I’m sorry, but that command would only connect while the bootloader is running, which is almost impossible to manually connect to. Should have told you this below instead…
Can you run through the IDE and just click the more details button when you get the timeout error? The IDE captures this information itself.
The detailed text in the error message box explains the error. If you could give me that text I can suggest something.
Just to verify:
If so, this is not a timing issue, it’s a permissions issue. The application is being blocked from accessing the HID port.
Given that the camera is in a static state and the command is failing, you can copy the command line argument from the IDE and try to run it yourself on the command line. This should fail, but it would be helpful if you could verify if running the command with sudo fixes it. This should be the case as the way to get past this issue is generally just to run the IDE in admin mode for this update if your PC’s policy is to block USB HID access by programs. If using sudo doesn’t fix this then something else is blocking access…
…
Note, I can confirm on my Mac that I do not need to use admin privileges to run this command though.
As expected, running the command line argument from the IDE as sudo also fails.
Maybe the SBL Mode is also a problem. I execute steps 1. and 2. like you describe it. At step 3. I run the IDE, click connect but then there is no difference in connecting it without connecting 3.3V to SBL. I have run it both ways and it doesn’t seem to make any difference. Same pins as in @jjrzov’s picture at the top.
Hmm,
I’m not sure what to do.
You can verify the camera is alive by running this command while it’s plugged into the PC in SBL mode:
ioreg -p IOUSB -w0 -l
And you should see a device with:
idProduct = 309 (0x0135), idVendor = 8137 (0x1fc9)
This is the serial flash bootloader burned into the ROM of the chip sitting on the USB bus. When you try to run the sdphost tool using the same command that the IDE tries this is running the NXP tool to try to flash the system with a first stage bootloader…
If you see it sitting on the bus with the above command, which is what the IDE runs in the background to find the device and display the little USB logo, then there should be no issue… but, you’ve tried this with sudo and it still fails.
…
Doing a google search: mac usb hid pheripheral cannot open device permission sdphost - Google Search… it sounds like this is definitely a security issue.
I found this post: [SOLVED] HID permissions in OSX - Questions - scsynth - It could be your Mac is blocking HID access thinking the application is trying to monitor keystrokes.
Here’s another thread: macOS IOHIDManager Permission Issue | John’s Blog (nachtimwald.com)
Unfortunately, NXP implemented their bootloader using USB HID versus a USB CDC port. So, you have to disable whatever is blocking HID access. Or use another computer to update the firmware.
That was it! 
Thank you so much for your fast and extensive help. I really appreciate.
Turns out it is indeed necessary to allow the IDE to monitor keystrokes. When it is enabled, the board updates as it should even without the SBL mode.
/OT
Is it possible that you improved the implementation for the global shutter module? It feels like the line artefacts and flickering have improved significantly. 
Yes, I turned on a row noise feature that removed some noise. It’s not an 100% fix though.
Note, the RT1062 electrical design for the global shutter sensor (and only this sensor) has a problem. It will be fixed in Rev5 of the board which will go on sale once we sell out of Rev4. On the H7 we have a linear regulator that takes 3.3V in and outputs 3.3V. You’d think it was doing nothing. But, it turns out that it is necessary to reduce the noise on the VDDA line for the sensor. Rev4 of the RT1062 doesn’t have this LDO. It will be added back in Rev5. Also, the design will be improved by adding the LDO to be powered by VIN so that it generally is actually working.
If you want a perfect image use the H7 Plus with the Global Shutter sensor.
Awesome 
That’s good to know. Image quality isn’t the top priority for me, I prefer the power of the RT1062, but it’s a welcomed improvement for the canny edge detection.
I really like your work! It is pretty much exactly what I was looking for, while still being accessible as a student. Thank you for that! 