Since the introduction of the GigE Vision standard, Gigabit Ethernet is rapidly gaining market share against other camera interfaces, and there seems to be a consensus that it will play an important role in the future. However, sales figures of digital cameras for machine vision applications still put FireWire as the leading digital interface. Will Gigabit Ethernet replace FireWire as well as other interfaces as the leading standard of the industry? Or will both technologies complement each other so that they can coexist in the market?
Cost-Effective Standards
Both interfaces have one thing in common: FireWire and GigE use highly standardized hardware and data transmission protocols, offering plug-and-play compatibility of components and easy integration in a system. The IIDC/DCAM standard guarantees easy integration of compatible IEEE 1394 devices from various vendors while the GigE Vision and GenIcam standards ensure the compatibility of Gigabit Ethernet components. This saves developers time and money in the design of their imaging systems.The relatively low cost of hardware components and accessories makes both interfaces a good choice for cost-sensitive applications. FireWire, originally a consumer electronics interface, is widely available in the market and many PCs have an IEEE 1394 port on board. Plugs, cables and other accessories are affordable and easy to purchase from local retailers. This is even truer with GigE Vision, as this interface uses Ethernet connectivity (Cat5), which has been used as the most common standard in IT networks for years. Mass produced cables and components are extremely easy to find and even cheaper than FireWire accessories.
The relatively low cost of hardware components and accessories makes both interfaces a good choice for cost-sensitive applications. Source: Allied Vision Technologies
When Size Matters: Form Factor and Cable Length
The physical size of the camera can be important for specific applications. Because Gigabit Ethernet components are less miniaturized than FireWire, GigE cameras still suffer from a slightly larger form factor than their FireWire competitors.An often-considered aspect in the choice of the interface is the possible transmission range between the system PC and the camera. In that respect GigE Vision clearly is the interface of choice as it allows up to 100-meter cable length, or up to tens of kilometers using low-cost fiber optic media converters.
However, the possibilities of FireWire in terms of cable length are often underestimated. It is true that the IEEE 1394 standard only guarantees data transmission over 4.5 meters, but this by far does not mean that longer distances cannot be bridged in practice. For example, based on extensive tests, some companies commit to cable length of up to 10 meters with IEEE 1394b and 17 meters with IEEE 1394a interfaces. Should longer distances between a PC and camera with a FireWire interface be required, repeaters offer a practical solution. It is even possible to bridge very long distances using optical fiber (GOF).
Bandwidth
With up to 125 megabytes per second (MB/s), GigE Vision clearly offers the highest bandwidth. Among GigE Vision camera vendors, some make the most out of this bandwidth by running sustained data rates of 124 MB/s. In FireWire, some companies achieve higher bit rates than specified by the standard with up to 84 MB/s with IEEE 1394b.
This exhibition demonstration illustrates the principle of a daisy chain network with direct FireWire connection from camera to camera. Source: Allied Vision Technologies
Power Supply and Heating
The FireWire and GigE Vision interfaces are quite different as far as the power supply of the devices is concerned. FireWire cameras usually are powered directly over the FireWire port, which highly contributes to the plug-and-play ease of installation: just plug in the cable and the camera runs. Power over Ethernet has now become available too, but it is not a standard feature of the interface and, therefore, not yet easy to implement.Most GigE cameras do not support power over Ethernet and even if it is the case, it is very likely that the Ethernet port of the PC will not, so additional accessories such as a new Ethernet card or a powered switch have to be purchased.
Gigabit Ethernet cameras tend to consume more power than FireWire cameras. This may turn out to be a disadvantage in mobile applications powered with batteries. Related to this higher energy consumption is also a higher heating of the camera, which might be problematic in specific applications. However, some GigE Vision cameras perform well compared to FireWire. For example, there are GigE cameras with a power consumption of 2.5 watts, which is as low as that of comparable competitive FireWire cameras.
Small GigE Vision cameras are available on the market today, as the physical size of the camera can be important for specific applications. Source: Allied Vision Technologies
Multi-Camera Operation: Network vs. Bus
Industrial image processing systems often rely on several networked cameras. Therefore, the capabilities of an interface in the area of multi-camera operation are often crucial for the choice of the most appropriate one.IEEE 1394 and GigE Vision have two different philosophies: FireWire is a bus standard, while GigE Vision is based on an Ethernet network. According to the strict communication protocol of the FireWire standard, devices on the bus can only transmit data on the bus after each other within clearly allocated timeslots. As a result, system developers can precisely calculate and plan what data from which camera is to be transmitted when to the processing computer. Using a resource manager, it is possible to make sure that the bus cannot be saturated with image data sent by several cameras at a time.
In a Gigabit Ethernet network, all devices can send data at the same time, which means that the limits of the bandwidth may be reached. A bottleneck risk exists specifically when a switch connects several cameras to the network. If the capacity of the switch is saturated by too high an amount of data, the data will not necessarily be lost, but delivery delays can occur and, more importantly, it is not predictable which data will be delivered and when to the PC. Some Gigabit Ethernet cameras feature a sophisticated bandwidth management function that ensures that image data is reliably and predictably delivered in multi-camera operations.
A unique benefit of the FireWire interface is the possibility of daisy-chain connectivity. Cameras equipped with two IEEE 1394 ports can be connected directly with each other like pearls on a necklace. That way, it is possible to easily build a multi-camera bus with only one FireWire connection on the PC side and no hub or accessories others than standard cables.
The GigE Vision standard does not limit the number of cameras that can be operated in one network. There is a limit of up to 63 cameras on a FireWire bus but this limit is far higher than the vast majority of machine vision applications require. Even if a system required more, it only takes a standard 4-port PCI card to operate up to 252 cameras with one PC. In fact, for both FireWire and GigE Vision, the total amount of cameras that can be operated simultaneously is rather limited by the available bandwidth than by the standard itself.
Reviewing these different criteria, it becomes obvious that the choice of the better interface really depends on the requirements of the specific application: for some systems the form factor is key, for others the cable length or power supply. Those using complex multi-camera systems will probably value the advantages of the FireWire bus standard as they did before, while GigE Vision is the interface of choice for applications requiring long distances between the cameras and processing computer such as traffic monitoring. Both interfaces complement each other and together they will surely dominate the machine vision market of the future.Q
Tech Tips
Standards 101
Quality Online
For more information on machine vision, visit www.qualitymag.com to read the following:“Machine Vision Achieves More”
“Machine Vision ‘Sees’ Color”
“Machine Vision Fundamentals”
