This topic was migrated from the BotBlox ticketing system after being summarized and anonymized.
Problem:
I’m using a SwitchBlox Rugged (Rev G) in a test setup and I’m running into a port link instability that I believe needs a static firmware configuration to resolve. I’ve done fairly thorough testing to isolate the cause and wanted to share the results.
Setup:
- Board: SwitchBlox Rugged, Board Revision G (please confirm whether my unit is the BB-SWR-G-1 PicoBlade or BB-SWR-G-1-PC PicoClasp variant — I can verify on my end if needed)
- Powered from a stable 12V bench supply; PWR LED is on solid
- Port A: Radxa CM5 companion computer
- Port B: Herelink Air Unit (Ethernet port)
- No other devices connected
- All cabling is correctly wired (device TX → switch R, device RX → switch T) and twisted-pair integrity is maintained
Symptom:
When connecting the CM5 to the Herelink Air Unit through the switch, the connection periodically drops out (in my application a video stream comes up briefly and then disappears).
What I tested:
- CM5 ↔ switch link is stable: 100 Mbps, full duplex, carrier stays up continuously (confirmed via sysfs, carrier=1 throughout).
- IP connectivity works and small/standard pings succeed.
- Running a sustained flood ping (ping -s 150 -i 0.01 -c 500) from the CM5 to the Air Unit THROUGH THE SWITCH gives 26.2% packet loss, and critically the loss occurs as one continuous blackout block of ~130 consecutive packets, not as scattered loss. Latency outside the blackout is rock solid (~1.0 ms). This pattern is characteristic of a link flap, not an MTU or bandwidth/capacity limit.
- Running the exact same flood ping with the CM5 connected DIRECTLY to the Air Unit (switch removed) gives 0.4% loss (essentially clean — the 2 lost packets coincide with the cable swap).
Conclusion:
The CM5<->switch link and the direct CM5<->Air Unit link are both healthy. The instability is specifically on the switch<->Herelink port. The Herelink Ethernet port appears to be a fixed-speed (likely 100full, no auto-negotiation) interface that I cannot configure, and I suspect the SwitchBlox port’s auto-negotiation / auto-MDIX is not settling into a stable link against it, causing periodic flapping.
Solution:
There is board you can purchase to stack onto SwitchBlox Rugged, called Rugged SOM hat implements a full routing OS (Open WRT), to allow full switch and routing management. That will give you the ability to tun off auto-negotiation and fix a port speed.
In all honestly, however, this is overkill for what you need. A much better solution would be to use SwitchBlox Industrial instead. It has the same number of ports, but runs a lightweight CLI that you can use to fix port speeds. It’s also smaller and cheaper than SwitchBlox Rugged, and we know people use this to connect to similar devices you are using.