Cabling industrial environments today
Q: I have a little dilemma. I am doing the network infrastructure for Pratt & Whitney, which has been complaining that the installation of a network cabinet in its shop areas is too expensive.
Q: I have a little dilemma. I am doing the network infrastructure for Pratt & Whitney, which has been complaining that the installation of a network cabinet in its shop areas is too expensive. To save money, the client is asking my company to employ the use of Cisco's 2900 Series LRE XL for some of its Ethernet-connected devices. I am against using this equipment because it defies the structured cabling standards specified in the TIA/EIA-568 standard. What are your thoughts on using the Cisco 2900 Series LRE XL equipment for long Ethernet runs?
Donald L. Rio
CSC Network Infrastructure
East Hartford, CT
A: I have a long-time friend who often tells the story of his father's tools. His father was the coach at the local high school, who raised a few head of cattle as well as seven children. Needless to say, back then there was not a lot of slush in the tool budget at the McCellen household. I have been hearing for 20 years, "Dad fixed everything with the same broken claw hammer and bent screw driver." His children all are grown, he retired from the high school, and now travels from state to state with his wife, visiting their many cattle ranches. But I am told, "Coach still uses the same old tools."
Well, that is how his story goes. But why are you limiting yourself and your clients to using "the same old tools"? The structured cabling system described in TIA/EIA-568B is for commercial buildings, not industrial and manufacturing facilities. What works well in an office environment tends to become problematic when you add temperature and humidity extremes, harsh chemicals, vibration, and all sorts of electrical "noise."
The TIA's TR-42 Committee on User Premises Equipment has established the 42.9 Subcommittee on Industrial Tele communications Infrastructure for the purpose of establishing open, non-proprietary standards relating solely to cabling infrastructures in industrial environments. But it is unlikely that your client can wait another year or two for the standards, so here are my thoughts.
In an industrial control system, there are two types of applications: information and control. Information applications typically are used to monitor processes and gather statistics, as well as to process data and diagnostics. While control applications require fast response and high data throughput, network availability is paramount. Information gathering is not as critical as control. Having an aircraft engine land on the shop floor due to a collision on your data network would be difficult to explain.
When planning your system, determine the number of devices and ports, as well as horizontal cable lengths and length limits. If your site includes spot welders, induction welding, electrostatic processes, motor-control centers, or high-current (greater than 100-A) devices, you should consider using either shielded twisted-pair copper cable or optical fiber. If you choose shielded twisted-pair, remember that it is the bonding of the shield that makes the difference.
Conventional wisdom is to bond the shield to ground at the network-equipment end and isolate at the device. Select the cable jacket carefully; you may need oil- or water-resistant cable jackets, or jackets rated to withstand weld spatters. You also may need sealed connectors to avoid dust or liquid contaminants.
Personally, I would use optical fiber with a jacket rated for the severity of the environment, or install the cable in conduit. This approach would allow longer lengths of cabling with no electromagnetic interference (EMI) concerns. You could also use Cisco 2900 extended-distance switches. Plan to locate the network equipment in an accessible area of the plant with clean power and ground nearby, but away from the EMI, vibration, and other electrical-noise sources.
Next, plan your cable routing to avoid lights, motors, drive controllers, arc welders and the like. You may want to consider using a metallic wireway or ferrite beads to reduce common-mode and shield-current noises.
And then there are the outlet locations. Realize that in an industrial environment, not every cable will necessarily be terminated on a jack with a work-area cable from the jack to the device. If you are cabling to a network-controlled machine in the middle of a shop floor and there is no logical place to mount an outlet, then field-terminate the plug (being certain that you are using a plug kit expressly for solid conductors) on the end of the horizontal cable, and plug it into the jack on the machine.
Q: I recently ran into a situation with a Digital Subscriber Line (DSL) provider. The provider sent a technician on an equipment install. The Regional Bell Operating Company (RBOC) previously had installed and tagged the circuit. The DSL technician found a spare four-pair Category 3 cable and, for no apparent reason, pulled out the green/white and brown/white pairs and terminated them in a separate jack.
The client called me because he was having a problem hooking up to the jack. I informed the DSL technician that he should be installing a new four-pair cable from the demarcation point to the equipment location. He responded that a new cable was not necessary. I further questioned why he pulled out the green and brown pairs instead of using the typical T568A or T568B termination models. He said they "do it all the time."
I raised the question of shared-sheath compatibility. Response: they do it all the time and the RBOC does not have any problems with multiple circuits in the same sheath. "What's the problem?" I was asked.
I was not on-site, but spoke to the client who had been waiting months for the circuit and did not want to wait any longer. The client decided to go ahead with the installation. In my experience, on most dedicated data circuits like T-1 or ISDN, the circuits are run in separate cables. Or, at least, the transmit and receive pairs are separated in binder groups. Maybe I missed a standards release, but the DSL provider's approach does not sound right.
A: Ah yes, the virtues of DSL. Digital Subscriber Line is a technology that brings high-bandwidth connectivity to you over existing outside-plant copper cabling. From an access or service provider's perspective, DSL's greatest virtue is its use of the existing twisted-pair infrastructure. Yes, POTS cable to the demarc and, in your example, beyond.
New cable or old, shared sheath or not, it all comes down to the result of the line test. If the technician tests at the modem and the signal is strong, the circuit will work. If it fails, a new line can be assigned from the central office to the demarc-just like an ISDN or T-1 installation-at no additional cost to your client. But a new cable from the demarc to the modem would be at the client's expense.
The extra telephone cabling inside the house acts as a combination of short bridge taps. So, caution your client to install a splitter-sometimes called a filter-either at the demarc (which would be before the modem), or at each telephone outlet in the house.
TIA's industrial-infrastructure standard * When published, the Industrial Telecommunications Infrastructure Standard will address the requirements for telecommunications infrastructure in industrial buildings, structures, and campuses, and other facilities that are beyond the scope of the Commercial Building standards.
The standard will include system topology, pathways and spaces, cable and associated connecting hardware, grounding and bonding, power-system coordination, and installation.
Donna Ballast is a communications analyst at The University of Texas at Austin and a BICSI registered communications distribution designer (RCDD). Questions can be sent to her at Cabling Installation & Maintenance or at PO Drawer 7580, The University of Texas, Austin, TX 78713; e-mail: firstname.lastname@example.org.
Why distance is an issue with DSL
Ever wonder why the distance from the central office in which DSL can operate is so short? Loading coils are the culprits. They are used to improve frequencies in the high end of the voice spectrum to the demise of frequencies above 3,600 Hz-where DSL would operate. Typical connection speeds for ADSL range from 1.544 Mbits/sec to 512 kbits/sec downstream, and are approximately 128 kbits/sec upstream.
An ADSL line allows for one copper pair to carry both voice and data signals, and for the data part of the line to be continuously connected. DSL achieves higher data-transfer rates by using more of the available bandwidth. POTS only uses the 0 to 3,400 Hz frequency range. DSL excludes the 3,400-Hz frequency boundary by outmoding the digital-to-analog conversion that modems perform, and connecting both ends digitally. Hence, more bandwidth, allowing higher transfer rates.