OAK-D S2 PoE can't operate strobe

Muu

Hi Luxonis team,
I’m trying to get camera strobe pulses from the M8 connector on an OAK-D S2 PoE FF, but I cannot see any camera-FPS pulses on M8 pin 6.

Device:
- Model: OAK-D-S2-POE-FF
- Board: NG9097
- MXID: 19443010F13D7B5A00
- Connection: PoE Ethernet, IP 169.254.1.222
- DepthAI tested: 3.7.1 and 2.29.0.0

Wiring / measurement:
- Analyzer signal/input is connected to M8 pin 6, labeled `Strobe ISO`.
- Analyzer ground/reference is connected to M8 pin 7, isolated GND.
- With pin 6 and pin 7 connected directly to the analyzer, the strobe line appears to pull the analyzer input low.
- I do not see camera-FPS strobe pulses on pin 6.
- Also the M8 pin 5 is labeled `5V`, but I measure only about 0.012 V on pin 5 while the camera is powered by PoE.

Code I tried:

python
import depthai as dai

pipeline = dai.Pipeline()

mono = pipeline.create(dai.node.MonoCamera)
mono.setBoardSocket(dai.CameraBoardSocket.CAM_B)
mono.setResolution(dai.MonoCameraProperties.SensorResolution.THE_800_P)
mono.setFps(30)
mono.initialControl.setManualExposure(1000, 400)
mono.initialControl.setStrobeSensor(1)

xout = pipeline.create(dai.node.XLinkOut)
xout.setStreamName("left")
mono.out.link(xout.input)

with dai.Device(pipeline, dai.DeviceInfo("169.254.1.222")) as device:
q = device.getOutputQueue("left", maxSize=4, blocking=False)
while True:
q.tryGet()

Result:
- Same pipeline without `setStrobeSensor(1)` streams normally.
- With `setStrobeSensor(1)`, the device crashes/disconnects.
- Crash dump shows:

RTEMS_FATAL_SOURCE_ASSERT
DrvGpio.c
DrvGpioSetPin
line 326

Earlier logs also showed:

Strobe enable for cam 1: gpio -1, active-level 1

Main question:
Why can’t I see strobe pulses on M8 pin 6 with this setup?

Specific questions:
1. Is `setStrobeSensor()` the correct API for generating camera strobe pulses on M8 pin 6?
2. If yes, why does it resolve to `gpio -1` and crash the firmware?
3. Since pin 6 is an isolated open-collector output, what exact pull-up wiring is required to see pulses?
4. Should the pull-up use M8 pin 5, M8 pin 1, or an external supply?
5. Should M8 pin 5 output 5V when the device is powered by PoE? If yes, why might I measure only 0.012 V?
6. Does pin 6 pulling the analyzer input low indicate that the open-collector output is present but not being driven correctly?

Oskar-Luxonis

Hey Muu - setStrobeSensor() is the intended API for enabling the sensor strobe signal. The argument is the active level/polarity, so setStrobeSensor(1) means active-high strobe.

For the M8 Strobe ISO output, note that this is an isolated open-collector output, so it will not generate a voltage by itself. It needs an external pull-up/reference on the isolated side in order to see pulses with a logic analyzer.

That said, the crash you are seeing is not expected. We were able to reproduce a similar crash on RVC2 with DepthAI 3.7.1: the strobe path resolves to gpio -1 and then firmware asserts in DrvGpioSetPin. I’ve reported this to the development team for investigation.

I’ll follow up once we have confirmation on the firmware side and the exact recommended wiring for the M8 strobe output.

Best,
Oskar

Muu

Hi Oskar,

Thanks for the quick reply and for reproducing the issue on your side. I'm looking forward to the dev teams findings and the recommended wiring.

I also wanted to ask about the 5 V output on M8 pin 5, since that wasn't addressed yet. Should pin 5 provide 5 V when the device is powered over PoE, or is it expected to remain unpowered in this configuration? I only measure about 0.012 V there.

Thanks again!

Oskar-Luxonis

Muu - thanks for the follow-up. On M8 pin 5, this pin is VBUSIN / 5V and is intended for sourcing 5 V power to accessories on the M8 connector. The M8 pinout/diagram is here: Connectors M8, M12, and the OAK-D S2 PoE product page shows the same pinout: OAK-D S2 PoE.

So measuring only 0.012 V on pin 5 while powered over PoE does not sound like the expected accessory 5 V output, could you check again? Check the diagram img, if you mesured the right pin, which pin did you measure this againts?

For the setStrobeSensor() crash: we reproduced it and reported it to the development team. They confirmed this needs further investigation on the firmware side. I’ll update you once we have their findings and the recommended wiring/usage path.