FW support replied quickly but stated that they could not help. I am in the midst of adding more test pads to the board to help debug further. Will order sometime this week
Boards should be here today
Boards arrived. I will test them today
I got the 100MHz REFCLK signal to appear however the GPU still does not appear in device manager. I remember someone mentioning that they had to disable Quick boot to get the Oculink board working. Does this seem familiar to anyone?
Yes, this is something I have shared with some folks (I did it on my laptop) and others have had success too after doing so.
I’m facepalming so hard right now. I swapped the RX and TX lanes on the interposer footprint somehow…
Oh no
am I interpreting that as you have to reverse the design and print another board because you had it backwards? If that is the case, you aren’t alone. I remember reading someone else’s post where they had done something similar due to that thorn in the side interposer. I can easily see how that can get mixed up.
Unfortunately i do have to order another board. But because my board is a 2 piece (one being significantly smaller than the other) i only have to redesign the smaller board meaning it will be cheaper than a full redesign
At least there is a way to see some positivity in the situation. You’ll get it squared away, you’re almost there ![]()
HUZZA!
Im still installing the drivers but was too exited not to post an update. I will post more details once i get them installed and run some tests
After installing drivers i rebooted my laptop. I pluged in the 4070 Ti Super (My desktop GPU that i daily drive) into a monitor and when the laptop displayed the memory test screen, it shows up on the monitor however once the memory test is over with, i no longer get any display output from the GPU (Laptop display works fine). I am able to use the GPU as a display output in the BIOS as well.
Additionally for some reason the GPU is visible in device manager under display adapters however it is not visible in task manager. Running GPU-Z i see that the GPU is seen however no bios version is found and a bunch of boxes are blank/unknown
(Ignore the x4 lanes, that is due to a poor solder connection)
I am unsure of how to further diagnose this.
unrelated to the weird half-ghost detection issues, but in the final design would it be dangerous to have the remaining power budget of the expansion slot avaible as two massive solder-able pads or it would be a dangerous and cause more complexity than neccessary as a small feature to add? i had the idea that since the expansion slot can provide up to 220w (if i’m not mistaken) and pcie slot only consumes up to 75w, using the renaining watts with a dc-dc converter to power a single pcie 8pin or something else.
That could be possible but the x16 connector takes up most of the rear opening leaving very little room to run wires or a connector
additionally the output of the laptop varies from ~15v to 20V so you would need another converter to make it 12V
yeah i did tought about a dc-dc converter between the expansion bay remaining power output and whatever i’m trying to power, but that can be a separate board and not something built into the pcie adapter itself as it would add complexity for a usecase that not everyone would need, possibly a warning printed onto the silkscreen of board that states the voltage range of this power output with a max wattage (since 220w are the truly maximum it can deliver, and 75w are already in use by the pcie slot + whatever overhead due to innefficiency, something along the lines of 110~watts max (estimating 80% efficiency of things that turn power into heat))
tldr just a couple solder-able big pads that can give access to the remaining power with a warning of max estimated available power output and the voltage range of the pads
I tested a 3.2 dual usb-c pcie riser and it works fine and i can connect devices to it and they function normally. I’m really stumped. I did a clean install after using DDU to uninstall all graphics drivers and still the same issue.
I also tried Ubuntu but still similar issue. Seems that the OS (windows or linux) sees the graphics card but the driver fails to initialize it (no errror 43)
do any other pcie device work via the adapter or nothing works? maybe its a signaling issue that cant handle the connection or its a conflict due to the gpu spoofing to enable the pcie and power (the microchip that makes the laptop think its a 770s, at least i remember your pcie adapter works that way)
these are my hypothesis not sure if they’re sound
I have gotten a PCIE to USB-C 3.2 hub working but when i attempt to connect a PCIE-M.2 Drive it fails. This points towards bad signal integrity.
I am working on another revision right now and will probably order them this weekend
Just finished redesigning the board. I added some circuitry so that the DC-DC converters are only enabled if a card is plugged into the pcie slot (to save some power)

