EYE ON FIBER

University Article Includes Severe Errors And Inaccuracies

Executive Summary

It is unfortunate that UTP proponents enlist others to spread inaccuracies, myths, lies and misconceptions about the use of fiber optic communications. This response to an article with glaring errors highlights the need to evaluate the accuracy of all information used to make decisions.

Professor:
I have read the press release (http://live.psu.edu/story/4917). I have found seven errors and inaccuracies regarding the use of fiber optic transmission. I am wondering how these errors emerged. Perhaps the PR department introduced these errors in its revision of the material you presented.

Errors 1-3
In the opening statement, (with a lower average bit error rate than fiber optic cable that is 10 times more expensive), there are three errors or inaccuracies.

First, fiber optic cables are immune, to EMI and RFI. No copper cable is immune. Some copper wires and cables may be more resistant to EMI and RFI than others, but none are immune. Therefore, the claim that the BER is less for copper wire than for fiber cable is false on a level of basic high school physics. If we believe the statement, transmission BER for wire will be less than zero!

In summary, the statement that wire has a lower BER than does fiber is impossible.

The second error is that fiber optic cable is not 10 times more expensive than copper wire, more specifically, UTP. A two fiber plenum rated cable is $0.20/ft (Fiber Instrument Sales catalog, page 17) while plenum rated UTP cable is $0.55/FT (Data Warehouse web price during 6/03). The favorable fiber optic cable price results from the common use of plenum cables. Fiber optic LAN cables are less expensive than UTP LAN cables because the fiber cables are smaller and require less of the expensive materials required to meet the plenum cable requirements. This fact is documented in the FOLS cost comparison model. This fact is supported by the FOLS members, many of whom manufacture both copper and fiber cables. See http://www.fols.org for the details.

The third error is the implication that wire and fiber component costs are the costs that should be compared. The UTP industry has been promoting this misconception and misdirection for many years. By continuing this misconception and misdirection, this document, does a disservice to the reader and to the fiber optic industry.
The cost that should be compared is the total initial installed network cost.
Fiber and UTP networks are not designed in the same manner. To do so would be to ignore the limitations of UTP and to ignore the benefits and capabilities of fiber transmission. Since these two media are not used in the same way, one of the valid comparisons is of total initial installed network cost. When one compares these two total initial installed costs, one learns that there are scenarios in which fiber networks have an initial installed cost lower than that of UTP networks. The savings can range from 3.3 % to an incredible 68.9 %! These numbers are in the FOLS cost model. In addition, many organizations are verifying this reduced fiber network cost through their implementation of FTTD. Their independent analyses support this reality.

The fourth error is the implication that cost of the cable is the important criterion. It is not: the important criteria are capacity (bit rate) and transmission distance. When cost, bit rate and distance are considered, fiber optic cables are less expensive per bit and less expensive per km than are UTP cables.

This statement depends on what the wire or fiber is expected to carry (bit rate). Since 1978, telephone companies have chosen fiber optic cables instead of copper cables because the fiber system provides a lower cost per bit than a copper system

This same relationship exists in some LANs. During the last four years, the FOLS of the TIA has performed an extensive cost comparison of fiber LANs to UTP LANs. This comparison, which is biased against fiber, shows that the total system cost of some fiber LANs is lower than that of an equivalent UTP LAN. This model does not show that all LAN configurations favor fiber LANs on the basis of initial installed cost. But many configurations do have an initial installed cost that is lower for the fiber optic LAN than for the UTP LAN. In addition, many organizations are adopting FTTD because their independent analyses indicate that FTTD networks cost less to install and less to maintain than do UTP networks.

The fifth error is in the statement Using copper wire is much cheaper than fiber optic cable and, often, the wire is already in place.

This statement ignores the network costs that are required by a UTP network but not by the fiber network. When these costs, often ignored or unknown by the UTP network designer, are included in a realistic comparison, the cost of the UTP network is higher. In some scenarios, such as fiber to the zone, the cost of a UTP network can be two to three times higher than the cost of a fiber to the zone network. See the K-12 and FTTZ scenarios in the cost model at www.fols.org.

These hidden or overlooked costs are the cost of the telecommunication room and the cost to support the equipment in the telecommunication room. The FOLS study indicates a typical cost of $20,000 for a telecommunication room on each floor of an office building. An earlier study by the Tolley Group indicates the cost of the telecommunication room to be closer to $40,000 per room. The fiber to the desk network does not need a room or any of the support costs. When your divide either of these cost values by the number of nodes supported by the telecommunication room, you get the hidden cost per node. With these two numbers, the hidden cost of a UTP network ranges from $416-$832/node at 48 nodes to $104-$208/node at 192 nodes per telecommunication room.

Any potential increase in cost from fiber components need be less than these crudely calculated values to produce a fiber to the desk network that is lower in cost than the UTP network.

The cost of the telecommunication room and the cost of supporting the telecommunication room are not new costs. I first heard of these costs in 1992.
In summary, the fifth error is that using copper wire can be more expensive, not less expensive, than using fiber optic cable.

The statement that often, the wire is already in place. is a contradiction of UTP history: in most cases, an increase in bit rate has resulted in a need for a increased performance UTP cable, so the cable in place needs to be removed, not used. Remember Cat. 3, Cat. 5, Cat. 5e, Cat. 6 and Cat. 7, now in discussion? With a fiber network, the end electronics can be replaced, the same fiber cable used, and the bit rate is increased.
The sixth error is that the installed UTP cable may not be able to be used.

The seventh error is in the statement Fiber optic cable typically achieves 10 to the minus nine. This statement is grossly incorrect. The standards for BER in fiber optic networks start at 10^-9 and go to 10^-14. For example, 802.3, Section 15.2.2 states that the BER of a 10BaseF link be better than/less than 10-9. The FDDI standard, which is referenced by the 100BaseF standard, states that the BER shall be less than 2.5 x10^-10. As a final example, 802.3, Table 38-4 indicates that the BER of a 1000BaseSX link be better than/less than 10^-12.

If you remember that the standards are negotiated political agreements (and not technically optimum solutions) and that the accuracy applies to the received electrical signal, the stated requirements are worse/higher than the expected performance of the link, not the cable.

The standards state these values, but the fiber optic cable greatly exceeds these values. Remember EMI and RFI immunity? If you believe in 10-9 from fiber cables that are EMI and RFI immune, from whence cometh the errors? There are no errors created during transmission through a fiber optic cable. Such errors come from connectors and electronics. Connectors and electronics can be a source of errors in both UTP and fiber networks. As an example of a realistic error rate, one of our clients, Grand Casino, ran an error test on its 155 mbps ATM fiber network for 30 days. There was one error in that time. This test indicates a BER of 2.489 x10^-15! This value is significantly better than the value of 10^-9that the document states.

In summary, the seventh error is that the typical BER value is totally incorrect. The BERs for optical fibers and their cables are so low to be unmeasurable (they are immune to EMI and RFI). Measured BERs for fiber links will be lower than measured BERs for UTP cables.

Finally, the document ignores the cost increase created by the need for signal processing: in a UTP link, the signal must be processed to recreate the electrical bit. In a fiber link, the received optical signal needs to be converted to recreate the electrical bit. As some point in time, the cost of the chip to perform the processing will exceed the cost to convert an optical bit to an electrical bit. As it stands now, the optical conversion process (recreate, retime and reshape) is a simpler process than the electrical conversion.
While a 10 Gbps copper cable will be useful for links within a data center, or in a central equipment facility, use of such a link to the desktop will result in network costs higher than those with FTTD. Note that all cost statements ignore additional cost factors that favor fiber. Two of these cost factors are lower maintenance cost and lower life cycle cost for fiber networks than for UTP networks. As a result of ignoring these cost factors, all cost comparisons are biased against fiber.

It is my hope that you will correct these errors and misconceptions in future publications.

Respectfully submitted for your consideration,

Eric R. Pearson, CPC, CFOS

President

Pearson Technologies Inc.

For Mr. Pearson contact information, click here.

Pearson Technologies Web Sites

http://www.ptnowire.com http://www.FTTDnow.info

http://www.fiberopticlawsuits.info http://www.sfoi.info

Contact Pearson Technologies Inc.