Optical Time Domain Reflectometer

[samarsingla]

Hi all, I am trying to build an Optical Time Domain Reflectometer. Does anyone know of any source on the internet where I can find some info about it? Please help.
Thanx

 

[Analog]

did you try the website:

 

[samarsingla]

Ya that was the first and obvious thing to do but I mostly got commercial links not tutorials... If someone can help, I would be thankful..

 

[evandude]

Are you aware of what you're getting into? It looks to me like even USED OTDR's sell for $1500-3000, do you really expect to be able to build one if you don't even know anything about them?

 

[samarsingla]

Well. thats the reason why..

 

[evandude]

Let me reiterate. This sounds like the kind of project that is far too complicated to just do some quick web research and then throw together; one of those projects where if you actually had the knowledge to successfully build it, you wouldn't be asking such a basic question like this.

The speed of light in a vacuum is 3x10^8 meters per second, and in an optical fiber with an index of refraction of 1.5 or so, would still be around 2x10^8 m/s, which would mean it would take 5 nanoseconds to travel a meter, so you would need resolution on the order of nanoseconds or less to be able to detect any kind of faults with a resolution of a meter or less.

Before you even think about the optical parts involved, perhaps you had better spend some time thinking hard about how you plan to measure an ANALOG signal (the reflected light intensity) at that kind of rate... and then perhaps you might realize why these things cost thousands of dollars to buy commercially, and why you don't see hobbyists building them.

 

[Sceadwian]

Just reading the wiki-pedia entry for it pretty well indicates it's way over the head of a hobbyist to create one of these. At least not for anything less than it would cost to go out and buy one, typical hobby problem. Reminds me of a guy that built an Electron microscope in his garage though, though he was retired and had large amounts of money to spend =>

 

[samarsingla]

Suppose I want to measure only lenghths of fibres greater the 10 kilometers with an accuracy of 1 km, then the least time interval involved is abt 3 micro seconds. Now, I think a 16 mhz microcontroller should be able to handle it. how should i time the events, ie sending and receiving of pulses. Any ideas??

 

[evandude]

well now you're just getting into extremely basic microcontroller stuff. If you can't think of how to time some pulses, you really need to go spend some time learning to use whatever microcontroller you intend to use, because it's a very common and basic task.

However, have you actually tried this at all? Do you know the magnitude of the reflection you're going to see? Properly detecting such a reflection is still the harder part of the project. If you get a nice strong reflection from 10km away then it shouldn't be too bad, but if you're expecting reflections from tiny imperfections/damage in the fiber, which could be very low amplitude, you may still have a lot more of a problem. And if you actually need to measure these reflections in an analog manner, rather than just a digital indication of a reflection or not, you will have even more of a problem, because a simple 16MHz microcontroller won't be able to sample an analog signal nearly fast enough.

 

[Sceadwian]

At 16mhz the time constant is only 62.5 nano seconds.
Asuming you could come up with a decent emitter and detector you would be able to time length intervals at about 19 meter increments. The cable would have to be at least as long as the routine that switches between the light pulse and the actual moment it's waiting for the detection. Asume that's 10 single cycle instructions you'd be able to tell cable length from about 20 decameters up to the detectors limit to receive the transmitters signal pulse in 19 meter increments.

 

[fat-tony]

If you push for something a little quicker, you might want to consider an ARM thumb. Still not overly complicated to use and can run way faster.

 

[Sceadwian]

Even high speed glue logic would allow at least the first echo to be received at a higher clock rate. Getting a nice precise clock would be a problem though as clock jitter will scew the results you get.

 

[mvs sarma]

Suppose I want to measure only lenghths of fibres greater the 10 kilometers with an accuracy of 1 km, ................ . Any ideas??

Hi Samarsingla,

If you are already an expert of Optical instrumentation and and related electronics and physics, Only then, i suugest, you may take up designing of OTDR instrument.

I was ofcourse an end user of the same, before i retired form Telecom field in India. you might be an institutional researcher , it is ok , otherwise , i wish, you must be rich enough to be able to spend for the project cost and its iterations beore a final design.

As of today, the resolution of 1Km in 10Km appears loose spec as better reults are being achieved with exhisting test eqpt.

i am not discouraging you, but feel an ambtious project----

 

[samarsingla]

Though what you say is very true, it is going to be a very crude instrument as compared to what is available today, but still i will learn a lot in the process and i am thinking of taking up this project as part of a semester course so i dont stand to loose anything. As far as optical instrumentation part is concerned, I have many excellent teachers to guide me in that part, Electronics is where I would need some help. Anyways, thanks to all for all theinformation. I will definitely out some info here when i am through with building something. Anyone knows something else, please share it here.

 

[Sceadwian]

It's nice to know you built it yourself as well. Any useable results no matter how crude would give you a good grade I think. If you think of it when your project is done post the results here, or to my e-mail at sceadwian@yahoo.com

 

[samarsingla]

i surely will...