adding a nearly fixed time delay and phase shift of 20 degrees at band center (17 GHz). In past efforts the shunt segments were often just short circuits, and here is where this new circuit differs. Looking at (d) and (e) you will notice that the shunt segments in the Q-TTD approach have some additional features: a shunt inductor terminating the line, with seven different inductance values switched in by more NFETs, which ideally would result in fixed phase shift across frequency in steps of 2-3 degrees. Finally, ultra-fine phase resolution is provided by changing voltages across varactors inside the shunts to vary their capacitance. There are a lot of voltages to keep track of, but thankfully, CMOS is a great media for controlling such a circuit.
The resulting device operates from 10 to 24 GHz with average loss of 8 dB. The phase shift is nominally 180 degrees at center band and increases with frequency, but not as much as if it was an ideal true-time delay circuit (which we thought explained the "quasi" modifier in the title). According to a follow up with the author, it was named for a different reason: "theoretically time delay would have infinite bandwidth while this has about 12 GHz." The bandwdth, of course, is limited my the 3dB hybrid couople, not the phase adjusting terminations.
By combining the course, fine and ultrafine phase adjustments, phase steps of 1 degree are provided. Output P1dB is a healthy 8.5 to 9.5 dBm, and return loss is excellent at better than 10 dB.
Thanks to Bal Govind for sharing this idea! We followed up with him on a couple of thoughts, starting with, don't you need 360 degrees of phase shift in a phased array? Beyond the idea of cascading two of the described 180 degree phase shifters, Bal offered a method for a adding a simple, 180-degree single bit using a balun-based phase inverter, you can read about it in reference 1 below. We won't review the phase inverter idea here on Microwaves101, unless someone requests more info.
References
1. Govind, B., Tapen, T. & Apsel, A. "Ultra-compact quasi-true time delay for boosting wireless channel capacity", Nature 627, 88–94 (2024).
2. R. B. Yishay and D. Elad, "D-Band 360° Phase Shifter with Uniform Insertion Loss," 2018 IEEE/MTT-S International Microwave Symposium - IMS, Philadelphia, PA, USA, 2018, pp. 868-870.