In the early days of machine vision, like with all new technology, there was a lot of confusion as to what constituted a video interface between camera and computer. It was known that a camera and frame grabber were needed, but because the frame grabber and camera manufacturers were from places all over, there wasn't much agreement on how the two should be connected. This led to a time of several cameras with several cables, often the same camera manufacturer may have had different cables for each family of cameras on offer. Couple that with the fact that frame grabber interfaces also had no interface standards and you ended up with a lot of cables, sitting around long after the frame grabber or the camera had served its purpose. This led to standardization, and agreement by all parties on what constitutes an acceptable number of interfaces.
One of the early adopted and agreed upon interfaces was the Camera Link interface, first introduced in late 2000 (v.1.0) and updated in 2004 (v1.1). The current revision (v2.0), was ratified in February 2012. In 2006, another interface, GigE Vision, was introduced and the latest version (v2.0) was released in November 2011.
As time moved on, it was generally accepted that Camera Link was the industry standard for vision systems requiring real time data in small volume systems (large image transfer with low latency) whereas the GigE Vision could support data transmission up to 100m albeit at a time loss. GigE Vision used standard CAT-5/CAT-6 network cables, which are cheap as a cable needed for data transmission. Camera Link was standardized so the customer had a vast array of cameras and frame grabbers to choose from.
While other interfaces (Analog, Firewire, USB2 etc.) are still out in the machine vision world, the design rule of thumb was, Camera Link for high speeds and short distances and GigE Vision for lower speeds (100MB/s) at long distances.
As one can see both of these are fairly new interfaces but each has its own limitations. These limitations weren't realized until sensor technology began to develop at a fast pace in the 2000’s. This development resulted in improved resolution and higher frame rates, meaning there was an increasing need for larger bandwidth. It was time to develop a new standard, something that would be an improvement over the GigE Vision and Camera Link interfaces.
With GigE Vision 2.0 now scalable to 10 GigE, this has resulted in some benefits, but the resulting power requirements and the fact that the transmission speeds are similar to those of Camera Link Full+ (850MB/s), meant that universal adoption wasn't immediate and trepidation by some manufacturers slowed its growth.
Another option is CoaXPress, which was adopted in early 2011. CoaXPress is the current interface that will be one of the fastest growing for the next 3-4 years. Because of the low cost of cables, and the data transmission rates and achievable distances, this interface looks to be a major player in the coming years. CoaXPress can currently be expanded to four ports, resulting in 25Gb/s of data throughput, so the technology expansion is resolved for now and the next few generations of sensor. CoaXPress also has the benefit of being able to operate on industry standard 75Ω coaxial cable.
Camera Link HS (a new interface, not Camera Link 2.0) has taken some of the benefits of Camera Link and enhanced them, while at the same time, taking the weaknesses and eliminating them. Features such as a low cost solution, scalable bandwidths and reliable data delivery are part of this new interface.
The Camera Link HS interface will be released in May 2012 and it's poised to be a big player in the Machine Vision market. However, the timing of its release will be a factor in its success. The recent adoption of CoaXPress and its many benefits has led several frame grabber manufacturers and camera manufacturers to build their newest product to this standard. At the most recent Vision trade show in Stuttgart (November 2011), arguably one of the most important trade shows for the vision industry, there were considerably more CoaXPress products than Camera Link HS.
Another interface is USB3 Vision which was also accepted as a global standard in late 2011. A draft standard will be available in May 2012. Looking back at Vision again, there were a lot of USB3 cameras in effect. USB3 cameras are popular but for companies looking to simply replace part of the their system, they will need to upgrade their laptops and motherboards to the latest version. USB3 also needs Windows 7 with a service pack to run, but when Windows 8 is released (due Q3, 2012), USB3 support will be native.
If I were a gambling man, I'd venture a good deal of money on the interface that's accepted first by G3. After all, an interface can change a good deal until it's released. On the CLHS side, no one has achieved the 20 links yet. This is because so many input levels, from sensor people, to camera people, to cable people each have their own agenda for certain parts of the standard. These changes may be seen in later revisions of the standard.
When a customer needs a vision system, cost can be as much of a factor as performance. For some customers performance is the ultimate factor and as such, they will simply use the one that does exactly what they want it to do. For others, the price point can dictate how “exact” they want their system to be, or how much they can live with certain factors, (bandwidths, distances etc.) Then there's the customer who wants the redundancy factor. Knowing that there are other cameras/cables or frame grabbers other than the one they selected gives them a sense of ease that this is an interface that’s here to stay and not the whim of some engineers in a back room of some lab.
For now, Camera Link is still the king of high speed data transfer, albeit over shorter distances, but the CoaXPress team is looking for that title. USB will always be available, and not just as a cheaper option. Each standard has it’s own merits. The Camera Link HS draft is now out for the required final review by industry stakeholders. While little is expected to change, it’s not set in stone as yet. But like all things we wait for, the delivery might not be exactly what we expect. A summation might be a phrase from the 1500's “Better the devil you know, than the devil you don't.”