Hey, hoping someone here can help me out. I’ve got three Arducam cameras — IMX519 AF, NOIR, and NOIR-mini. All three work perfectly when I plug them straight into the Raspberry Pi. But the moment I run them through my 15m cable extension hat, they all fail. So it really looks like the problem is the extension hat itself and not the cameras. My main use case is video capture (I’m only taking videos, not stills). Anyone seen this before?
Hi, sorry to hear you’re running into this. Can you give us a few details so we can narrow it down? Which Pi model(s) you’re testing on, the exact camera modules you have, and how the extension hat is wired up. Also, what commands are you running and what error output do you get? If you can grab the libcamera-still / libcamera-vid logs, that would really help.
Sure. The cameras are the IMX519 AF, NOIR and NOIR-mini — part numbers B0386, B0389 and U6248. I’ve tested on both a Pi 3B and a Pi 4B. The accessory is the 15m cable extension hat. One thing I did notice: the hat requires pin 3 (SDA) and pin 5 (SCL). The failure happens on all three cameras as soon as they go through the hat.
Thanks, that’s useful. Two things worth checking: first, is I2C actually enabled on the Pi? The hat uses pin 3 (SDA) and pin 5 (SCL), and on a lot of Pis I2C isn’t enabled by default, so that alone could explain it. Second, can you paste the exact libcamera-vid command you’re using, plus the console output from when it fails through the hat?
Good call — I enabled I2C in raspi-config already, but there’s no difference, it still fails exactly the same way. Here are the libcamera-vid parameters I’m using, tested with the NOIR camera. I uploaded the test file here: https://drive.google.com/file/d/1wh2Dg9MNiy9c2cs1zd2sqKL4pQJwumAF/view?usp=drive_web — let me know if you need anything else from me.
Thanks for the file and for confirming the I2C step — and sorry for the slow reply. I’ve gone through this and we’re now tracking it as an IMX519 low-speed driver issue. It’s the low-speed driver path that’s misbehaving when the camera goes through the extension hat, which lines up with why the cameras work fine when connected directly. We’re working on a fix for the IMX519 low-speed driver.
Thanks for the update. Do you have any kind of timeframe for when the IMX519 low-speed driver (libcamera) fix will land? My suspicion is that libcamera 0.0.10 broke something in the IMX519 low-speed kernel driver — I don’t see anyone else on the forum hitting this, which is what makes me think it’s version-specific. Camera extension functionality is needed soon, so if possible I’d really appreciate guidance on installing the older working driver (libcamera 0.0.8 or 0.0.9) in the meantime. I’ve got the IMX519 with the low-speed kernel driver, and 0.0.8/0.0.9 reportedly work.
Understood. We’ll try to get the fix finished this week. We don’t have installation/downgrade steps to hand out yet, but I’ll update you as soon as it’s done. One thing that would really help us: can you tell us exactly what functionality broke on 0.0.10? Sometimes you’re the best software QA engineer we’ve got — the more specific the symptom, the faster we can pin it down.
Quick recap of where this stands: the extension-hat failures on the IMX519 AF / NOIR / NOIR-mini are being tracked as an issue in the IMX519 low-speed driver, and the fix is in progress with a target of this week. No installation steps or downgrade instructions just yet — we’ll post them here once the fix is ready. If you can share what specifically broke on libcamera 0.0.10, that will help a lot. — Dion, Arducam Support