If the serial's dead, then it's probably a proper brick, unless you want to play around with a SPI programmers and such.
I wouldn't waste my time on an old unit as the C20.
I am sorry to hear you have this problem, I would never use a version that required TTFP to install, not because it is bad but because it shows a lack of maturity in the development of the software.
Luci is the only girl for me and even then I am very careful on choosing the right hardware version and making sure the file is an absolute match.
To make matters worse, TP-Link are the only routers I have ever bricked and that was installing an upgrade of their own firmware with no apparent cause.
I do not have the expertise to advise how to fix a bricked device but according to google there are around 850 posts that might be relevant on this site, the link below should help you find them
I agree, if you put a value on your time then cheaper to just buy a router off ebay, but this will not apply if you are in a developing country or even one of the BRIC countries. No pun intended.
@srmarchi You said you can use serial, nothing on there that shows up? Did you try setting up serial?
If you can get into the bootloader you could TFTP an image over (heck, I even used kermit to transfer an image and boot from that with my bricked Archer C2...). I see the layout is documented in the wiki page and there's a picture of the location as well.
Since the headers aren't populated, unless you insert and fixate a set of male headers, it's going to be difficult. I haven't ever soldered any headers myself, but you can ply them underneath the board to fixate them (one to the left, next to the right, etc). It's a poor man's serial header, but it will do.
Trzeba zewrzeć pola gdzie mają być rezystory R600 oraz R601 inaczej nie będzie komunikacji ze sprzętem.
I used the Google Translator:
You need to close the fields where the R600 and R601 resistors are to be, otherwise there will be no communication with the equipment.
TP-Link very often intentionally disables the serial console, either by breaking the lane (no populating a zero-ohm resistor) or by pulling down a pin to ground. The image is not quite clear enough to answer this for sure (and the red frames don't help either to follow the lanes), but in combination with the translated text it appears to be the former. At least R601 is quite obviously breaking up the rx lane, and R600 looks very similar as well.
