Re: USB connection hangs up after about 1 minute
Posted: Sun 19 May 2024 2:03 pm
I have had an extensive and prolonged battle with my FO 3083 and later 3085 stations locking up and not being able to reset. Sometimes it would run uninterrupted for months and then I'd get half a dozen crashes in a couple of hours. First, I went looking for the cause. I tried stabilising the power by cutting the USB power feed and running only from batteries. I also tried bridging one battery position to reduce the voltage to 3v. None of that made any difference. I changed the computer I use to run CumulusMX and now run on Linux, not Windows. That also made no difference. I fitted chokes to usb cables and tried shielding them. Again, no effect.
One evening, I was web browsing on the machine which runs the station. I heard a car being driven very fast outside and at that exact moment, the station crashed. I reset it and less than an hour later, the same thing happened again. I guessed that radio interference from the car caused the station to crash. Long story short, I bought an RTL-SDR and ran rtl_433 to monitor the 433MHz ISM band. I wrote a database to capture and correlate the rtl_433 outputs and the weather station outputs including alarms. The system also detects and monitors TPMS codes (handily also on 433). To back this up, I also installed AgentDVR and installed a security camera which watches the passing traffic. The results have been impressive. In my case, the interference and locking up is caused by people using one of two devices in proximity to the station, HackRf and FlipperZero. This is confirmed by video capture. I can now identify the different types of interference and predict when they are likely to occur. I have a short list of vehicles associated with it that I can identify by their TPMS sensors even when they are not seen by the security cameras. These devices are sometimes carried by pedestrians. I'll leave you to google what they are doing...
At this point, I have a Cumulus alarm set up which runs a python script and resets the station by interrupting the power with a serial relay. This is only required with the most disruptive interference. Other events occur which cause data spikes - usually in the humidity, solar and UV data - but those events do not cause a lock up. They are detected on camera and in some rtl_433 output though. The only problem I still have is that the pressure sensor needs to be recalibrated after every reset. Still working on ways to do that.
I'm not saying that the cause of interference is the same in every case but it definitely is radio interference which the base station does not correctly reject as it should. Devices operating on 433 have become increasingly common in recent years and I'm guessing that the number of them which can cause this is increasing all the time.
One evening, I was web browsing on the machine which runs the station. I heard a car being driven very fast outside and at that exact moment, the station crashed. I reset it and less than an hour later, the same thing happened again. I guessed that radio interference from the car caused the station to crash. Long story short, I bought an RTL-SDR and ran rtl_433 to monitor the 433MHz ISM band. I wrote a database to capture and correlate the rtl_433 outputs and the weather station outputs including alarms. The system also detects and monitors TPMS codes (handily also on 433). To back this up, I also installed AgentDVR and installed a security camera which watches the passing traffic. The results have been impressive. In my case, the interference and locking up is caused by people using one of two devices in proximity to the station, HackRf and FlipperZero. This is confirmed by video capture. I can now identify the different types of interference and predict when they are likely to occur. I have a short list of vehicles associated with it that I can identify by their TPMS sensors even when they are not seen by the security cameras. These devices are sometimes carried by pedestrians. I'll leave you to google what they are doing...
At this point, I have a Cumulus alarm set up which runs a python script and resets the station by interrupting the power with a serial relay. This is only required with the most disruptive interference. Other events occur which cause data spikes - usually in the humidity, solar and UV data - but those events do not cause a lock up. They are detected on camera and in some rtl_433 output though. The only problem I still have is that the pressure sensor needs to be recalibrated after every reset. Still working on ways to do that.
I'm not saying that the cause of interference is the same in every case but it definitely is radio interference which the base station does not correctly reject as it should. Devices operating on 433 have become increasingly common in recent years and I'm guessing that the number of them which can cause this is increasing all the time.
