If you are testing long fibers, then you likely are using a dispersion shifted fiber. You will not find these in the list of fiber types the instrument knows about. However, you can create a custom fiber and select it as the fiber type, or set manual values in the third Setup tab (Cable). The manual describes this process for each instrument.
- The Index of Refraction affects the reported fiber lengths, especially long lengths.
- The Backscatter Coefficient primarily affects the reported reflectance measurements.
- Modal Bandwidth does not apply to singlemode fiber
- Core Diameter is nominally 9.0 µm for singlemode.
In premises environments, the overall loss is typically dominated by insertion loss at the connectors. Using a launch fiber prior to the fiber under test and a receive fiber after allows the entire fiber under test to be measured, including the end connections. Premises standards such as IEC 61280-4-1 require a launch cord and a tail cord (also called a launch fiber and a receive fiber).
When testing longer fibers, the fiber attenuation dominates and so a launch fiber is not as important. For simple troubleshooting, a launch fiber would not have to be used. If a launch fiber is not used, a short (1 meter) patch cable is still recommended to minimize wear on the end face of the instrument’s OTDR port.
Using a launch fiber and a receive fiber is still a good idea, because it provides the most complete results and some standards may require it for certification. However, the standard launch fiber from Fluke Networks, such as part number NFK3-LAUNCH, may not be appropriate. To determine this, ask three questions:
- How long is the longest fiber under test to be tested?
For this first question, it is important to understand that the longer the fiber is, the more the fiber loss is, and so the more dynamic range is needed to measure the end of the fiber. Dynamic range is increased by using longer test times and/or wider pulse widths, which have wider deadzones. The launch fiber needs to be longer than the deadzone; however, longer launch fibers eat into the budget for length and dynamic range, are more unwieldy to carry and more costly.
Determine the maximum dynamic range needed for testing using the following formula:
Dynamic Range Needed = fiber attenuation (dB/km) x length (km) + event losses (dB)
Use the following table as a rough guideline for determining the appropriate length to use for the dynamic range needed. The instrument will attempt to keep the pulse width compatible with the launch and receive fiber lengths determined through launch and receive compensation, though this may increase the test time.
Dynamic Range Needed (dB)
Typical Pulse Width (ns)
Min. Launch Fiber Length (meters)
10.4 to 13.5
8.5 to 11.4
13.5 to 21.8
11.4 to 19.6
As an example the following table shows the launch fiber length needed for fiber with typical fiber attenuation (≤ 0.35 dB/km at 1310nm; ≤ 0.2 dB/km at 1550nm) and event loss less than 0.8 dB.
Max. Fiber Length (km)
Typical Pulse Width (ns)
Min. Launch Fiber Length (meters)
0 to 27
0 to 38
27 to 36
38 to 53
36 to 60
53 to 94
- Is the fiber under test similar to the launch fiber?
OTDR loss measurements rely on fibers having similar backscatter properties throughout. When dissimilar fibers are mated, the OTDR will measure extra loss (if the further fiber reflects less light) or not enough loss (if the further fiber reflects more light). Sometimes this difference is enough to measure the loss as negative, resulting in a “Gainer”.
It is best to use launch fibers with glass properties that are similar to the fiber it is mated to. The standard Fluke Networks launch fiber is made from Corning SMF-28e fiber. It is suitable for testing most OS1 and OS2 singlemode fiber (B1.1 and B1.3 per IEC 60793-2; G652 per ITU-T), whereas NZD fiber (B4 or G.655) can produce significant gainers of 1 dB.
Even when mating the same type of fiber from the same manufacturer, a difference of 0.1 dB is not uncommon . To account for these differences, bidirectional testing can be used as described in the manuals.
- Is the launch fiber connector compatible with the fiber under test?
One end of the launch fiber should mate to the instrument using a SC style connector. The other end should be a style and polish compatible with the fiber under test. Patch cables between the launch fiber and fiber under test are not recommended, as the insertion loss between the launch and patch cables will be included in the reported overall loss, launch event loss, and receive event loss. For a tail cord (receive fiber), the unconnected end should be terminated with a UPC polish to provide a consistent end location.
Fluke Networks sells launch fibers with SC, ST, LC, and FC style connectors with 130 meter lengths.
If you require a custom launch fiber, be sure it is good quality since it provides a reference for all of the tests. The end faces should be UPC (or APC). Obtain test data on the end face geometry if possible.
A second launch fiber may used as receive fiber, except where standards require them to be different lengths. Although testing with a receive fiber requires more steps, it also provide the most complete results. Also, the unconnected end of the receive fiber does not have to be terminated with a UPC connector; rather, it needs to be terminated with a non-angled polish. [It could be PC, UPC, flat, etc.]
Finally, set an appropriate test limit. The default General Singlemode test limit fails fibers longer than 5 km. To test longer fibers, create a custom OTDR test limit. The manual describes how to do this. If you also created a custom fiber type, be sure select it when creating the custom limit. The custom limit length must be at least as long as the longest fiber to test, or set to N/A to ignore the length when determining the pass/fail status.
In some applications, the overall loss may be a fixed value. In other applications, the overall loss should account for fiber attenuation, connector losses, and splice losses. For instance, a 30 km fiber with 1 splice and just 2 connectors (connected to the launch fiber and receive fiber) might have an overall loss budget for 1550 nm of:
0.25 dB/km x 30 km + 0.75 dB/connector x 2 + 0.3 dB/splice x 1 = 9.3 dB.
For long fibers with wide pulse widths and wide deadzones, short patch cables will not be detected, and so the combined loss of both patch cable connectors will be compared to the limit. If each of the connections within the fiber under test will use a short patch cable then the reflective event loss limit should be set to double the acceptable loss of a single connection.