Unexpected coupling through QM12112 RF switch at 6.5 GHz UWB

RF Solutions RF All Other (switches, controls, attenuators, ..)
1

I’m working on a UWB PDoA/TWR board based on the Murata Type 2AB (which uses the QM33120) operating at 6.5 GHz. Two 2-element phased arrays on opposite sides of the PCB are selected using QM12112 RF switches.

While investigating antenna selection performance, I physically disconnected one array from the PCB by removing a series matching component. Surprisingly, when the RF switch is set to the disconnected array, the system can still receive signals and perform ranging.

When the connected array faces the transmitter, RSSI is only 3-8 dB lower than when that array is actively selected. When the disconnected array faces the transmitter, RSSI drops by approximately 15 dB, suggesting the signal is actually being received by the connected array and somehow coupling into the selected receive path. This causes PDoA performance to be very unreliable and inaccurate, but the ranging performance seems to be unaffected.

The QM12112 datasheet specifies approximately 15-18 dB isolation at 6.5 GHz, so I would expect RSSI degradation closer to that value if switch leakage were the dominant mechanism. The observed 3-8 dB reduction suggests there may be another coupling path involved, such as PCB coupling, antenna-to-antenna coupling, ground plane coupling, feed network leakage, or a switch integration/layout issue.

Has anyone seen similar behaviour in switched antenna systems? Does this point to a likely RF layout mistake, a misunderstanding of how switch isolation translates into system-level RSSI, or simply insufficient switch isolation for this application?

Please find attached an image of the PCB design around the RF switch. Thank you in advance!

{42F3DCCF-914C-4484-93B1-5185B7AD4425}
{42F3DCCF-914C-4484-93B1-5185B7AD4425}875×849 90.9 KB

2

On the layout then the lack of ground vias directly under the switch could be causing issues and degrading isolation. I’d recommend following Qorvo’s evaluation board layout. There is also a higher isolation switch available that was designed for UWB applications and would likely give better performance in your application, QM12202. Contact appsupport@qorvo.com stating your company name and location if you need the evaluation board gerber files, or info on the QM12202.

3

Hi Chris, thanks for the reply. I have a few follow up questions:

  1. Could you elaborate more on how a ground via should be placed under the switch? The switch pads are only 0.2x0.2mm in size, which is much smaller than even a reasonably sized via hole (0.3 mm).
  2. Could you provide more information on the QM12202? I cannot seem to be able to find any information on this product at all. It is not present on Qorvo’s RF switch catalog, and no datasheets seem to be publicly available.

Thank you so much for your help?

4

Below is a screenshot of the area around the QM12112 on the eval board. There are five 6mil (0.152mm) vias in the ground area under the switch, plus a lot of 10mil (0.254mm) vias around the IC. The vias stitching along the RF lines are also much closer spaced than on your design. Your design also has thicker 50R lines, which is slightly problematic as the QM12112 pins are narrow. You may need to add a taper to avoid discontinuities around the RF pins.
The QM12202 is in production, please contact appsupport@qorvo.com for the datasheet.

image
image622×572 112 KB