Factors affecting network bandwidth

Bandwidth is defined as the carrying capacity of a circuit. It's usually measured in bits per second for digital circuits, or in hertz for analog circuits. Though a hardwired network is also limited by the speed of its devices and microprocessors, a typical 4-20 mA signal over typical 18-gauge wire usually has about a 10-ms response time.

By Jim Montague September 1, 2003

Bandwidth is defined as the carrying capacity of a circuit. It’s usually measured in bits per second for digital circuits, or in hertz for analog circuits. Though a hardwired network is also limited by the speed of its devices and microprocessors, a typical 4-20 mA signal over typical 18-gauge wire usually has about a 10-ms response time. Fieldbus response times can range from 2-50 ms or more.

“Network bandwidth comes down to two measures: 1) transactions per second—how many communication cycles is the network accomplishing per second, and 2) latency— how long did it take to get the data from the time you wanted it?” says Wally Pratt, chief engineer at the HART Communication Foundation. “When you’re talking about throughput, bytes per second, latency means how many times can you measure your primary variable? When you’re talking about a multi-drop network, you have to take transactions per second and divide by the number of devices on the network, which will show the number of times you can talk to each one. Then, you have to ask is that time enough for the application?”

Perry Marshall, author and consultant, adds that users need to determine their speed and data requirements. “Does your I/O device need to be scanned every 2 ms? Or is 500 ms okay? Do your devices send a few bits of data, or hundreds of bytes? Are you mixing simple and complex devices on the same network?” asks Marshall. “To answer these questions, determine the ‘hard limits’ and worst-case requirements of your design. Which networks are available on specific devices you’ve chosen? This issue may force your hand, or at least limit your choices, if you are required to use a specific brand of PLC or other component. “There’s only a loose relationship between bandwidth and speed. The real tradeoff is among bandwidth, response time, determinism, and noise immunity.”

“A lot of the improvements in network capacity is due to refinements in the technology used to pump data down a wire at ever higher rates,” says Dick Caro, a well-known fieldbus expert.” Users need to be sure that the network in their production areas is designed for real-time data transmission, which means about a 1-second response in process applications and 5 ms in applications with factory safety switches.

Jim Montague, news editor, jmontague@reedbusiness.com

Fieldbus Comparison ChartThe following includes sections of the Fieldbus Comparison Chart that deal directly with capacity and bandwidth issues. The full chart, developed by the former Synergetic Micro Systems and now known as Grid Connect Inc., includes other fieldbus protocols and is located at www.synergetic.com/compare.htm .

Name
Max. devices/distance
Transmission properties
Data transfer size

Profibus PA/DP
127 nodes/100 m between segments at 12 Mbps
DP: 9.6,19.2, 93.75, 187.5, 500 kbps and 1.5, 3, 6, 12 Mbps PA: 31.25 kbps
0-244 bytes

DeviceNet
64 nodes/500 m depending on baud rate
500, 250, 125 kbps
8-byte variable message w/ fragmentation for larger packets

Foundation fieldbus (H1)
240 per segment and 65,000 segments/1,900 m at 31.25 kbps wire
31.25 kbps
128 octets

Foundation fieldbus (HSE)
Unlimited with IP addressing/2,000 m at 100 Mbps fiber full-duplex
100 Mbps
Varies; uses standard TCP/IP

Industrial Ethernet
1,024 nodes, expandable via routers
10, 100 Mbps
46-1,500 bytes

Modbus Plus
32 nodes per segment; 64 max./500 m per segment
1 Mbps
Variable