I don’t yet have my FW16 so, I’m going to make some assumptions, sorry if I get it wrong.
It seems to me like the plastic shield only needs to be slid over the very edge of the WiFi card. So it should be possible to put it on the card by just slightly raising the card, without removing it completely from the slot.
So, if I’m seeing the picture correctly (again, I may be completely off), I would say you need to install the card into the slot, install the wires, slightly raise the card edge (so, release pressure on it), slide the shield onto it, bring the card back down and then screw it in.
Ok, so it could be that it is not a compatibility problem, but an actual hardware problem, that happens.
EDIT: That information from Framework also says “limited validation”, I am not sure what exactly that means, but that would indicate to me that it has not been tested enough to guarantee that it will work yet.
Hi,
I am learning stuff in this thread. For example, memory training up to 15mins often with no error codes or indication of what it is doing !!! :
Yep, the memory training is relatively new, and has recently started taking a lot longer. All a result of pushing things faster and faster. Just reading that initial summary, I can say that is not accurate though. All new systems (at least ones without soldered memory) will do some amount of memory training when RAM changes, CPU changes, BIOS is cleared, or on firmware update. Some desktop motherboards have an indicator showing memory training, or some will show what looks like a DRAM error. Often it will reboot at least once during memory training. This mostly seems like a new issue because before DDR5 it was uncommon for it to take more than 10 seconds, and usually more like 1-2 seconds.
At least this doesn’t need to be done very often, just really scary when it always happens on first boot making it seem like something went wrong with the build unless you know about this.
Well that doesn’t bode well for me - almost 50 and also wear reading glasses!
(And to all the zoomers reading who are thinking “oh, some old geezer, lol” - 51 is NOT old!! And yes I will keep telling myself that.)
Anyway Mark, I just wanted to say that everything that seems bloody difficult at first becomes easier with practice. So don’t send it back - practice! Give yourself a chance to get the hang of it (and let the rage wear off a bit).
Maybe it won’t make a difference but maybe it’s worth a try?
Don’t think so, more a problem with being unfamiliar with the connector type and the techniques for dealing with them.
another one is FPC connectors and flexible PCBs. The number of different ways to unlatch those connectors … well, I’ve lost count, and they are REAL easy to damage.
I suspect the OPs real problem is the antenna connectors haven’t latched onto the card, and that is why they are coming off so easily. Maybe try squeezing them between thumb and index finger to get enough pressure to make them latch.
It sounds to me more like a faulty RAM socket than a faulty module. There have been reports of faulty sockets on the FL13 machines.
As a check, once you had the machine boot up, did you try the other module in the same RAM socket (again just one RAM module)?
could also just be stuck memory training.
If you know why you can’t really get 5600 on a SODIMM DDR5, the answer is signal integrity. It’s running at like, 2.8GHz (transfer on both the rising and falling edge of clock). The small, cramped socket that’s SODIMM literally is struggling to keep up. Most DDR5 SODIMM systems run at 4800MHz, and just for comparisons, on desktop you can easily push 6000, 6400, (and beyond, although we have yet to see a CPU that have a sufficiently capable memory controller) because the DIMM is much larger, and have better signal clarity, despite the PCB trace sometimes being longer. But even that is starting to affect things, especially for overclocking.
Knowing this, memory training can become tricky and sometimes nigh impossible. Older systems aren’t affected nearly as much, since even something like DDR4-3200 run at 1.6GHz, almost half.
I assume people walked into Framework 16 expecting to sacrifice some memory speeds. I don’t know why it took the 16 this long to fall back to a slower speed. Could be because it’s mostly working, except for like the 5% on the edge of the PCB. It have to do much more training. Or it could pass basic training but fail to boot due to still insufficient signal clarity.
This is why the entire “CAMM” memory module is invented – pogo pins, direct PCB-to-PCB connection. Excellent signal integrity, also slightly higher density. But it’s after Framework had basically finished making the Framework 16. This AMD version, at least. It is yet to be seen whether they are going to be making a new Intel board featuring CAMM modules.
It’s not that it’s small that’s the problem it’s that it’s so wide and the pins are so far apart which makes the minimum trace length super long.
Desktop also doesn’t need to care about battery life and can brute force it by using higher voltages and more power. The dimm being larger is still worse but higer signaling voltages more than make up for it.
And most importantly much shorter min trace length and a form factor that can be put closer to the cpu which means even shorter effective trace lengths. To an extent where you can even use lpddr which is even more sensitive to signal integrity.
Been rather frustrating to watch people be dismissive of the issues reported. A laptop selling itself on repairability which makes servicing things like these so hard would make me very frustrated as well. I understand the decisions being made are due to things FW doesn’t have direct control over, but this isn’t “the ethernet port expansion card sticks out of the body because of the port form factor”, this is an accessibility issue. I’m sorry your experience was like this, I hope FW looks at this experience and thinks of ways to improve the design.
Repairability doesn’t necessarily mean trivial or not requiring time, skills or patience.
Having installed or replaced wi-fi adapters several times, those antenna connectors are awful to get on and off, but they’re what we have.
And I’d rather it was something standard, than something specifically bespoke to Framework.
I don’t understand: what about this order
- connect the wires
- insert the card in the connector (letting it going up for the moment)
- pushing the card down by screwing the cache + screw.
This is what I always do on my FW13.
I second that. If Framework were to invent a new WiFi-connector to counter this issue, I bet they’d be criticized for not being standard-compliant and manufacturers would charge an extra customization fee. Maybe someone could come up with a connector adapter to improve usability, but most people would consider that to be overkill.
Welcome to the dark side of miniaturization.
From just that one picture it’s quite hard to tell, but I fathom the helping hands and magnifying glass combo used for soldering works in conjunction with a pair of pincers should do the trick.
The FL13 doesn’t seem to have said plastic cover.
This isn’t accurate: it has the same kind of cover part, it’s just in metal instead. The cover is used in the same way at the same place.
So, the same steps should work, unless I am missing something.
I’d be really happy if you gave this laptop another chance. It seems like you share the same dream of a laptop upgradeable/moddable like a desktop.
Maybe you can get in touch with Framework support and have them take a video (or maybe add proper documentation directly) so you can keep this laptop.
I get why you’re a bit furious, after waiting so long. But now that you have it, do you really want give in so quickly?
You got this
I am not asking for non-standard solutions. I’m saying if you sell your device on being able to be serviced by end-users, and it requires more than the included tools to make it not be a frustrating experience for the average user, maybe you should include those other tools.
Independent of that, the FW people are very talented and dedicated to making a device no one else has, and have had to solve a lot of complex problems along the way. It’d be nice to see some attention put to this to let more people seriously consider a purchase like this, instead of having to pass on it because they don’t have the dexterity or tools to service their device properly.
The generic screwdriver with a flat end they include should be all you need to do this.
You just need to make sure the connectors are really connected before putting that cover back on.
I’m not sure the OP has gotten them fully in place, otherwise the cover should be doing the opposite of what he’s saying.
There is no special tool to accomplish this, unless you’re saying they need to include jewelers glasses with each laptop?..
I’ll point out this:
Our philosophy is that by making well-considered design tradeoffs and trusting customers and repair shops with the access and information they need
Some things will be beyond end-users, be it because they have poor eyesight, fat fingers or just no aptitude for it, and that does not make the Frameworks any less sustainable or repairable as a result.
The one on mine is plastic