AWR2243: Is there any suggested number for LO phase matching in the cascade design

Hi Guangfu, 

There are a few issues that will come up with increased LO mismatch on the PCB. 

• Range correlation benefit decrease
• Phase-mismatch between beam-formed TX elements devices
• Phase-mismatch between RX elements across devices

Phase mismatch between TX and RX elements will need to be dealt with in system-level TX and RX path calibrations. A consistent phase offset on the LO due to PCB mismatch might be able to be calibrated out, but due to the potential non-linear phase effects across temperature on PCB substrates, it's best to just limit the effect vs. living with it. The result would be drop in SNR in beam-formed scenarios and angular resolution loss in MIMO scenarios. The range correlation benefit effect cannot be calibrated out. 

LO mismatch from the PCB will result in a proportional decrease in the range correlation benefit from the single VCO being used as the TX and mixer input across the AWR devices. See the references below for more info on this effect. 

To summarize though, in an FMCW radar, like the AWR devices, the LO and RX paths have correlated phase noise due to the same VCO generating both signals. This correlation is a function of delay between the two signals. With increased delay, this correlation drops off. This decrease in correlation happens with longer range targets just due to the round-trip propagation delay. Unless well managed in layout, the delay mismatch on the LO PCB paths, between devices, can also be on the same order as the normal delays due to the RF propagation, further increasing this effect.

This effect is most noted in "bumper reflections" in automotive radar context since the phase noise skirt around larger, closer targets (like vehicle bumper radomes) may mask smaller targets near the same range unless phase noise is sufficiently small to provide adequate SNR for those close, smaller, targets to peak out. Our AWR devices provided excellent phase noise quality, and largely mitigate this problem. 

Without some additional experimentation on a specific AWR system I don't know if it is possible to quantify the specific increase in quick terms of dB/deg of phase mismatch, but the effect is there. That is why in that quoted e2e post I simply stated that PCB designers should match these lines to whatever minimal tolerances can be achieved. Our method of dealing with the potential problem on the MMWCAS-RF-EVM was to completely minimize its effects by length matching the LO lines to within 1 mil. 

You can experiment with this phase correlation and mismatch effect on the MMWCAS-RF-EVM by looking at target scenes in cascaded vs. single-device mode, comparing AWR #1 and AWR #3 on the same cross-device virtual channels between the two modes. AWR #1 and AWR #3 both include their own 40 MHz crystal source and can both be run in single-device mode. That may be one way to figure out if it is a problem for your use-case. 

References: 

1. A. D. Droitcour, O. Boric-Lubecke, V. M. Lubecke, J. Lin and G. T. A. Kovacs, "Range correlation and I/Q performance benefits in single-chip silicon Doppler radars for noncontact cardiopulmonary monitoring," in IEEE Transactions on Microwave Theory and Techniques, vol. 52, no. 3, pp. 838-848, March 2004, doi: 10.1109/TMTT.2004.823552.

2. M. C. Budge and M. P. Burt, "Range correlation effects on phase and amplitude noise," Proceedings of Southeastcon '93, Charlotte, NC, USA, 1993, pp. 5 p.-, doi: 10.1109/SECON.1993.465731.

Thank you,

-Randy