That ‘Multiple DTBs found’ error is a bit unusual. It could mean leftover DTB files in /boot from previous install attempts are confusing the detection script. Can you post the full output from the installer? There might be more clues in the complete log about what’s failing and when.
I’ve uploaded the full installer log — I’ll PM you the details. But here’s the key sequence: the .deb downloads and extracts correctly, then during the post-install phase it hits the DTB detection error. The script seems to be looking for a specific device-tree naming pattern and the ‘Super’ variant doesn’t match what it expects.
The arducam_csi2.ko module never actually gets built or placed in /boot/arducam/, so that mv failure is expected — the file was never created in the first place.
0V on the power rail plus no I2C response narrows it down — either the power delivery isn’t being configured, or there’s a physical connection break. Let’s check two more things to see if the device-tree overlay is even taking effect:
Check if camera power regulators are being registered:
/sys/kernel/debug/regulator/regulator_summary — no entries with ‘cam’ at all. No camera regulators anywhere in the output.
/proc/device-tree/ — no camera nodes present either.
So the device-tree overlay definitely isn’t being applied. That explains the 0V — the power rails for the camera are never getting enabled because the kernel doesn’t know about the camera.
Just finished going through all the physical combinations:
Both CSI ports (A and C): same result, 0V and -121
Both cable orientations at both ends of the ribbon: no change
Two different ribbon cables: no difference
Also tried swapping which end of the cable goes to the board vs the camera module — no luck. I’m fairly convinced this isn’t a cable or port issue at this point.
Okay, let’s try a different approach on the software side. The Arducam install_full.sh script clearly isn’t handling the Orin Nano Super variant’s DTB naming convention correctly. NVIDIA’s own jetson-io tool should work better here:
sudo /opt/nvidia/jetson-io/jetson-io.py
In the menu, select ‘Configure Jetson for compatible hardware’ and try the IMX477 overlays. There should be IMX477-A for CSI port A, IMX477-C for CSI port C, and possibly a Dual configuration. Apply one, reboot, and run i2cdetect. Try each overlay and let us know what you get.
Then use the resulting DTB and run i2cdetect again. This will tell us whether the overlay itself merges cleanly for this board variant or if there’s a dtbo compatibility issue.
Did the manual fdtoverlay merge. It completes without errors and produces the merged DTB file fine. But after booting with the resulting DTB — still getting -121 on i2cdetect on both buses. No difference from any of the other attempts.
I think I’ve identified what’s going on here. There are actually two separate issues at play:
Physical issue — the cable orientation on the camera module side: I’ve seen this many times with the Arducam IMX477 module. The MIPI ribbon cable on the camera module side is very likely inserted upside down. The orientation can be counterintuitive — the contacts need to face a specific direction and it’s easy to get wrong. When it’s flipped:
The power lines don’t make contact → 0V reading at the CSI connector
The I2C data lines are open → error -121 on every probe
The sensor is completely electrically invisible
This is why no amount of DTB tweaking, overlay merging, or port swapping made any difference — the sensor was never getting power in the first place.
Software issue — installer incompatibility with the Super variant: The Arducam install_full.sh script doesn’t recognize the device-tree naming convention used by the Jetson Orin Nano ‘Super’ variant. Its DTB detection logic hits multiple matching patterns and throws the ‘Multiple DTBs found’ error. This is why jetson-io is the correct approach for this board on JetPack 6.
For the physical fix: Power down the board completely. Unlatch the CSI connector on the Arducam IMX477 module itself, flip the ribbon cable to the correct orientation, and reseat it firmly. Make sure it’s fully seated before latching.
That was 100% it! I flipped the ribbon cable on the camera module side and now I’m reading proper voltage at the 3V3 pin. Can’t believe it was the orientation — it looked correct to me. So now that the hardware is sorted, what’s the recommended software setup? Should I go back to jetson-io?
One more thing — since the Arducam installer failed multiple times, you might have leftover DTB files in /boot/ from those attempts. Worth cleaning those up to avoid confusion down the line. The ‘Multiple DTBs found’ error from the installer is a known limitation with the Super variant — stick with jetson-io and you’ll be fine.