04-08-2019 11:06 AM
I acquired a Rocket 5AC Prism Gen1 unit that had a broken reset switch, but was otherwise in immaculate condition. I removed and replaced the switch, no problem. However, now I'm trying to reset it and I'm having issues. It boots up to a mode where the signal LEDs flash in 1,3 / 2,4 pairs, alternating back and forth. I gathered that this was recovery mode, so I sent it XC.v126.96.36.199181.190213.1104.bin via TFTP. It took the firmware, seemed to reboot (signal LEDs did some stuff where they went up and down at one speed, then another, faster speed), and then it just reboots back to the alternating LEDs mode. The reset button does not appear to do anything, even when I just short the pads.
Does anyone have any ideas? I can ping it on .1.20 just fine. Perhaps I'm flashing the wrong firmware?
My next idea was to probe around for a serial port on the board but I figured I'd ask here first.
04-08-2019 01:12 PM
I read on the TFTP knowledgebase article that you need to flash the exact same firmware. I went through the list of old firmwares and tried flashing each one, but never found one that worked.
I found a serial port on the PCB that seems to print some debug info on startup -- it's too laborious to decode the messages with my oscope though, so I need to get a RS232-USB bridge to get the full output. My hope is that I can figure out what FW is currently on there so I can reflash it (perhaps it's a beta firmware?).
04-10-2019 04:03 PM
@UBNT-James bump, see SNK's post, does the TFTP recovery firmware have to be the same as the original for the recovery to work? Is there any special firmware I can TFTP that will get around this restriction so I can put a new firmware on?
I borrowed a logic analyzer from a friend, but it can't do the 24V signalling on the debug output on the board. So, I ordered a board on ebay to drop it down to a lower voltage, but that won't come for a week I think.
04-10-2019 09:56 PM
I got my serial output working, it appears the PCB outputs an inverted polarity serial signal, so I fixed it with a MAX232. Here's the output from the main processor during bootup and when I send it a firmware bin via TFTP:
The reason this device was being sold was the reset button was physically broken off the board. I'm wondering if maybe the circuit has failed and it thinks the reset button is constantly held down. There's still one more UART on the board that I haven't looked at yet, but I will investigate that one soon, too.
04-11-2019 08:45 AM
We recommend using the same firmware that is already on the radio. TFTP should be used for recovery only, not to upgrade/downgrade firmware.
There isn't much we can do for a hardware issue like this. I would suggest replacing the unit.
04-18-2019 03:29 PM
It turns out that the POE injector that I got with this radio was causing the issues. Although it is a legit UBNT adapter, and outputs 24VDC @ 1A, it doesn't have a reset button on it like I think the R5AC POE injectors are supposed to. I learned that these radios have a remote reset functionality, and I guess this injector was wired in such a way that it was constantly resetting the radio. I swapped it for another injector and was able to properly reflash the firmware. The radio functions perfectly now. I am glad I did not discard it and buy a new one.
In the course of the troubleshooting, I did connect to the board's UART and poked around in the u-boot bootloader console. To do this, you need to use a MAX232 chip to invert the polarity of the signals so you can hook the UART up to a RS232-USB adapter.