Difference in event rate between 2 DAVIS 346 camera

[barbiert...@gmail.com]

Hello,

We have 2 DAVIS camera, id: 0270 and 0335 that we bought a year apart aproximately.

When recording with both cameras active, and with the exact same biases, I got a difference in the event rate between the two.

The older one records at higher event rate than the new one (the difference can be up to 10  times more events per second).

The test was made with both camera next to each other, looking at the same scene, so the illumination characteristics should be the same.

I am not sure why this is happening. I am guessing I could change the bias to get similar event rates for both cameras, but an insight on the reason it might happen would be nice.

Thanks in advance,

Thomas

[Tobi Delbruck (UZH-ETH)]

It is probably mismatch in the bias generators.  See

Z. Yu and T. Delbruck, “Self Calibration of Wide Dynamic Range Bias Current Generators,” in 2020 IEEE International Symposium on Circuits and Systems (ISCAS), 2020, pp. 1–5, doi: 10.1109/ISCAS45731.2020.9180623 [Online]. Available: http://dx.doi.org/10.1109/ISCAS45731.2020.9180623

It would be great if you could share these two recordings.

[Tobi Delbruck (UZH-ETH)]

The other possibility is that the difference is caused by long tail outliers in the hot pixel distribution. We made recent measurements of background activity (BA) leak events that show a significant fraction of BA activity under bright conditions is due to hot pixels. Example histogram of event rates looking a blank white area is below (it is part of a paper submission). (Not sure if this inline PNG will post)

[barbiert...@gmail.com]

Thank you for your responses.

After some investigation, I found out that the problem was due to the camera temperature. The older camera (0270) when active for a certain time starts to get quite hot, which slowly increases the event rate. On the contrary, the new camera (0335) does not seem affected by the same problem.

Could it be a defect on the camera ?

Thanks in advance.

[Luca Longinotti]

Yes, higher temperature will give higher noise, this has been
characterized well in:

Y. Nozaki and T. Delbruck, Temperature and parasitic photocurrent
effects in Dynamic Vision Sensors
https://ieeexplore.ieee.org/document/7962235

What kind of difference are you seeing between your devices? You can
try looking at the IMU visualizer, it shows temperature too, if you
hover over the visualizer info bar (top of the visualizer).
Normally between 40 and 50 °C should be fine.

--
Luca Longinotti (llongi)

Software Engineer
iniVation AG - https://inivation.com/
Zurich, Switzerland
Office: +41 44 500 3214

Support - https://inivation.com/support/

[barbiert...@gmail.com]

I measured the temperature for both cameras for a while. The camera were static and connected to the DV-software. Here are the results:

 2 min:
 0270 = 37°C
 0335 = 30°C
 
 5min:
 0270 = 44°C
 0335 = 32°C
 
 10 min:
 0270 = 50°C
 0335 = 34°C
 
 15 min:
 0270 = 53°C
 0335 = 35°C
 
 20 min:
 0270 = 55°C
 0335 = 36°C
 
 30 min:
 0270 = 57°C
 0335 = 36°C

For a static scene with neon lighting, I have a final event rate (after 30 minutes) of 3,7 million events per second for the 0270, and only 100 thousand events per second on the 0335.

[Luca Longinotti]

The results you see for camera 0335 are normal in our experience.
0270 definitely gets hotter, which will generate more noise.
The camera itself still works and all components easily work at that
temperature, but the extra noise isn't optimal.
If the two cameras are both DAVIS346 with the red metal case, please
contact sup...@inivation.com to discuss a replacement, as the camera
is outside our QA parameters.
Thanks and have a great week!