Linksys WRT54GL

[jaclaz]

Not necessarily on that model, but sometimes leds blinking mean "cannot boot".

Sometimes these bricked devices can be resurrected with some magic spells (actually fiddling with inserting the power supply, pushing the reset button in a certain sequence, etc.), since it costs nothing you can try:

https://forum.dd-wrt.com/phpBB2/viewtopic.php?p=98742

jaclaz

[D.Draker]

I once had a case where the lightning bricked the device's BIOS, and I was able to resurrect it with magic flashing a new BIOS version from within DOS environment.

[Saxon]

4 hours ago, D.Draker said:

I once had a case where the lightning bricked the device's BIOS, and I was able to resurrect it with magic flashing a new BIOS version from within DOS environment.

Why new? What was wrong with the original, curious?

[D.Draker]

3 hours ago, Saxon said:

Why new? What was wrong with the original, curious?

The original BIOS was nowhere to be found, and the one I used was meant as an "update", so I used it. I had nothing to lose anyways.

[D.Draker]

Oh, how could I forget to tell?!? I even saved a bricked GPU that wasn't recognized as a GPU anymore, I had to short the dedicated pins (while it was booting up), it cleared out the chip off the remnants of the invalid BIOS someone stupidly flashed on it, so I was able to re-flash with the original. Probably it will work for the router, too.

[UCyborg]

I've read the linked thread and some others as well, but these simple measures with reset button and power cycling don't change anything. There is an anecdotal evidence that suggests the possibility that rewriting software and data on the router could help, assuming the hardware survived and it's just choking on bad bits. Apparently some people were employing the shunned upon practice of shortening specific pins on the flash chip, which presumably nukes the core operating system and may allow it to be re-uploaded via TFTP (may also need valid settings in NVRAM), but I will avoid shortening and if I'm going to try anything, I want to do it more professionally. It would also indicated whether hardware is still OK if done properly.

I have an idea, but being more of a software guy, I have to ask what is probably a banal thing for people tinkering with electronics, but here goes. It's about unused holes on the PCB, where you can solder pin headers, the holes' diameter is 1 millimeter, they're apart about 2,5 millimeters, center to center, presumably this might actually be 2,54 millimeters, at least that's supposed to be the standard, but don't have any more accurate tool to measure exactly. The question is, does the male end of the jumper wire such as this one fit properly in the hole? There is another more detailed picture of presumably the same kind of wire, maybe just different manufacturer. Or would there be an issue, maybe loose and wouldn't conduct properly? I only found two vague hints on the internet it could work, one is below:

Source: https://electronics.stackexchange.com/questions/249140/how-can-i-make-connection-on-pcb-circuit-board-holes-without-solder-for-proto

The picture below:

Another way? This appears to employ press-fit mechanism. I found a pin-header recently that looks like this where you push it in the hole, doesn't require soldering, but not available on Slovenian market. I'd have to find it again as I didn't note the link, I think it's available from DigiKey, unsure about buying from DigiKey as a Slovene, first post about it I found on the other reputable (Slovenian) forum suggests there shouldn't be an issue though.

Right now, I'm most curious about whether plugging that wire in the hole directly is a valid step in making a properly working connection from the board's hole that doesn't have any header attached/soldered to it. I'm not going to need a permanent connection to a router's board (in case discussions around the subject mostly imply the need for permanent connection).

Edited March 5 by UCyborg

[Saxon]

That CapXon caps don't look good. It's from the 3rd lowest tier.

[UCyborg]

Ugh, I see it's possible they could've taken out another component.

https://www.badcaps.net/forum/troubleshooting-hardware-devices-and-electronics-theory/troubleshooting-designing-networks/35202-linksys-wrt54gl-v1-1-bad-capxon-caps

You'd think the modding guides for this router would warn about it: "Before you do anything, might as well replace those".

By plugging onto a serial port with USB to RS-232 TTL converter, it could be seen if the console says anything, maybe erase NVRAM as well while plugging onto JTAG port would allow to re-install the bootloader. Apparently you can do the latter with Raspberry Pi, the older method is crafting special JTAG cable to plug onto computer's parallel port. My computer has that port, but this method doesn't seem worth the bother while Raspberry Pi looks interesting for other uses.

Regarding pin headers, I was looking at these:

https://www.digikey.si/en/products/detail/samtec-inc/pht-112-01-l-d/19020741

https://www.digikey.si/en/products/detail/samtec-inc/pht-110-01-l-d/10218549

One for JTAG (marked JP1), one for serial ports (marked JP2, I know only first row would be strictly needed, but what the heck), if headers are really needed to make good connection with jump wires and if I'm even looking at the right ones??

I realize replacing caps would make sense if the device still works in the first place, where soldering is unavoidable.

The whole thing is mostly about satiating my curiousity at this point.

[D.Draker]

Despite quite common belief, CaXpons can live up to 7 years just fine, after they need to be replaced 100%. 

[UCyborg]

Asked around a little and got things cleared up regarding connections to the PCB. Those wires employed on the picture up there, they totally slipped by me, the method is described here. Another good solderless method would be press-fit headers. Just putting male headers in the hole wouldn't make a good contact.

I didn't know we have a bunch of plain pin headers available at my workplace, the coworker from production said it was not a problem for him to solder them on my router, so I got that done in no time.

Then I got a Raspberry Pi 5 and related accessories, set it up, got a look around the Raspberry Pi OS a bit, tried a bit of web browsing, put on Visual Studio Code, first time trying the latter and finally, I believe I have, at least in theory, a working version of tjtag-pi, my version is not published anywhere ATM, but generally it can access the router's CPU and flash memory over JTAG port, adapted to work on the new Pi, with the help of WiringPi library. I added one new function in the latter, which should flip the states of certain GPIO pins at once, or at least should happen a bit faster than the sequential way offered by the library, I basically just wrote a function that takes bit mask of pins to be set/clear, to closely emulate the GPIO pins maniupulation code of tjtag-pi.

Whether this is necessary or not, I have no way of testing or otherwise knowing whether sequential flipping is really incorrect or just slightly less optimal, it's not a long intensive operation. But I've read about "bit banging", which is how communication is done through JTAG, that generally when it comes to bit banging, timings matter. Just in general, but in the case of these routers...the program has an option to wait between changes/reading of pin states, though judging by the comments, it was just a workaround for flimsy physical connection to JTAG port.

As it turned out...

On 3/3/2024 at 3:53 PM, jaclaz said:

Not necessarily on that model, but sometimes leds blinking mean "cannot boot".

...I realized after trying another good working power supply, which compared to the other two, has a LED and it was blinking as well when powering the router while it should remain lit, that means only one thing, short circuit. So guess in best case scenario, only capacitors are faulty and in the worst case scenario, something else is fried. Ehhh...

Oh well, I've got a working Raspberry Pi 5, so that's something.

Edited March 14 by UCyborg
Additional clarifications