Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
OTDRs provide extensive information on the condition of your fiber optic lines. They may tell you there is something at 11.457 km with a loss of 2.346 dB and a reflectance of 46.7 dB. But it really doesnt tell you what it is! If you are unfamiliar with OTDR waveforms and particularly if you dont have any cable maps or original test results for comparison it can be difficult to correctly interpret the OTDR. Even the automatic measurement software found in modern OTDRs doesnt identify what the item is, it just gives its characteristics. This application note shows the most common types of shapes, you will see on an OTDR waveform. By combining your knowledge of the system under test, and sometimes a little extra testing, you can identify what the waveform event actually represents on the fiber. Once you have identified the event you can compare it against your system specifications to see if it is good or go fix the problem if it is not acceptable! Tests shown in this application note are available from Tektronix for reference and training. The tests can be viewed using a TFP2A or TFS3031 OTDR, or using the Tektronix OTDR software program, FMTAP. The file name is shown for each picture so you can reference the events on your OTDR or PC. Pictures for this application note were taken with FMTAP.
Figure 1, Fiber Backscatter When you see the backscatter type of line it means there is cable, glass, present. If you dont there is no cable or the OTDR signal has been stopped before reaching the point of interest. The backscatter is important in measurements. Where it stops, that is the distance to the event at that location. Vertical changes in the backscatter is loss or attenuation.
Page 1
indicates a flat precision (polished or cleaved) end. The OTDR is able to test all the way to that connector. This does not mean that the connector itself is good, just that the OTDR can get to it. The only way to completely test a connector is to test through it. To check the quality of the far end connector you have to make and OTDR test from that end, or attach a long jumper cable to the far end connector. 2. Low Reflection End Some systems use angled end connectors (APC connectors). The end of the fiber is cleaved or polished at an angle to reduce return reflections. You need to know if your system used flat (standard or PC) or angled (APC) connectors to know what type of end reflection to expect. The angled connectors will have a reduced Fresnel reflection at the open end (see Figure 3, file APC-end2.cff).
Fresnel reflection
End of fiber
Figure 2, Flat Polish End Reflection When you see a large Fresnel reflection, typically greater than 30 dB reflectance, along with an end to the backscatter this
OTDR Waveform Interpretation
An APC end reflection is typically less that 45 dB reflectance. In some cases you may not see a Fresnel reflection at all. As with the flat polish connector seeing to the far end tells you the fiber reaches the connector but doesnt tell you if that connector is good.
Page 2
3. Non-Reflective End A Fresnel end reflection is caused by a precision end, either cleaved or polished. A broken fiber usually does not have a Fresnel reflection because the end is shattered (it is glass). There is no precision surface to reflect large amounts of light. After the broken end the backscatter slope from the cable drops to the noise level at the bottom of the screen (see Figure 4, file brkend2.cff).
No end reflection
End of fiber
PC reflection
Figure 4, Non-reflective End When you have no end reflection usually this indicates that the light is lost at that point due to a break or bend in the cable. Some very low-reflection APC connectors may look similar to this case. One non-reflective case is actually the end of the OTDR test, not the end of the fiber. If you see the fiber slope continue all the way to the noise level at the bottom of the screen the OTDR has run out of range before it found the end of the fiber (see Figure 5, file rng-end1.cff). Check your OTDR instructions on how to increase the range of its test or check for events that are causing much of the signal to be lost. This may occur on very long fiber or where a connector/splice has excessive loss.
APC reflection
Figure 6, Connector Reflections APC connectors and gel-filled mechanical splices will have a smaller reflection than PC or flat polish connectors (see Figure 6, bottom waveform, file apc-mid1.cff). The APC connector angle reduces the amount of Fresnel reflection. The matching gel in mechanical splices serves a similar, Fresnel reflection reduction function.
Page 3
The OTDR provides loss and reflectance measurements for the connector. The measurements can be compared to system specifications. Zero or very low loss reflections may be an echo of an event rather than a real event. Waveform echoes occur when light re-reflects off mechanical connectors (one of the connectors could be the OTDR connector). Echoes create a reflection that repeats the distance between the connectors. For example, if you have two connectors 1 km apart you may see an echo 2km from the first event. Some OTDRs will warn you if a reflection is a possible echo but most will not.
Figure 7, Splice or Bend Determining if this is a splice or a bend may be critical in fixing a bad (highloss) point. The loss may actually be a bend near the splice (e.g., going into the splice case) rather than the splice. You need to know what you are looking for or you could waste time trying to fix a good splice! On singlemode fiber there is a simple test to differentiate a bend from a splice. One characteristic of a bend is that its loss is higher at 1550nm than at 1310nm. A fusion splice will have similar loss at both wavelengths. It is common to have a bend near a splice so the measurement is a mix of the splice loss and bend loss. For example, the loss of the event seen in Figure 7 was 1.21 dB at 1310nm. However, the same point has 4.75 dB loss at 1550nm! Both tests are shown in Figure 8 (Top waveform 1310nm file bend1310.cff, bottom waveform 1550nm file bend1550.cff). The problem at this point is a bend, not the splice. The bend is at, or close to the splice. OTDRs will group events close together so the bend may be several meters away from the splice and the OTDR is only able to show one loss event.
Page 4
Figure 8, Bend at Two Wavelengths Figure 9, - Gainer Testing the event at 1310 and 1550nm wavelengths is an excellent method of determining the type of fault you are looking at. Bends may also be present in reflective events and will show similar wavelength dependant loss variation. Sorry, this method does not work with the 850/1300nm wavelengths used in multimode systems. Gainers are a function of different (not bad, just different) sections of fiber. At the opposite end of the fiber section from the gainer you will get a larger than normal loss. The amount that the loss measurement is pushed up at the gain, it is pushed down at the opposite end of the fiber section. You may see (in more extreme cases) a whole section of fiber that seems higher than adjacent sections (see Figure 10, file gain-1.cff).
Figure 10, Raised Fiber Section Intrinsic measurement offsets can cause problems in determining if the splice loss meets your system specifications. The gainer measurement cant be correct -- you dont get real gains in joined sections of glass. The extra large loss at the far end of the section is only partially splice loss. You might try to fix a splice where the issue is the cable not the splice. Re-splicing will not fix the large loss measurement in these cases.
Page 5
As with bends, intrinsic events take a little extra testing to resolve. When you see a gainer you know you have intrinsic effects but you dont know how much it is affecting the loss measurements. To resolve the issue, find the true splice loss, you need to test the splice from both ends and average the measurement. The amount the measurement is pushed up from one is the amount is will be pushed down when testing from the opposite direction. By averaging the two measurements you get the correct splice loss, without the intrinsic caused offsets. For example, you may have a splice that measures as a 0.20 dB gain (-0.20 dB loss) from one direction. Testing from the opposite direction the splice measures a 0.30 dB loss. Averaging the result: -0.20 + 0.30 2
2. The 0.30 dB also was not an accurate measurement, though we dont have the gainer clue to realize this if you were only looking at the loss of that event. For example, if your maximum allowable splice loss was 0.2 dB you would have tried to fix a splice that was already good. Many OTDR documentation software programs will do the bi-directional or two-way average for you. They will match up the splices and calculate the average splice loss. Figure 11 shows a bi-directional waveform comparison done with the FMTAP software program. Waveforms shows are A-demo01.cff and Bdemo01.cff that come with the FMTAP software program.
0.10 2
= 0.05 dB
The averaged measurement solves two problems: 1. The 0.20 gain was obviously not an accurate splice loss measurement. Figure 11, Bi-directional Waveforms Figure 12 shows the bi-directional event table from the A and B-demo files.
Page 6