RAK5802 Ground needed?

I am using a RAK5802 to communicate with an RS485 sensor probe. I have everything working great and I am able to pull the data from the sensor. However after a period of time, I begin to experience timeouts on the RS485 side. I’ve replaced boards, and wisblock items, and pretty much everything but this still randomly happens. Today once I got the failure notice, I decided to bring the device in and touch a ground to the ground pin on the RS5802 quick attach block and then my device started working again. Aside from the probe the entire setup is RAK Wisblock devices. I am utilizing the RAK19013 for the Solar charging and a 3.7 LiPo pouch battery. I am using the RAK19002 to power the sensor probe as it needs 10+ volts to operate.

I suppose my questions are, has anyone else had an issue where a ground was needed? Is the RAK19013 producing noise in the power plane that might be skewing the differential signaling of the RS485 chip? Is there a way to emulate ground without needing a ground? Does the RAK19002 GND meet with the rest of the GND plane on the board? Since the VCC and GND for the probe are coming from the RAK19002, do I need to find a way to link the GND back to the RAK5802 pin?

Probe device is a 4-pin RS485 with VCC, GND, RS485-A and RS485-B.

VCC and GND connect to RAK19002.
RS485-A and RS485-B connect to the quick attach block on the RAK5802.

Everything thing else is running like a champ, the device still connects via lorawan, even with failed values. Again, simply touching a ground to the ground pin on the quick attach block suddenly fixed the probe not working so I’m thinking this may be a electrical issue/signal issue on the board components.

Pig

Welcome to RAK forum @wetpig ,

RS485 can work without GND but GND connection together with the A/B lines is not uncommon. If the GND of the probe is coming for RAK19002 and the GND of the RAK5802 is on the IO Slot, you can see that a long trace/path between them which can potentially causes reference offset (my speculation).

I am not sure how long the probe is from the RAK5802 but connecting their GND together will likely make the differential signal more stable just like what you experience. Potential down-side is the ground loop created but in your case, it seems the ground loop doesn’t cause issue.

What you can also try is connect the GND of RAK19002 and RAK5802 together if connecting the GND from the probe to RAK5802 is not possible. I am not sure if this has effect on your setup but should be easily done since they are in the same WisBlock Base.

Hi there @carlrowan ,

Thanks for taking the time to reply. I ended up doing the item you mentioned last. I am still having the issue. I also changed out the probe with another and exeperienced the same issue, however I noticed in the database some of the probe information is retrieved while some isn’t. So now I’m going down the path of possibly changing the probe baud rate to see if that’s the core issue.