Good examples of the wrong way

Optical time domain reflectometers(OTDRs) have been the topic of several recent columns, including last month where I detailed how to set up the OTDR properly to make measurements. This month, the series concludes with the wrong way to use an OTDR and information on what kinds of trouble it can cause.

Since an Internet search on “fiber optic testing” will get you directly to me or to The Fiber Optic Association, which I help run, many people confused about fiber optic testing end up calling me. These are two such cases.

The first call was from an end-user, and it started, as many do, with the question, “What is an acceptable loss for my fiber optic cables?” I sent the caller to the FOA Web site for “Tech Topics” (www.thefoa.org/tech), where there are two pages addressing this topic, calculating loss budgets from the standpoint of the network designer and expected measurement results.

The second caller asked if he could send me some documentation submitted to the end-user as confirmation the installation was completed and if I would examine it and give my opinion. After reviewing the OTDR traces he sent me, one of which is reproduced below, I confirmed they were what the installer provided for documentation.

Close examination of the traces revealed the OTDR was connected to the testing cable with a launch cable of adequate length to allow the OTDR receiver to settle down after the test pulse, but the cable under test was about 76 feet long, awfully short for OTDR testing. The pulse width at 850 nm was short enough at least to see both ends of the cable, so we know no receive cable was used in the testing. Thus the OTDR test was useful only for measuring the loss of the first connector and perhaps finding faults in the fiber of this short cable.

At 1,300 nm, the pulse width was so great the reflection at the connection between the launch cable and cable under test overwhelmed the OTDR, masking any details of the 76-foot cable. Since the pulse stretched the full length of the cable under test, it was impossible to measure the connection loss or see any details in the cable.

But wait—at 5 points on the OTDR documentation, it says “PASS” anyway. The 850 nm data gives a loss of 0.21 dB, but since we don’t know the location of the markers, we don’t know if it’s just the loss of the connection to the launch cable or if it includes the fiber loss. Since 76 feet of fiber has only about 0.08 dB loss, it’s not important. Let’s give the OTDR the benefit of the doubt and say a “PASS” is OK here.

However, for 1,300 nm, it shows a loss of –0.39 dB. That’s a gain, not a loss, but it says pass anyway. Anytime an OTDR indicates a gain, it needs some operator interpretation of the actual trace. And a knowledgeable operator would have thrown this trace out because it’s too short for the OTDR to measure at all.

To satisfy my own curiosity, I asked the OTDR manufacturer why the OTDR gave a “pass” to this data. Bottom line, it’s all in the programming. “Garbage in—garbage out.”

I explained all this to the caller; plus, I pointed out acceptance testing to industry standards required insertion loss testing, not OTDR testing. That caused a conference call involving the contractor, where we discussed the relevant testing standard and procedures for loss testing. In that conversation, it was revealed there were about 30 cables with a total of 1,000 fibers involved in the installation, and the installer would be required to retest most of them.

All told, I spent about 10 hours on this project. This issue became so significant that I created a new document on the FOA Web site to explain fiber optic testing methods: www.thefoa.org/tech/VDV4W.pdf. If you need to know about fiber optic testing procedures, I suggest you download this document and study it.

And if you think it sounds bad to have to retest 1,000 fibers, someday I’ll tell you about a contractor on another installation who removed and destroyed about $100,000 worth of perfectly good cable due to a similar OTDR testing mistake.          EC