How can I connect multiple cameras of type mvBlueCOUGAR-XD (Dual GigE) to one PC?

mvBlueCOUGAR-XD

Where to place the quad NIC card, if there is more than one slot available?

A typical block diagram of the actual Intel PC hardware is given below:

Bockdiagramm Intel PC-Hardware

Please note that PCI Express x16 (PCIe) slots usually are connected directly to the processor whereas the other PCI Express lanes come from chipset.

It should be further noted that all remaining chipset traffic goes via DMI (direct media interface) which itself has according to Intel a bandwidth capability of 20 Gigabit/s. This simply means that not all interfaces can transmit data at the given maximum individual limits. E.g 8 x 5 Gb/s PCIe 2.0 lanes = 40 Gb/s which is twice as much as possible.

Depending on the network cards PCIe revision, the bandwith transportation capabilities may vary. PCIe 1.0 can transmit 250 MByte/s per lane which means one dual GigE card with PCIe 1.0 is needed per dual GigE camera. PCIe 2.0 can transmit 500 MByte/s per lane, but it requires the corresponding network card.

How to check what’s inside your PC and with what revision?

The usage of HWINFO tool can be recommended: http://www.hwinfo.com/

HWINFO

The above screenshot shows e.g. a single port GigE network Express card V1.0 adapter, which is connected to a Intel Cougar Point PCIe single lane capable of V2.0.

Thus the following rules apply to exploit the max. bandwidth capabilities:

  • Place the first (quad) interface card in the x16 slot if possible.
    • Connect max. two dual GigE cameras here
  • Place the next quad card in a x4 slot
    • Connect max. two dual GigE cameras here
  • Place dual NIC cards in x1 slots
    • Connect one dual GigE camera here

Configuring network interface cards (NIC) for point-to-point multi-camera usage

All cameras must use fixed (aka persistent) IP addresses and the NICs must be configured to work in different subnets, due to the fact that LLA otherwise configures NICs in the same subnet and may mix up cameras subsequently.

Use the mvIPConfigure tool, as described here: http://www.matrix-vision.com/manuals/mvBlueCOUGAR-X/GUI_page_0.html#GUI_section_mvIPConfigure

Conclusion

Practically thus between 4 to 5 Dual GigE cameras can be connected and operated at full bandwidth in the above scenario with “reasonable” CPU load. Usually the machine vision application brings the CPU load up and will limit the overall performance of the PC.

Back