Hello,
Not sure if this is the right venue for this, so please let me know where to send if not. Got my new RT1062 OpenMV cam today with a couple extra camera modules. The stock one and the autofocus one work great, but the Global Shutter Module gives a ‘sensor not found’ error when I power up and connect to the IDE with it attached. Assuming I may have gotten a bad module?
Thanks,
-Adam
Please update the device firmware on the RT1062 camera. It should be found after doing that.
That did it. Thanks! I do seem to have a stuck pixel and I’m getting black bars intermittently. Is that to be expected?
Hi, the Global Shutter camera works better with the H7 Plus. The RT1062 has a higher noise floor than the H7 Plus on it’s 3.3V rail. So, the image will have flickering horizontal bars. I’ve already turned up the row denoising on the module to get the picture as good as possible.
(What’s weird is that the RT1062 has more filtering than the H7 Plus on the analog supply to the sensor pixel power rail. It should technically produce a better image…)
On the H7 Plus you get a perfect image without issue. But, not so on the RT1062. However, if you increase the exposure of the module from the default using sensor.set_exposure() or use the sensor in a brighter environment the issue goes away.
Sorry about this. We have new much better cameras in prototyping right now which will fix these issues. If you can use the H7 Plus for your app or are okay with a longer exposure with the RT1062 both will get you past this issue.
Ok, thanks. That’s a bummer, but I appreciate the info. Want to make use of the on-board wifi, so need the RT1062, but I’ll see if I can make things work with a longer exposure.
Thanks again for the super fast response!
Sorry about this, that said, new global shutter modules with color and more res are around the corner.
I did some debugging on this and ran some tests:
RT1062 VDDA Rail Results:
VRMS = 3.33V
VPP = 90-130mV
VAMP = 100-120mV
VAVG = 3.33V
VAREA = 399mVs - 10ms window
H7 Plus VDDA Rail Results:
VRMS = 3.29V
VPP = 210-230mV
VAMP = 220-240mV
VAVG = 3.29V
VAREA = 394mVs - 10ms window
From these results the RT1062 clearly has better line regulation than the H7 Plus… and yet, has a worse camera image than the H7 Plus.
The only reason I can guess for this is larger global voltage swings area happening on the RT1062.
For the next production batch of the RT1062 we will probably update the design to include a linear regulator on VDDA going to the camera to drop that rail down to 3.0-3.1V. This should resolve the issue. It’s not in spec for the MT9V034 exactly as it’s VDDA is supposed to be 3.3V. But, 3.0-3.1V is pretty close. All other camera modules except the OV7725 drop VDDA down to 2.8V. The OV7725 uses it directly but can work down to 2.8V. Interestingly, the OV7725 with the RT1062 has a great image.
We’re working on some board designs right now I can upstream this change into and should know if it fixes the issue in about two months.
…
I did find that the image quality greatly improved by adding a filter cap to the VDDA rail. If you are not afraid of soldering onto your camera:
Add like a 220uF tantlum cap there: F930J227MBA KYOCERA AVX | Mouser
And the image will improve. However, bright areas will still show some lines in them. Dark areas though will be perfect.
Hi Kwabena,
Man - thank-you for taking the time to dig into this - I really appreciate it. The level of support here is off the charts.
I will definitely give the cap a shot the next time I pull the board out of the assembly.
Thanks!
-Adam
Did some more testing.
Swapping the ferrite bead out for a resistor of like 2.2ohms yields a perfect picture on bright scenes. On darker scenes the image can be seen flickering. Even after increasing the resistor to 10ohms and adding a 47uf tant the issue still happened on darker scenes. This is probably because the RC filter has a variable voltage drop based on the current draw from the camera which is variable.
So, a regulator is definitely needed.
…
I did notice the camera didn’t really care that the voltage was 3.2V or 3.1V. So, it seems fine to lower the VDDA rail slightly.
Probably this chip: AP2202K-ADJTRG1
…
EDIT: 5/7/2024 - Tested if putting the main regulator into forced PWM mode mattered. It did not.
