We have a few Wio-E5 mini, added shielding and filter to get our range and stability. However, when the device move pass some 5G towers, the device will freeze/hang. It can only be recover with reset.
We re-create scenario in our lab, and can repeat it all the times. With a RSSI of -75dBm, at 868MHz, and when we send a signal of -20dBm at 871MHz at the device, it will immediate freeze. Removing the jamming signal will not recover it. Either by resetting it, and sending a Lora Signal at -50dBm, then it recover.
Test SF6,7,8 with various FEC and bandwidth, with the SAMe result.
Anyone seen this before?
Hi there,
Wow man, this is a classic (5G related) RF failure mode playing out hard. Props to you for recreating the issue in a controlled lab, that’s half the battle right there. 
It’s sounding like front-end overload, or more specifically:
Receiver gets so flooded with energy near its tuned frequency, the internal LNA or AGC path collapses. That’s why the receiver wedges and needs a new signal injected to recover. the PLL or AGC is getting jammed by it.
A similar thing occurs in RadaR with some RF signals and 5G signals.
this why 5G is a problem for other tech as well. doesn’t help it’s Lora is close to cellular freq’s. 
AL, thinks
you could also experiment with:
radio.standby();
radio.sleep();
radio.begin();
as a forced un-wedge sequence., YMMV 
In the current form I’m thinking Shielding MORE is they only other option, Around the LNA (if possible) or add a watchdog recovery logic no RX for extended time. A low-loss SAW filter will almost certainly help
HTH
GL 
PJ 
BUMP… to see if we get anything from support.
I would be curious to know if the Other Lora device have the same issue, like the
The problem is, there is no errata or way to detect it (yet). My setup is extremely well shielded, and very advanced filter in the front end. But having signal that close, just not possible to filter (for now).
Current problem, we are using it for “high speed” data, so watchdog recovery or “reset” cycle, ain’t going to fly. There should be a errata or suggested workaround, I hope.
Paul