New Framework 16 DIY won't start when RAM Channel 0 Occupied

The TLDR is I can get my new DIY to where I could start installing the OS from my USB, but only if nothing is in RAM channel 0.

While I successfully built my desktop 2 years ago I am still rather amateur so I am trying to be thorough. I have already searched through the forums without luck.

It works with either stick of RAM in channel 1, but only if channel 0 is empty.

I have tried every combination of 1 or the other sticks or both sticks in every combination of ram channels 0 or 1 or both. Also empty, that was how I found it was not a display issue.

I also tried the suggestion of letting it finish training with 1 stick and then after that is successful, turn off, install the second and that also didn’t work. I tried that with each stick too just in case.

It seems to train on 1 stick of RAM in under 2 minutes, with both slots occupied it will start blinking the error/boot code in a similar time.

With it fully assembled with both sticks of RAM it gave me: white, 12 greens, orange, blue, green, blue, green, green, green, blue, blue.

From this:My Framework Laptop (Intel 11th Gen Intel® Core™) is not powering on . If the codes are the same for my Framework 16 as the 13 it passed the boot check, but I cannot find anything on what the post code at the end indicated.

I am open to if anyone knows what to do?

The instructions were to ask here before support but from my limited experience and extensive looking around it looks like a motherboard issue.

The last hope I had is to disassemble it again and reassemble it with only channel 1 occupied. Go through installing the operating system and make sure everything is up to date. And then rebuild it again with both sticks. But I have already had to unscrew and rescrew that mid-plate 4 times now and I am getting tired. I wouldn’t mind doing it once a year but that about an hour every time I have to go through those 16 screws.

I have a Framework Laptop 16, AMD 7040 Series, just the cooling shell.

RAM: 2x16GB T-Create Classic 5600

SSD: WD Black SN7100 2TB

Help would be appreciated,

Tyler

could be mem training wait about 15~ minutes

please inspect channel 0 socket. there were reports that some pins were bend.

2 Likes

FW support is needed to interpret post codes. Since the beginning, post code flashes have never been documented officially. There was / is a github gist with “port80 codes”, but no info on if or how flashes convert to codes, it’s years old at this point, unknown what changes there are on newer boards, and the codes are hard to interpret, since they are low level and won’t correspond directly with what the source of the issue is. I don’t recall anyone actually finding them that useful.

Not fun, I can imagine.

If I may make a suggestion, next time you open it, take a lot of pictures, multiple angles, overall board, different regions, ram slots of course, angled to see pins if possible. FW support has a tendency to ask for pictures, I think they want to be sure there aren’t any other issues that need to be addressed in order to get everything working. Anyway, taking an abundance of pictures next time you’re in there can save having to open it up yet again when they ask.

The blink table is indeed supposedly the same. 12 greens suggest “everything ok”. Clearly it’s not.

“blue, green, blue, green, green, green, blue, blue.” means 0x1010 0011, which is “code A3”.

Sounds .. normal?? Usually “Ax” is good. For AMI this means “IDE enable”, so it probably means it went through everything.

Not really a thing. On the 16, it turns on basically immediately, with a CMOS reset as well.

You can run the computer open, you dont need the midplate. If you do that, you can then just ignore the errors with “ketboard or midplate missing”. The computer will still work. Just grab a USB keyboard and mouse.

So it just don’t post. How curious.

For the port 80, correct. But you can try your luck with a generic table on the internet. It’s a Insyde BIOS, and I dont think they have a reason to change anything.


Idk that sound like bad memory channel; reach out to support and do the RMA process.

A very curious bad channel, since it pass initial training.

You can try a CMOS clear. Plug the laptop in, remove the midplate. The LED should blink red.

With the chassis intrusion (midplate) switch. Press it 10 times. But you must press it slowly.
So blink, press, orange, release. Blink, press, orange, release. Do this at least 10x. It might flash white at the end, it might not. Either way, do that, and then attempt to boot

1 Like

First of all, thank you for the quick and helpful responses.

Knowing that the Input modules will still work even before the screen backlight turns on was really nice to make testing faster.

As was knowing that even with both RAM modules installed it should still be only a few minutes not 2 hours to wait between attempts. I noticed today that while I stated the successful time, I had edited out that the first thing I did after confirming that it is normal for the screen to stay off during training was leave it overnight. (For me it always takes from 1-2 minutes.)

The good news is after removing everything I had installed to get photos, while I could not see anything suspicious I did take the advice and gently used some compressed air into the slot and then inserted the module into the slot and then it worked! (I get then “default boot device missing[…]”, but then I can get to the setup utility. Which is what I expect.)

Until I installed the midplate again. Or rather fully reassembled it and was then very confused why once again the screen never turns on and I get the same light code as above.

So I took the midplate out again and it starts at “boot device missing” and works as expected. (borrowing the mouse & keyboard from my desktop).

Gently placing the midplate but not connecting anything and it still works.

Plug in the connector #1 it works, but starts at “The input modules not detected” screen and then I can go to the boot device screen.

Screw it in and blank screen. Unscrew, it is fine.

Testing one at a time It will work after the first two screws are tightened, but will stop once the third screw(the one numbered 4) is tightened.

I have not had luck so far finding a closeup view of the backside of the midplate to see if maybe a thermal pad is missing, I am not sure but there might be one missing here.

The other side has one.

I still have more work to do but I wanted to thank everyone for the help so far.

P.S Sorry if this is hard to follow, I should have gone to bed awhile ago but I thought it was over.

3 Likes

I’m glad you made progress.

How curious.

Is the lid close switch somehow getting activated? Is there something thats shorting against the board? It’s somewhere along the bottom of the top cover, right above F7 (if your keyboard is in the left position)

you can try reseating the connectors that go to the display panel, its next to the Wifi card. Two big boys, don’t mix them up.

I can’t tell what is shorting if that is the cause.

I reseated the display cables a few times and that didn’t help either.

I tried assembling it with the screws just barley threaded as well as as tight as I can get them and still no change.

I don’t know what it is and I’ll try reaching out to support tomorrow.

1 Like

I don’t know what I screwed up, but I had to reset the motherboard.

I don’t know why it didn’t work the previous times, but my advice is to wait for the second red light between presses. There is no indication if it worked so I don’t know it that is neccassary.

Bazzite is impossible to install I cannot get past the error messaages.

Windows 11 worked, so I have that running now. If I can catch up on work I will try installing Bazzite now that it can boot without error messages finally.

This was far harder than building a desktop from scratch, I would not recommend this to anyone who does not build computers every day and who is ready to spend 10x longer every time something needs to be fixed. If this is the “Most Repairable Laptop” I am never working on a laptop again and the next time something goes wrong I am just sending it to someone else.

TLDR This thing rattles like a $300 laptop and I am only going to work on desktops in the future. This was a mistake and I am done.

1 Like

I’m glad resetting it worked. Not sure how, maybe a forced memory training.

I tried steamOS (the SteamOS), the install image doesnt even start.

I think all “mainline” would work (Debian branch, RHEL branch …)
I like how in the grand scheme of things arch is such a twinkle in the sky