Durability is important, but when you’re using expansion cards that are as easy to replace as these, durability becomes a lot less important.
If the ethernet port on any other laptop broke, it’d be a huge problem, and a difficult repair. If the ethernet expansion card for my framework laptop breaks, it’s trivial to replace.
Durability is important, but not of the highest importance thanks to the modularity of the device.
It actually costs a lot of plastic to also do the upside down style in support and produces a bad result due to say from the large overhangs
Don’t know until you try 
Extremely highly doubt they’ll be the lowest quality part of the gig. 3D printed parts are usually the lowest-quality, lowest-durability part of any given project.
Well, that’s a quite ambiguous statement 
I don’t see how it could be different one way or the other. It comes as 2 halves, doesn’t it? And the plastic is pretty minimal anyway - if you were to print a fully filled box in the shape of the card, we’re still only talking single-digit dollars of material (if that much). Sure, support can be a waste, but Turbo Supports in Markforged for example are pretty effective at reducing plastic waste for supporting everything that’s needed.
But in this case, I’d imagine a case in 2 halves, and each half would still require no support / no overhangs. Each side has its flat surface against the bed. Join the two together and 
magic, a card is born 
No, it’s a whole piece. Too hard to easily build a strong locking mechanism.
Well hey, congratulations, now you’ve got me cranking the gears towards proving-out a feasible design using these parts 
Just printing some sample parts here for starters - one, the basic frame from Framework’s Github, and a few others from Thingiverse including a solid block.
I’m open to being proven wrong, or right, either way… but cracks knuckles time to warm up that TinkerCAD again.
I’ve found the hinge mechanisms never clasp aswell as the full size, multiple people have said that only want full size none of this half size rubbish. I’m already too deep into this design to change, I have 10 Ethernet jacks that are quite expensive. I can’t change. I might bring out a later revision but not now.
Best policy, bring product to market and pay for the equipment and time you’ve invested then iterate a better product, perhaps better than the official one if/when Framework delivers one of their own
Personally I’d be willing to pay more for an aesthetically pleasing implementation but not all would and I certainly want this project to get off the ground first
Lots of us rooting for you, Josh, and looking forward to ordering!
I wish we could just buy the xjack ports they used in these cards
https://www.amazon.com/3Com-Megahertz-Card-Xjack-Connector/dp/B00004Z8B8
that may actually be possible to replicate.
as soon as the network expansion card is out I can order my laptop. Thats the only thing im still waiting on.
It may be worth testing what power the card draws. I guess it should be possible to try this as soon as the chipset is chosen, since the casing probably won’t affect the operation. It’s particularly worthwhile to test what power it draws when it is not in operation and when the system is suspended. As it turns out, the HDMI and USB-A do use a noticeable amount of power, even when the system is suspended. That means that for many on-the-road scenarios, using a portable USB-C hub is actually a more practical scenario than having non USB-C expansion cards installed (particularly for wired ethernet, where one will be fiddling with wires anyway; less so for a quick plug of a USB-A stick to transfer a file): unplugging a hub is much easier than prying one of the cards out and leaves the laptop in a better shape for transportation.
RTL8153
-20 characters-
Just had a look, at gigabit and heavy usage it should draw around 400 milliamps and can go into a deep standby but can’t get the current figure.
OK, that looks like a very common chip for USB-C ethernet adaptors. Mine sit in a powered dock or a multi-port hub, so I can’t get reliable data on the power use when suspended.
The experiment would be simple:
Difference of charge divided by suspension time (in hours) should get you the power-consumption while suspended. The difference between the consumption with/without dongle should be the power the dongle draws while suspended.
It would be useful information to have for pretty much any expansion card. Devices that draw quite a bit of power while suspended may be better in dongle/hub/dock format than as an expansion card for people travelling (because then the device naturally unplugs when not in use).
What figures would we be aiming for? It does promote deep suspend internally that will put the RTL8153 in a suspend when no network cable is connected.
I’m not sure there’s anything to aim for and/or if there’s something you can do in the card (other than a switch to break the connections to the plug). The measurable power drain from USB-A and HDMI expansion cards while suspended may well be because some devices lead the hub to not power down completely. An ethernet port will likely be susceptible to that, due to support for “Wake on LAN” etc.
There’s probably a market for the card even if the power drain is measurable – hopefully Window’s hibernate does kill the card completely – but for people running Linux it’s a good data point: weighing increased suspend time (at the price of a dongle) against the convenience of a port. It’ll just allow people to make an informed choice.
If it turns out possible to change the firmware so that the board can actually power off the USB hubs on the main board upon suspend, this will probably all be moot, but as of yet, Framework hasn’t given any indication they can change the situation.
EDIT: Here is a message from @nrp that acknowledges they are aware of (at least) the HDMI issue and that they’re working on it: Linux battery life tuning - #22 by nrp
It will suspend itself, no contact with windows or linux even if its being supplied power it will go to suspend after 10 seconds without a ethernet cable.
Being suspended may not be the same thing as being disconnected from the USB hub. I would have totally expected that the USB-A expansion would draw no power whatsoever if nothing was plugged into the USB-A connector, but that is not the case either.
I’m sure the RTL 8153 does a great job minimizing its power use (which might not be completely 0 if it needs to keep an eye out for an RJ45 getting plugged in), but the power drain might actually happen on the other side of the USB-C connector.
That makes sense now.
