Disable faulty soldered RAM on Acer A315-22

[mat2k]

Hello, I've been googling for some pointers on this but couldn't find anything that applies to this board. I have an Acer A315-22 (board NB8607) with faulty RAM between 142 and 174mb. I just wanted to disable the onboard RAM altogether, but this one doesn't seem to have any memory down ID resistors or strap in order to disable it. I'm attaching the schematic I found, the boardview for it also seems to be available somewhere in this forum but I still don't have a use for it lol.
How would I disable the RAM on this PC? Thanks for the help.

[MOD EDIT] LINK to boardview + schematic https://www.badcaps.net/forum/troubl...ew#post1709776

[mcplslg123]

Shift
RU180 to 188
RU179 to187
RU178 to186

Refer to page 4 & 9 of schematic. You'll need boardview to try this.

[mat2k]

Shift
RU180 to 188
RU179 to187
RU178 to186

Refer to page 4 & 9 of schematic. You'll need boardview to try this.

Thank you, I missed that entirely.
Unfortunately it doesn't seem to work, after moving RU178 to RU186 (the others were already set in 188, 187) the BIOS still reports the onboard 4gb and i can access them from Memtest. I suppose the "no onboard memory" code could be something else, or it isn't even coded in?

[Davi.p]

Do a ram reballing.. try also freezing them for test..

[mat2k]

I guess I'll try, but I would still prefer to disable the RAM to avoid any issues. Btw, I have also tried to set the three bits to 1 (shifting resistors to 178, 179, 180 - I tried this since some other Acer models seemed to use this scheme for no onboard RAM) and nothing changed - exactly the same behavior as with all the bits set to 0, or with the original configuration

[Davi.p]

Hi, since modern cards has so tiny resistors, have you checked three resistors has not been cooked by the soldering?

[mat2k]

Hi, since modern cards has so tiny resistors, have you checked three resistors has not been cooked by the soldering?

Of course, I used leaded solder, sufficient flux and a heatgun to move them, and then I measured all of the connections from other pads using the boardview. All of them measured very close to 10k.

The reballing seems to have fixed it. I was extremely skeptical that it would, because the issue was just a few bits but in two different pages. Since the issue was around the 150mb mark, I just reballed the 1st and 4th chip (because I didn't know which of them would be the first gb) and tested a few minutes afterwards, no errors. now i left it overnight so it cooled to ambient temp (around 10°c) and I still don't see any errors. I'm still intrigued about the resistors though, right now they are all set to 1 and the memory ID is the same as before, plus everything works.

[Davi.p]

Ok just the last step to do, seal the chips around with hot glue, better all the bga.. this is ex tremely important, do it, don't ask me, if you don't want a short time recurrence..

[piernov]

The reballing seems to have fixed it. I was extremely skeptical that it would, because the issue was just a few bits but in two different pages. Since the issue was around the 150mb mark, I just reballed the 1st and 4th chip (because I didn't know which of them would be the first gb) and tested a few minutes afterwards, no errors. now i left it overnight so it cooled to ambient temp (around 10°c) and I still don't see any errors.

Unlikely to be a soldering problem, heat probably affected RAM IC itself.
Also no relationship between memory address and IC here, all ICs are used for all addresses since there are 4× 16-bit RAM ICs to form a full 64-bit RAM data bus. Affected bits correlate directly with an IC though.
If there were more ICs (or larger data bus width per IC), multiple ICs would use the same data bits but at different addresses, however the address map is not linear, there's some sort of complex interleaving going on and the mapping is often not public.

[Davi.p]

Do it, with confidence, take it as an experiment, if it will fail you'll do no more.. you can't damage things..

[mat2k]

Unlikely to be a soldering problem, heat probably affected RAM IC itself.
Also no relationship between memory address and IC here, all ICs are used for all addresses since there are 4× 16-bit RAM ICs to form a full 64-bit RAM data bus. Affected bits correlate directly with an IC though.
If there were more ICs (or larger data bus width per IC), multiple ICs would use the same data bits but at different addresses, however the address map is not linear, there's some sort of complex interleaving going on and the mapping is often not public.

oh well, i still don't trust it and would definitely disable it if I could, but it doesn't seem to be possible. I was considering trying random memory ID bits, but I doubt they do anything at all in this board.

I didn't know that's the way RAM is laid out, makes sense. In any case I was able to install Windows without issues, and if the bad pages come back I guess I can just add them to {badmemory} with bcdedit