The GigE Vision Cost Advantage Debunked

If you’re involved with “high-end” machine vision applications, you’ve no doubt heard that because CoaXPress uses a frame grabber it’s more expensive to implement than its main competitor, GigE Vision, which does not. Simplistically speaking, this would put CoaXPress at a cost disadvantage, especially for multi-camera systems. But as with all things technical, the true picture emerges only once you dig into the details, especially when considering that CoaXPress is the only standard that provides video, camera control, and triggering over a single coaxial cable, along with 13 W of DC power at 24 VDC.

First of all, a frame grabber is a lot more than simply a device that connects the camera to the computer. It performs all acquisition and IO synchronization functions, manages external devices including their triggering and synchronization required for line-scan applications without physical triggering, and provides a wide array of advanced imaging functions such as color correction.

As frame grabbers can achieve latencies of less than 1 µs, they’re extremely well suited for applications such as industrial inspection and sports broadcasting where the ability to achieve superior image analysis from different angles and perspectives is required. In a high-speed industrial inspection system, the frame grabber synchronizes image captures with triggers while simultaneously processing the images, which makes extremely fast pass-fail decisions possible.

CoaXPress allows cameras with different resolutions and frame rates and a mixture of CMOS and CCD cameras can be linked to a single frame grabber, each one performing a different inspection task. The ability to connect a single camera to more than one frame grabber (that can also be located in different PCs, further increases the frame grabber’s contribution).

A GigE Vision system makes it necessary for all data de-encapsulation to be performed in the host PC’s CPU, so the CPU can be overloaded when a camera is capturing images at 300 frames per second or higher. In addition to its primary functions, the CPU must now handle others with the result that it will begin to drop data packets.

Now, you can always mitigate this to some degree by using a network interface card (NIC) in the PC, but as they cost 50 to 80% as much as a frame grabber card, some of GigE Vision’s touted cost benefits go away. And after all, NICs are designed primarily for transferring data rather than processing images like a frame grabber, so they’re hard-pressed to deliver image processing ability comparable to a frame grabber. In short, to come even close to what CoaXPress can achieve using a frame grabber, GigE Vision requires a more powerful processor and a more formidable NIC, negating any benefit in cost reduction.

The issue becomes even more onerous in multi-camera systems when, for example, a four-camera system simultaneously captures two side views and the top and bottom of an object. This is best achieved with CoaXPress because of the very precise triggering and synchronization enabled by the frame grabber on each channel.

While GigE Vision systems can provide reasonably precise synchronization for pre-defined acquisitions via the Precision Time Protocol (PTP), they cannot support real-time triggers from the PC system to the cameras. In contrast, CoaXPress can link multiple cameras with a single frame grabber over long runs of coaxial cables with near-zero latency and precision synchronization via real-time triggers.

Another issue, often overlooked, is power consumption, which in the case of GigE Vision, is quite high, which requires a power supply with larger components. They all create heat, potentially making the camera quite warm. Unfortunately, camera sensors are very sensitive to temperature, and when the temperature rises so does the noise, which significantly degrades image quality.

As for CoaXPress itself, the latest version (CoaXPress 2.0) is also faster than GigE Vision whose maximum data transfer rate is 10 Gb/s. CoaXPress 2.0 adds new CXP-10 (10 Gb/s) and CXP-12 (12.5 Gb/s Gb/s) tiers with a range greater than 65m and with repeaters several hundred meters. It continues to support CXP-1 (1.25 Gb/s up to 210m) to CXP-6 (6.25 Gb/s up to 90m.

Doubling the maximum data rate in CoaXPress 2.0 also reduces by half the number of cables required to achieve a given data rate. Upgrading from tier to tier can be made without replacing all the cable infrastructure, and the system runs at the speed the camera requests. The uplink path for camera control and triggering has been doubled in CoaXPress 2.0 to 41.6 Mb/s, which further reduces latency, increases trigger rates, and allows more data to be handled by high-speed cameras without the need for a dedicated high-speed uplink. The increase in speed also lets the host send trigger messages at much higher rates—nearly 600 kHz in single trigger message mode or 300 kHz in dual trigger message mode.

With four-link CoaXPress 2.0, a frame grabber can now transfer data at 50 Gb/s, much faster than the 10 Gb/s of GigE Vision, and more than 100 Gb/s can be achieved with an eight-channel frame grabber card or two 4-channel frame grabber cards. This performance may be a lot more than is required by today’s advanced imaging systems, but it ensures that CoaXPress will be able to support machine vision applications far into the future.  

Other features introduced with CoaXPress 2.0 include improved time stamping, error reporting, and data sharing. Unified Time Stamping integrates reporting events coming from cameras, hosts, and software into a unified time reference that helps keep track of when events occur. A new event channel adds a data path that allows a camera to precisely inform the application when specific internal events occur, such as the start of exposure. The CoaXPress control channel now has a tag field into control packets to allow the host and camera to consistently recover from error conditions. CoaXPress 2.0 also now defines rules for data sharing in which a camera simultaneously streams data to more than one host. 

To support CoaXPress 2.0, Microchip Technology’s latest transceiver SoC, the EQCO125X40, has an integrated high-speed equalizer, re-clocker, and cable driver. It consumes only about 125 mW at its full 12.5 Gb/s speed and less power at slower speeds. At the other end of the link, a frame grabber typically uses the same SoC. The EQCO125X40 can perform link signal integrity and cable margin tests in real-time before and during operation, so worn cables, and connector discontinuities can easily be detected.

The EQCO125X40 also has clock data recovery that performs at all speeds, not just those dictated by the CoaXPress speed tiers, and on-camera low-frequency clock recovery that eliminates the need to program a separate clock in the FPGA. Finally, the EQCO125X40 also supports shielded twisted-pair (SDP) cable and 50-ohm coaxial cable in addition to the traditional 75-ohm coaxial impedance.

So, when all factors are considered, there is no cost advantage afforded by using GigE Vision, and considering the many benefits delivered by the frame grabber and the inherent performance advantages of CoaXPress 2.0, should you really consider anything else?