@Gmanny when I saw this thread it made my day so thanks for posting, but I wanted to add that it is not strictly necessary to file down anything. You could also just exchange the screws, i.e. use the golden normal screw to fix the adapter down (it works fine as it is not a full scale m.2 ssd), and I then opted for screwing the retention screw in where the normal one was to make sure I do not lose it, see picture.
5 Likes
Hey, I’m glad this helped in the first place! And an awesome idea with the screw swapping, I’m editing the top post to include this so that people don’t get discouraged by the necessity of using a file.
Edit: Seems like you can probably even put the screw into the 2260 hole (without breaking off the 2280 portion) on the M.2 PCB for more stability 
2 Likes
I am really glad that we now already have a workable solution regarding performance. I have been looking more for m.2 to oculink connectors and in all honesty simply did not identify one that is angled and could help us more.
What I am, however, wondering is if this one here could be of help. If I understand correctly it would enter the Dual.M.2 expansion bay module upside down which would then fit well with the second adapter pointing in the right direction to be closer to the hole at the back. This leaves a couple of questions.
- would it fit?
- how to secure it if it is upside down?
- is there any 32 pin contacts m.2 adapter that is angled and would be more practical for our use case?
- Would signal integrity suffer from yet another contact (32 pin)?
- It says it’s PCIE 3.0 only, but it should be available at higher speeds?
PS: I am mainly posting this here to see if maybe the smaller part can be of any help in combination with something else, or if it could make the design of a DIY board by the community easier.
And here is another one. If it is actually possible to “fold” it as shown in the picture, then this could be quite interesting too.
2 Likes
Ok, that one is really interesting! Someone who has a FWL16 + Dual M.2 should consider trying that. Or at least maybe map out how that length cable would fit.
Those that already have a different M.2 to oculink board, how well is it working? I’m guessing there is no redriver on the Dual M.2 board. Wondering if it’s really stable with all these connections and routing over multiple PCBs and different wiring. .
If this one would fit, it seems like the PCB would be sticking out of the back of the laptop, which is not ideal, IMO. The best solution would have the port fixed to the chassis of the laptop so that no signaling interconnect receives stress from plugging/unplugging the OcuLink cable.
This one’s pretty cool-looking! I’d get one of those for tests, thank you!
1 Like
Please someone with the Oculink 4i setup do some GPU performance tests! I would really weigh my options, given that the M.2 → Oculink 8i is not in the works by anyone…
I actually managed to fit this adapter in the enclosure, something like this (but ended up using even more capton tape):
I know ~nothing about EE/signals, so I have no idea if this is remotely safe, but it seems to work for me.
- I used own M2 risers to avoid filing, much like Philipp.
- I drilled a hole for an extra M2 riser to fix the port in place.
- I drilled a hole for the Oculink port in the back cover of the expansion bay.
- The expansion bay fit (barely!) inside the chassis . There was a little bit of resistance/friction, but with enough patience it clicked in and nothing seemed to flex too badly.
2 Likes
Yes, completely agree, it was as mentioned mainly to show what is out there.
Just to confirm, is this the 10cm version or the 15cm version?
Yes, I bought it and will send an update too. Regarding signal integrity, I am using the DEG1 from Minisforum, it has a redriver and the connection works like a charm so far, no problems at all. It is also quite mobile and reasonably priced. Finally, with all the DIY things we are doing, I wanted to have a brand EGPU (although I think also cheaper ones will work completely fine).
2 Likes
15 cm. I might try to replace it with the 10 cm in the future.
right. since no one is doing it. I am trying to take the dual m2 to oculink 4i to 8i route. all parts shall arrive tomorrow so wish me luck
4 Likes
Sick! I am this close to finish the R&D phase to make a plug-and-play Framework OEM M.2 → OcuLink 4i conversion kit, including a 3d-printed custom designed I/O cover and a retaining system for the cable.
Stay tuned!
2 Likes
M.2 TO OCULINK USER - BE CAREFUL!
The cable can (and probably will) get stuck when trying to get the Expansion Bay Shell out of the laptop. Make sure to keep it tucked with tape or screws. I had to disassemble the midplate and ventilation port above the keyboard to take it out. Lesson learned.
3 Likes
Out of curiosity, do you turn the DEG1 on via its power button before turning the laptop on?
1 Like
I certainly have to do it before with my setup
1 Like
I have tried to combine the 2 4i singal into 1 8i. it didn’t work.
1 Like
Yes, you should turn it on before (and wait for a couple of seconds before turning on the laptop apparently). Here is a good enough overview of the EGPU with instructions of setup (starting at minute 15).
Really good to see that the version on the right bends so well. Still waiting for it to arrive from China and will then post a picture as well. Final step will be 3D printing a black plastic cover.
Nice! I just mounted mine with one screw, it seems to be fine for now. Eventually I want a real Oculink board when they come out, until then … I’m on about 32hrs of playing Satisfactory on my GPU (an old PSU and a cheap Oculink to PCIe PCB.) and it works just fine.
3 Likes
EDIT: Did anyone here have success with AMD eGPUs? With my 6900XT the games crash as soon as I enter fullscreen. Windowed mode works fine tho.
Same behaviour on Adrenaline 25.1.1, 24.10.1 and 24.12.1.
Using Minisforum DEG1 and one of those flexible OcuLink → M.2 adapter. I’m using the Framework LCD.