How to setup a VC MIPI MODULE on a Raspberry PI Compute IO Board 3

1 Hardware Setup

1.1 Hardware Pre-Check: Install Raspbian

First step is to install Raspberry Pi OS from

Operating system images – Raspberry Pi

The driver is known to work with the kernel version 5.10 and 5.15, so download the appropriate Raspberry Pi OS version. Raspberry Pi OS Buster Lite is sufficient, and this guide expects this version to be installed not only for the framebuffer output handling.

For more installation instructions see the Raspberry Pi OS Installation Manual; the procedure depends on the platform type where the OS is going to be installed.

Open

(Original site may look different) Install Raspberry Pi OS by following the instructions provided there

The display shows a login prompt after successful installation. If this is not the case, you have to check your Raspberry Pi OS installation. The most relevant information to succeed can be found at the Raspberry Pi OS website or on the web.

Open

Raspberry Pi OS showing login screen (user is usually pi with password raspberry)

1.2 Connect the MIPI module

The ends of the MIPI module connector cable is marked with the hardware to connect to. Open the socket connectors first by raising their lid, insert the cable and press their lid back when mounted correctly. You should then not be able to pull the cable out.

Open

Open the MIPI module socket, put in the cable, close the MIPI module socket

Open

Harmfully angled (left) and good (right) cable fixation

The socket type is also not protected against wrong orientation, so compare your setup to the figures below before switching the power on.

Open

Watch the orientation of the cable (left: bad, right: good)

Connect the MIPI sensor to the socket named CAM1.

Open

Connect the cable to the CAM1 socket and watch out for the cable orientation

Open

At the J6 connector: I²C cable bridge to the CAM1 socket.

The I²C bus named VC is used for sensor communication and must be connected by a cable bridge to the CAM1 socket's I²C bus pins.

Therefore connect the

• pin named 44 with the pin named CD1_SDA as well as the

• pin named 45 with the pin named CD1_SCL

at the J6 GPIO connector.

Reconnect the other peripherials to the Raspberry Pi.

Open

Connection setup for the first image acquisition test.

You should have the login prompt back after switching the system on.

Open

Raspbian showing login screen (user is usually pi with password raspberry)

1.3 Optional: Trigger input (if supported)

Open

Additional cables at trigger input

If the MIPI sensor supports it, you can trigger it externally for an image capture. To do so, connect - for example - a frequency generator between CAM1_IO0 and GND.

Open

Default: Trigger input at rising edge.

To trigger the sensor,

1. choose an appropriate sensor mode which supports external triggering (see the driver's setmode), and

2. give a +3.3V impulse on the CAM1_IO0 pin

Open

Connection setup with an external trigger source.

1.4 Optional: Connect a second MIPI module

We assume that the same type of sensor is connected to the other port, but this is no limitation. You may connect a different sensor to the other socket.

1.4.1 Additional cable setup

The second sensor will be connected to the socket named CAM0. Its corresponding I²C pins at the J6 GPIO connector are CD0_SDA and CD0_SCL. We will use the second I²C bus i2c_arm to connect to the second sensor. Its data lane is at pin 2 while its clock signal resides at pin 3 of the J5 GPIO connector.

Therefore connect the J5 GPIO connector's

• pin named 2 with the pin named CD0_SDA as well as the

• pin named 3 with the pin named CD0_SCL

at the J6 GPIO connector.

Open

I²C cable bridge to the CAM1 socket.

Open

Connected second sensor to CAM0

2 Software Setup

2.1 Get the driver and demo code

You can download the driver and demo code from the following links.

• Driver: vc-mipi-driver-bcm2835-dkms_0.2.6_all.zip

• Demo code: vcmipidemo_0.7.0.zip

2.2 Install necessary Raspberry Pi OS packages

Before beginning with the installation, do the following steps first. This requires your Raspberry PI to already have an internet connection; otherwise you have to install the packages mentioned manually, search the web for the procedure needed.

1. Update the raspberrypi-kernel package and your system by calling:

   sudo apt-get update && sudo apt-get upgrade

2. Reboot.

3. Install the raspberrypi-kernel-headers and device-tree-compiler package by using the following command:

   sudo apt-get install raspberrypi-kernel-headers device-tree-compiler

4. Test if the version of the running kernel matches the version of the kernel headers, the following command should show the directory for compiling the sensor module kernel module driver:

   ls "/usr/src/linux-headers-$(uname -r)"

5. Install the dkms package with:

   sudo apt-get install dkms

2.3 Driver Installation

1. Copy the driver debian package (vc-mipi-driver-bcm2835-dkms_x.x.x_armhf.deb) to the /tmp folder on the Raspberry Pi.

2. Install the driver package by calling (replace x.x.x by the current version number):

   sudo dpkg -i /tmp/vc-mipi-driver-bcm2835-dkms_x.x.x_armhf.deb

3. Edit the file /boot/config_vc-mipi-driver-bcm2835.txt (for example with nano with the command: sudo nano /boot/config_vc-mipi-driver-bcm2835.txt).

   • choose the correct platform by uncommenting the corresponding line

   • choose the correct overlay according to your MIPI module and platform by uncommenting the corresponding line

   • choose the desired sensor mode (see chapter Sensor modes below for mode description)

   • change the IO configuration if necessary (see chapter IO configuration below for IO description)

   • activate the self-triggered mode (see chapter Self-triggered mode below)

   All these settings can be done for cam0 and cam1 in case you are using a Raspberry Pi CMIO or a VC CMIO.
   
   

   Open

   

Part 2:

Open

Example of configuration file for the OV9281 MIPI module on a Raspberry Pi 3B+ in sensor mode 0.

4. Reboot.

2.4 First Image Acquisition Test

A sensor device should be listed as Video input at the following command output (from the Video4Linux-Control):

v4l2-ctl --all

The following command dumps sensor data:

v4l2-ctl --stream-mmap --stream-count=-1 -d /dev/video0 --stream-to=/dev/null

It will output subsequent lines ending with a frames-per-second information (in the example named [number]) until pressing CTRL-C:

<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< [number] fps <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< [number] fps <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< [number] fps <<<<<<<<<<<<<<<<<<<<<<<^C

2.5 Running the demo

The demo itself is a program named vcmipidemo and its source code is mainly in the file vcmipidemo.c. However more programs are provided, namely the vcimgnetsrv, a network image server, and its counterpart vcimgnetclient.py. The vcimgnetsrv is started as background service, and the vcmipidemo connects to it. Then you can use the vcimgnetclient.py on your PC to view live captured images.

But for the first run it is better to just run the vcmipidemo and check if it shows the ascii representation. This works without any network cable attached. You can then output the captured image to the framebuffer of the display by using the -f command line switch.

2.5.1 Compile the programs

1. Unpack the previously downloaded archive vcmipidemo_x.x.x.zip and copy the folder vcmipidemo to the Raspberry Pi (for example to /home/pi/)

2. Change to the subdirectory named vcmipidemo/src.

3. The source directory contains a Makefile to compile the driver. Do so by calling:

   make clean all

2.5.2 Execute the demo

Just run the demo itself:

./vcmipidemo

or with framebuffer output:

./vcmipidemo -f

or with live view over ethernet:

./vcimgnetsrv & ./vcmipidemo

For live view over ethernet, execute the vcimgnetclient.py at the client. This needs Python 2 and PyGTK. Install both following packages in this order (Windows):

• Python 2: https://files.vision-components.com/ImageTransfer/python-2.7.11.msi

• PyGTK: https://files.vision-components.com/ImageTransfer/pygtk-all-in-one-2.24.2.win32-py2.7.msi

2.6 Switching Sensor Configuration

2.6.1 Sensor modes

The sensor driver provides different modes which support several features. They can be switched by changing values of sensor driver parameters.

To list available parameters of the sensor driver kernel module, you can use the following command:

dmesg

You can also check the table below (Sensor modes description) for a complete list of available sensor modes.

Open

Example of the dmesg command for an OV9281 MIPI module, showing the available sensor modes.

To set the desired mode, edit the file /boot/config_vc-mipi-driver-bcm2835.txt (for example with nano with the command: sudo nano /boot/config_vc-mipi-driver-bcm2835.txt). Change the sensor mode by modifing dtparam:

dtparam=cam0_sensor_mode_1

The number after the underscore is the sensor mode. For mode 0 the correct setting would be dtparam=cam0_sensor_mode_0.

Open

Example of setting the sensor mode to 1 for cam0.

2.6.2 Sensor modes description

This table lists the available modes for all MIPI modules.