Here's a pretty good image of the impulse response of the fiber (Figure 6):

http://www.jimhayes.com/OTDR/otdrs_c.htm

(also seems a good tutorial)

Note the spike at the fiber tip. We normally expect 4% "Fresnel reflection" from the change in index going from glass fiber to air. (Much less if you angle-cleave the tip.)

What averaging are you referring to? The simplest version would be, send a pulse train at known frequency, observe the impulse response after each pulse, and average the impulse responses for all pulses. This would clean up noise and enhance your signal (as long as you had the same physical setup for each pulse obviously).

Other more elaborate schemes are used to obtain the impulse response, but once you've got it, I would think you just repeatedly average your impulse responses to increase SNR.

@Kevin and Luke : How about if there are two or more connetors that are used in a link of communication using fiber optic? Because as I know, there is a gap between fiber optic and connector, and the gap is air.

Fiber connectors (like fc/pc and such) to exhibit attenuation and reflection. i assume you will have separate it in the time domain.

@Ran : Could you please explain your answer? I still don't understand your answer.. =)

i didnt use ODTR dierctly so if you explain your goals more specifically it would help.

basiclly, any interface will cause refraction and reflection due to boundary contition etc. usually fiber are connected through phisical conatact (PC) and this coupling indroduces losses (as avery fiber coupling) and reflection (fresnell). when u calculate for example the absorption of a fiber you have to consider the connectors losses.

i hope i understood what your question was. :)

I agree with Ran that the connectors will reflect somewhat, so they will show up as spikes in the ODTR impulse responses. To figure out which one is the connector and which one is the fiber tip, you may find that connector returns are smaller, or you may just have to know where your connectors are. In any case, the fiber tip should be the last real return, so that should help ... though there could be ghost returns after, hmm ... I'm betting you'll find the connector returns are lower (normalized) amplitude than the fiber tip. Since the connectors push the fibers into physical contact with each other, there shouldn't be a full air gap, so I think you won't see 4% reflection. But, you won't know until you try.

@Ran : I'm going to make a simulation of how an OTDR works, and I used 3 connectors in that, to show the Fresnel Reflection process, but if it's like that, it should have be a different between the connector that connect fiber to fiber cable and the connector that connect the fiber to air.

@Luke : At the books that I read, I only found about Fresnel Reflection that connect a connector with air (the gap between is air), how about the connector that connect fiber to fiber?

i assume u r right.

Luke is also right that connector-connector is different from from connector-air. i cant tell u exactly how right now.

@Ran : Could you please tell me then? =)

@Joaquin : How can we calculate the glitch power or amplitude? And what is the main different between fiber-connector-fiber and connector-fiber-end? Actually, I'm also a bit confused about the Noise that happen at the end of fiber at OTDR, what caused that?

well i dont remember... nut i can recall that when i wanted to measure the relection of my FBG i took -0.5db fpr every connector.

check me! it should be easy to find.

Might be worth taking a few minutes on one of these videos, I know I learned something!

http://www.youtube.com/watch?v=0h_KHvl0AxI&feature=endscreen&NR=1

http://www.youtube.com/watch?v=4jz7d_IvhF8

I think the best answer on the connectors is: they are IDEALLY full physical contact, in which case there would be no reflection (as Joaquin points out, there's no reflection in a fusion splice since there's no change in index of refraction). BUT they usually do involve crossing some (minimal/partial) air gap, in which case there is some reflection. See 1:49 "Information in the OTDR Display" in the first video for the OTDR view of a splice vs that of a connector.

@Luke : if you say that there is a full physical contact at the connector's joint, why is there a spike at the OTDR when it sense a connector? If you say that there is also an air gap (minimal or partial), then what's the difference with the last connector at the end of fiber (it also have an air gap)?

I'm still a bit confused about how the OTDR knows the end of the fiber and shows the Noise after the end of the fiber.

Okay then, now I have an additional question:

What process does the microprocessor do at the OTDR, so it can give the graphic dB vs km? As I know, the input for the microprocessor are bits and how can microprocessor convert the bits to dB?

The mechanism of detection by OTDR is based on Fresnel reflection that occurs at the end of the fiber which is the result of silica and air interface at the fiber end face.

The optical pulses which are launched by a OTDR to a fiber under test, are reflected back from the end of the fiber via Fresnel reflection and then the aggregate average of pulses are calculated by the microprocessor along with the calculation of the to and fro distances traveled by the pulses that are averaged in a given time. This averaging time is set by the user prior the measurement. More the averaging time, more accurate calculations. Moreover, to add that the fiber loss is calculated on the basis of the same mechanism with a difference that along the fiber under test, the back reflections are due to Rayleigh scattering that are caused by in-homogeneous structure imposed by Germanium dopant in the fiber core.

In optical connectors usually the Fresnel reflection is made minimum by applying index matching oil at the splice joint at the time of production, hence minimum Fresnel reflection. If at all there is a such reflection, it will be negligible as compared with the reflection of the fiber end face. Therefore, the prevailing reflected power when detected by the OTDR would appear as a spike at the output monitor. Moreover, the back-reflection by Rayleigh scattering along the fiber is also small as compared with fiber end reflection.

@Faramarz : Do you know what is the used of microprocessor? I mean, how is the microprocessor works?

From some materials that I read, the inside of OTDR that used as a receiver, consists of Avalanche Photodiode (APD), I-V Converter, Analog to Digital Converter (ADC), microprocessor. The APD converts the power of the light pulse to current, then the current is converted to voltage by I-V Converter and this voltage is converted to bits. Then the bits is processed by microprocessor. And then, there is no further explanation about how the microprocessor converts the bits to the graphic.

Dear Stevan: My dear this question of yours does not relate to OTDR. How a microprocessor work is related to digital electronics which is a general topic not at all related to optics and OTDR. You better read some books on electronic devices. By the way, what is your specialization, i.e., University Degree? If you are an engineer you can easily refer to electronic eng. books. VIVA

@Faramarz : I'm studying Telecommunication Engineering. At first, I want to know how the OTDR works, and now, after I searched and read some materials. the 'brain' of OTDR is microprocessor, and I want to know what process does it do? I have one material that tell that the microprocessor do logarithmic function, but it only tell me that and no further explanation. I want to know, what is the comparison and how?

@Stevan: When the light pulses are reflected back from the fiber under test, they are coverted into electrical pulses by a photodiode located in the OTDR. These electrical pulses are then processed by a microprocessor which is already programmed by a pre-set codes. From this stage onward nothing is related to the OTDR operation but is handled by the electronics of the system. As I already said this part of the job is related to electronic engineering and no optics is involved. Therefore, you should know about electronics and microprocessor programming, that is all. This latter issue is not usually given in the literature which are talking about OTDR, it is solely electronics.

@Faramarz : Just want to ask a little, is it true that the microprocessor does a logarithmic procedure, so the OTDR can show the attenuation graphic?

For every pulse returning back to the processor, its amplitude and the corresponding distance, that is the distance between the input end of the fiber and the returning point of the pulse, is calculated. Therefore, for each point along the fiber length there will be a returned power which is the reduced form of the initial input power launched into fiber, i.e., the returned loss value (RLV). In this way, all the RLV of all the points along the fiber is processed and calculated. At the end, the software (used by the processors) draws the RLV versus the distance (that is the length of fiber). This graph will give the total loss of the fiber in terms of fiber length. Now if you like to call it "logarithmic procedure" or so, does not make any differences.

@Faramarz : As you said that the RLV is processed and calculated by microprocessor, what I want to know is the process that microprocessor do to convert the bits from ADC to a graphic that shows dB vs km. As the unit for the ordinat is dB, so I think that it will do the logarithmic process. Am I right?

@Stevan: Ok, To obtain the curve of RLV versus the fiber length, its value is converted to dB which is [10 * log(input power/output power)]/L, where L is length of the fiber under test.

@Faramarz : Ok then, but how about using 10 log (present power/power before)? Since the OTDR detect backscatter power. And, should it be reversed by microprocessor, since it gets bits from ADC, to become a power in mW?

OK, it is also right

@Faramarz : Should it be reversed by microprocessor, since it gets bits from ADC, to become a power in mW?