PCI CAN Card selection and installation

Dear Felix

Could you please tell me what PCI CAN card is used for the linux computer? As far as I understand this card, its purpose is the reduction of package losses happening in the CAN communication. How can I select the right one for my linux computer? Can you give me a guide to find that out or can I choose an arbitrary one? I wonder also how to connect the card to the computer.

Thanks a lot and have a nice day!
Yannick

Hi Yannick,

we use the dual channel PEAK PCI Express cards.
Part number: IPEH-003027
https://www.peak-system.com/PCAN-PCI-Express.206.0.html?&L=1

I am not sure about other options.
My coworkers might be able to give you some advice.

Best. Felix

1 Like

hi Yannick,

unfortunately, there is no good way to find out if the CAN card will be compatible with your computer or not. We have good experience with the one Felix linked.

That said, besides using the CAN card you should also make sure to install a realtime version of Linux. We use the rt_preempt patch from real-time linux, see here: https://wiki.linuxfoundation.org/realtime/start. We have an install script for building the kernel on Ubuntu here: https://github.com/machines-in-motion/ubuntu_installation_scripts/blob/master/rt-preempt/ubuntu18.04/install_rt_preempt_kernel.sh.

Best,
Julian

1 Like

Hi Felix and Julian

Thanks a lot for your information! I didn’t know there was an installation script for patching ubuntu with rt_preempt. I did it with a guide (guide to patch ubuntu with rt_preempt) some days ago.

All the best
Yannick

Hi Julian

Do I need a realtime CAN driver aswell if I use ubuntu with rt_preempt?
Because on the website, it is written that for xenomai it is needed.

Best
Yannick

Hi Yannick,

for rt_preempt with the peak CAN cards we are using the default drivers coming with the kernel. The documentation you found was valid when we were still using Xenomai.

Note that you might have to initialize the can channels. For this, we have a script which you can find here.

Best,
Julian

1 Like