but to tell you the truth I am too scared to try it because not enough information is given so how can I decide what to set the init value. If you have any luck please report back.
I soldered a single chip from a KVR400X64SC3A/256 to my wrt54gl and used
nvram set sdram_init=0x0113
nvram set sdram_ncdl=0x000
and got the full 64MB. I first tried a different setting and had to solder a jtag cable (my old one had been 15km away at 4am in the morning so it was easier to solder a new one *g*) and clear nvram and than tried the settings above and voila, 64MB.
Some pics of the somehow rude version of doing stuff... just used a hot-air fan to remove the chip (covered the rest using a bit of tinfoil but still removed the one cap accidentially *g*) and a standard soldering-station to apply the new one, so no special SMD-Tools were used...
I'm looking at doing the 64mb upgrade is there another memory module I could butcher as the one you used doesn't seem readily available in the UK.
Or can anyone recommend a chip that will go straight in ie without the need for extra jumpers.
The chip is a 56-TSOP package, but it can be soldered on the 48-TSOP footprint of the original flash chip. For this, pin 3 of the
S29GL128 has to coincide with pad 1 of the footprint (see pic). This is possible because pins 3 to 26 and pins 31 to 54 have the same assignment
as the original chip, pins 1 to 24 resp. 25 to 48. (see datasheet). Before soldering the chip, I slightly bended up the remaining pins
(1,2,27,28,29,30,55, 56), to avoid any risk of shortcut. This is barley visible on the picture.
Now, there are still pins of the S29GL128 that need to be connected: pin 2 (A22), pin 15 (A21) and pin 29 (Vio).
We can use small wires to connect them directly to traces or pads present on the board:
For A21, I used the same connection I made for the S29GL064A (prev. post).
For A22, this signal can be found on a trace close to the pin (upper left side of the picture).
And for Vio, I connected it directly to VCC found on pin 43. I also added a 100nF decoupling cap (just in case...)
Before being able to flash anything to this chip thru JTAG, I added the following line to the tjtagv2 source code chip definitions: