I wonder if I missed a decimal point… lol!
This thread sure grew fast overnight! I’ll read later. Thanks for a detailed reply Platos, Sui, probably others.
Edit: oh gbit vs gbytes! That old thing. That would be where I wandered off.
Sounds exciting. I like the dual cable idea BTW, from a broader compatibility view. If it works.
As far as I know, thunderbolts protocol allows it to tunnel PCIe signals. The drop in bandwidth between the Thunderbolt and PCIe speed comes from the overhead of this tunneling. And they also need to hit one of the specs (i.e. 2x 4x or 8x). I’d assume thunderbolt 5 (and usb 4 for that matter), will get support for a PCIe 4.0 x4 connection via their respective 80Gbit/s modes for an effective bandwidth of ~64Gbit/s.
One thing to consider, newer Framework motherboards could launch with an expansion bay connector that supports a higher PCIe gen. At least if they don’t run into signal integrity problems.
All in all a great write up 
Originally it was of course not meant seriously, why “harress” a company. But well, if we are seriously talking about it, then I would assume it depends on how cool a company is. For example, in these threads I am definitely trying to push for things I hope for, given staff seems to actually spend time here ;).
Of course it’s all relative. In DIY circles I’m in, having a soldering iron and ordering PCB files is relatively common.
I’m in the same boat as you regarding design, however the design only needs to be made once. In theory connecting the traces is easy, and I’d probably get it working somehow (with proper impedance and length). As for passive components to improve the signal integrity across connectors - I’d have to rely on existing designs for that (and those do exist usually). But I’m also not comfortable in a PCB designer.
But the resulting board is of a limited complexity in that you could get them in one of two ways:
One person solders some in addition to theirs, since ordering more parts is increasingly cheap, and sells them in the dozens at most. I’d be up to do that, at least the 4 additional boards (MOQ of PCBs is usually 5).
Alternatively, you go and use one of the mentioned PCB services, and if you throw enough money at the problem, they will put the components on for you - Oculink 8i port is for sure going to cost extra since that is a non-standard component they’d have to get, but otherwise, it’d be fine. You could sell the remaining few boards you got at a small profit.
That’s why I said the opening of the shell is actually the most difficult. I’m not comfortable enough with PCB design that I’d rush out and do it months before I get one (and then some months before I’d get an external GPU / PSU combo to connect) when I truly do believe somebody experienced with PCB design could come along and do it.
But yeah, these options, while realistic, are still not going to be easy, you’re right. Would love a Framework santioned way that doesn’t involve a dremel (I don’t have a clean way to open a hole in an aluminium (or magnesium alloy) shell)
EDIT: Well, scratch that. In LTTs original FW16 video, it is clear that not only is the back plastic (duh, forgot), but there seems to be a plastic slit that can be conceivably removed (and replaced with 3D printed custom port covers). So no dremel necessary. Makes selling to consumers also much easier, just a PCB and 3D printed cover. Nice going!
Yeah i do also think, thunderbolt 5 comes with 64GBit/s for PCI-E 4.0. But it is still no “nativ” PCI-E and for now it seems thunderbolt is not good in handling it. So im not really sure, if they really can fix that.
And still oculink 8i would (already now) still offer double bandwith of that.
But as i said, im not sure if it would just be better to make a direct PCI-E 4.0 x16 Port (with 8 electrical lanes). So i don’t know, what on a engineering side is easier. But probably it would also be more difficult to handle for consumer with these short and wide riser cable. On the other side it would maybe be a better signal, but i have no clue.
But i really don’t think framework will come up with PCI-E 5.0 in the next 2 years. Or do you think so?
Only because you can build your laptop (diy edition) and change parts of it, does not mean, you can design your own pcb. So i would say 90% of framework community (or more) can not be design a PCB nor would they ever be able to.
So that’s why people ask for official support.
And to be said: Not everyone lives in a country, where you have access to some of these PCB-makers (or it will be a expensiv shipping cost). And also it will not be easy to sell several PCBs, because you have to order a lot of pieces. Not all lives in a country, where there are a lot of Framework users or they live in countries, where shipping-costs are high.
So what you describe is maybe “easy” for people, who are building their own PCB, but thats far from normal. It’s like the difference between playing lego and producing your own lego blocks. Only because you can assemble parts, does not mean you can made your own parts.
I personally think designing a pcb is way more difficult for an average person, than drilling a little hole somewhere.
But let’s go back to topic. If you have a real plan to design a PCB, share it, but just to say "it’s actually quite easy " does not help anyone.
Again, designing it only needs to be done by ONE person, once. Worst case, I could do it, might take a couple days reading through PCI-E spec to find the right reference design for reducing signal degradation, but I would be able to get it done somehow. And I’ve only done some very basic designs before.
Next stage, “building” one. I already told you anybody with enough money could get it done using external manufacturers. More reasonable if somebody would take lead and sell them in low quantities, there’s a lot more people with a soldering iron in these communities than you migth expect - though soldering ports with fine legs isn’t something everyone is comfortable with, but worst case you have 5 PCBs minimum to train on:).
Finally, installing, which until now I thought required a dremel, but that doesn’t seem to be the case, again refer to my previous post.
So I find it decently realistic we can get a basic version going eventually, so any normal person can order one from another friendly DIY enthusiast, with little effort to installing it.
Shouldn’t surprise you, that is how the default shell has been designed after all, for people to replace the internals with something actually useful. Wouldn’t be surprised at all if they start selling some “official” modules and you get some PCBs and plastic port plate to install into the shell you already have (or can get separately). Why spend 110€ on another shell (yes, that is the price of the shell /w cooling system) when you can just get a PCB+plastic cover.
On the country: Fair enough. I’m not sure if chinese PCB makers have specific restrictions, but one of them ought to ship to you. If you think shipping is high, they are for everyone. That’s just one way they can advertise low costs and then charge you a lot anyway. Worst case, find somebody who ordered and build their own, and ask them to ship to you, as I said in my second post.
Again, I’m sorry for offending anyone by saying it’s “pretty easy” - I was saying that in the context off how difficult making an own damn external-GPU module sounds. Compared to that, it is surprisingly easy, and I stand by that. Designing PCBs is not magic, and this one has very little passive components (and MAY, just may, even work without them, though I wouldn’t count on that). That doesn’t mean I expect anyone to do it, or for you to do it, but I believe if you are not comfortable making it yourself, there is a decent chance you will be able to get a community made one eventually anyway.
Hot-plate reflow baby! I don’t want to do any fine pitch by soldering iron if it’s avoidable. Reflow is so much easier, imo. Toaster-oven reflow if you need components on both sides.
I don’t really know how PCIe tunneling is implemented so I can’t really comment on anything there :c
If it can be done, all depends on the signal integrity. Going from PCIe 3.0 to 4.0 or even 5.0 made keeping the signal in line with the spec way harder. I don’t know how much noise their interposer adds. If it is still fine for 5.0, I don’t see why they wouldn’t offer that down the road. PCIe is fully backward compatible, so they wouldn’t lose support on any of their upgrade modules.
But the big question here is how much the signal can handle and for that we’d need to ask an engineer that has played around with it.
yeah, lets hope framework does offer somehow an oculink 8i in future.
Sure, community-wise would also be nice, but as long as you can’t buy something somewhere (doesn’t matter if it comes from framework or community), it will not be usable for most people out there, who can’t do it by themself or “hope” for someone, who have 3 or 4 units “left”.
So at the end, someone “has” to do it, if it should be available to the majority of the users.
Continuing in this thread for development of a simple Oculink adapter card (might be dual Oculink 4i, not 8i, since I can’t find the pinout for the 8i)
Ok nice, let’s see if someone will have any idea how to develope that. Would help i if i can, but im probably useless in this field of topic.
Well, we need to wait for the expansion bay deep dive, and then let’s see. Hopes to see something on Etsy or FW marketplace eventually are high. 
Update from the other thread, there’s now one design with OCuLink 8i + DP by @Josh_Cook and one dual OCuLink 4i + DP by me.
There’s plenty space left on the board both for a battery (though not too big) and a charge controller on the left - routing would be pretty easy. If anyone of you has experience with that, feel free to add it, even if someone wouldn’t need it and wouldn’t populate it, it is a no-brainer to have IMO. Doesn’t add cost to PCB manufacturing after all.
Speaking of cost, if you want to order yourself, on jlcpcb it costs 34€ for a budget thin version and 49€ for the full size card + shipping (~10€ for me) for 5 PCBs, and it scales really nicely (e.g. 63€ for 20 full size cards). Connector costs are a bit higher at 10-12€ + shipping if you buy individually. Plus charge controller costs.
I expect some people to build and resell the boards, so don’t worry if you can’t reflow solder, and there’s also the 8i board by Josh which you might be able to buy
Also don’t forget to account for the costs of a dual 4i to 8i cable (~55€) and a 8i to x16 PCIe Slot board (~55€) - pretty expensive. If you go for an equivalent 4i setup, that’d be much cheaper at 40€ with combos made for the GPD win series. Win for us!
Nice work from you both!!
Can’t look at this thread without smiling a bit, looking forward to the oculink port! So, I have been looking for oculink EGPU options and saw this comment on egpu.io:
“Pretty much OSMETA GK01, NFHK N-P114-A and an upcoming JHH-Link OCuLink eGPU adapters are the most readily available OCuLink eGPU adapters.” (see: https://egpu.io/forums/which-gear-should-i-buy/need-help-with-oculink/). And I think there is a new One Dock coming as well.
I am looking for an EGPU that definitely provides the full speed of oculink obviously and can support high-end desktop cards. Any advice on which one to get and which one would be best compatible with what is being built in this thread?
I personally have only interests in a 8i dock. But it’s always good to see what’s around.
But where did you hear sth about a new One Dock? Isn’t it just a new batch ?
Yes, you are right, I called it “new” as it if I understand correctly has a “new” feature which is to support a PSU directly. But yes, it is a new batch, see here: https://egpu.io/forums/custom-egpu-chassis/one-dock-unique-dual-port-egpu-dock-thunderboltoculinkm-2-0-26l-2/paged/12/
Ah ok, didn’t know that. Then i would say it’s like a Version 2 of it.
Sadly it does not support 8i Oculink.
This leaves me for now with the plan to buy OSMETA and hopefully an oculink adapter for FW16 in late Q4, but hoping someone will maybe design a dock in this community as well by around then.
Josh, who also did the 8i design, expressed he planned to make an 8i dock way back. I believe he will be able to sell a bundle for a decently good price. Only downside, it’s not open source (yet?), so no extras possible unless he does it. For the dock, that’s fine ofc, doubt I’d want anything besides a standard 8i dock implementation that takes standard power supply input to support the 75W PCIe slot power
