Light coupling in fiber optic cables

[ace1719]

Hi Everyone! I have a photonics problem that I was wondering if anyone may be able to shed some light on (no pun intended).

I work for a company that is designing a fiber optic emitter/detector as a part of one of our products and we've noticed something that would appear to defy physics. Obviously we aren't breaking any physical laws, and my suspicion is that there is a nuance of photonics that we are unaware of.

Basically, we have observed that we are coupling light into a fiber optic cable through its sheathing and cladding, which should not be possible.

Our detector fibers are connected to a photodiode and signal processing hardware. This is how we measure light in them. Our company is ostensibly comprised of electronics engineers, so this part of the system works, is robust and is far away from any source of light.

I'll explain an experiment I did without getting into too many detail as to why this is important. You'll just have to trust me that this is important.

I taped the ends of our detector fibers with electrical tape to ensure that no light could be entering from the fiber tips. I then lay them on a lab bench and shone a 500 mW, 520 nm laser at them. I angled the laser away from the tips to ensure that light would not be entering through the tips.

I observed that when I brought the laser very close to the fibers (the emitter is only 200 um away from the detector fibers in our product), I would observe a voltage at the photodiode. The closer I brought the laser to the tip of the fibers, the larger voltage I would see on the photodiode.

To me (as an electronics engineer), this would indicate that the laser light is being coupled into the fiber through the sheathing and cladding.

It's possible that light could get in through cracks in the sheathing and cladding, but I inspected it, and it looked I was not able to see any defects. This should not be possible, so there must be something else going on. Could anyone provide any insight into this?

 

[Drakkith]

Is it possible some fraction of the light is simply passing through the thin sheathing without being absorbed? I'm absolutely not an expert in fiber optics, but a 500 mW laser is about 100x more powerful than a typical laser pointer, which is already VERY bright. Since the sheathing isn't supposed to protect against anything other than relatively low intensity ambient light, perhaps some of the laser light is getting through? Even a 99.9% reduction in light would still leave upwards of .5 mW of light getting through.

 

[ace1719]

Our laser is a Class IV that can go up to 5 W, so 500 mW is it's minimum setting. The light could certainly penetrate the cladding, but it shouldn't be able to be coupled into the fiber. I'm certainly not a photonics expert either, but my understanding is that it SHOULD either reflect off the surface of the fiber, or transmit through the fiber. I understand that this is a very nuanced question and a deep understanding of real-world photonics is required.

 

[Cthugha]

What kind of fiber are you using and how long is it? Is it single-mode or multi-mode? What is the sensitivity of the detector?

Cladding modes exist (see, e.g., here )and their properties depend among other factors on the refractive indices of the cladding and the overcoat. These are typically damped away after rather short distances on the cm scale, but at high powers or for some fiber designs it is not unrealistic that some light makes it through the fiber. Are you sure that it really couples to the fiber core or is it also possible that the light propagates at the interface between cladding and overcoat?

Also, at such high powers it may be worthwhile to check that the electrical tape is really fully absorbing. If your detector is highly sensitive, it is not unlikely that some light may still make it through.

 

[ace1719]

Our fibers are 50 um, glass core, multi-mode fibers They are 2.88 m long. The photodetectors are exquisitely sensitive, although the system itself is extremely lossy. An irradiance of 11.1 mW/cm^2 will yield a voltage to 2.33 (buffered, but unamplified) at the photodetector.

The proximal end of our fibers have an MPO connector. Even if the light was propagating through the cladding, it still wouldn't transmit into the connector. However it isn't inconceivable that the light propagates through the cladding and then couples into the distil end of the fiber. This would be consistent with my observation that the closer the light source was to the distil end of the fiber, the more light was being couple into it.

I wasn't putting the laser right up against the electrical tape, and I was angling the light away from the distil end of the fiber, although I'm not sure how light would do a U-turn and couple back into the fiber. If this is possible, then this would explain our observations.

In our actual system, our light source is 11 mm from the distil end of the fibers. The three detector fibers, plus the emitter fiber and a 1 mm spherical diffuser are all in a 1.2 mm nylon lumen. So they are indeed in very close proximity.

 

[berkeman]

I observed that when I brought the laser very close to the fibers (the emitter is only 200 um away from the detector fibers in our product), I would observe a voltage at the photodiode. The closer I brought the laser to the tip of the fibers, the larger voltage I would see on the photodiode.

Is the emitter amplitude continuous or modulated at some frequency?

 

[tech99]

It sounds to me that it might be experimental error. But in principle, EM waves can couple across a boundary which is providing total internal reflection. For instance, the "reflecting" faces of two prisms can allow transmission if they are very close. If your laser is exciting a dielectric which is very close to the fibre, then this type of coupling might occur. The electrons in the dielectric vibrate in sympathy with the incoming EM wave and create a local reactive near field, E and B. This then couples to another reactive near field region which closely surrounds the fibre (within a small fraction of a wavelength). The reactive waves involved in this process are referred to as evanescent.

 

[ace1719]

Is the emitter amplitude continuous or modulated at some frequency?

We're running the laser CW.

 

[berkeman]

We're running the laser CW.

Can you run it with AM to see how that affects the RX pickup waveform?

 

[ace1719]

It sounds to me that it might be experimental error. But in principle, EM waves can couple across a boundary which is providing total internal reflection. For instance, the "reflecting" faces of two prisms can allow transmission if they are very close. If your laser is exciting a dielectric which is very close to the fibre, then this type of coupling might occur. The electrons in the dielectric vibrate in sympathy with the incoming EM wave and create a local reactive near field, E and B. This then couples to another reactive near field region which closely surrounds the fibre (within a small fraction of a wavelength). The reactive waves involved in this process are referred to as evanescent.

It isn't experimental error. I didn't run this experiment out of interest. I ran it because it was the logical reduction of a real world problem we were seeing. I went back, and looked at data from our earliest prototypes, and even re-tested some of them. The phenomenon is extremely consistent. Perhaps it may help if I explain our testing procedure.

There are three parts to our testing procedure: testing the emitter, testing the detector, and then testing them as an assembly.

To test the detector, we put it in an integrating sphere, and put 500 mW of optical power into it through a patch cable. This will yield a maximum irradiance of 11.1 mW/cm². My observation is that the observed irradiance is usually around 9.6 mW/cm², which is fine since there are losses. More importantly is that the three detector fibers have a pretty tight distribution. They'll usually be within 2% of each other.

After we make the assembly, we then re-test it in an integrating sphere, but using the emitter rather than a patch cable. Sometimes the detector reads higher than the initial reading, and sometimes it reads lower. Occasionally it reads significantly higher, upwards of 11.1 mW/cm², and as high as 15 mW/cm², which is impossible unless light is getting into the fibers from a means other than reflection from the interior of the integrating sphere.

Out of interest, I've tried switching between using the patch cable and the emitter. Even if the emitter is high, when I use the patch cable, it is usually pretty close to the initial measurement.

One other experiment I tried is that I dipped the detector fibers in black ink. This seemed to mitigate the issue, but not get rid of it altogether.

 

[ace1719]

Can you run it with AM to see how that affects the RX pickup waveform?

Unfortunately our laser can only run CW.

 

[berkeman]

Unfortunately our laser can only run CW.

Fair enough. If you block the beam, does the RX signal change? If so, I support the proposal by @tech99 , if not, it is probably some electrical ground coupling issue.

 

[ace1719]

Fair enough. If you block the beam, does the RX signal change? If so, I support the proposal by @tech99 , if not, it is probably some electrical ground coupling issue.

What do you mean by blocking the beam? If I turn the beam off, the signal goes to 0 (as it should). I can't exactly block the beam from inside the lumen.

 

[berkeman]

Oh, there is no way to block the optical laser beam and leave the laser drive running? That complicates the debug a bit...

 

[ace1719]

What do you mean by blocking the beam? If I turn the beam off, the signal goes to 0 (as it should). I can't exactly block the beam from inside the lumen.

The emitter and the detector fibers are certainly very, very close together. So evanescence is possible (not that I really know what that is).

 

[ace1719]

Oh, there is no way to block the optical laser beam and leave the laser drive running? That complicates the debug a bit...

No, the emitter is buried 11 mm into the lumen, the tips of the detector fibers poke out of it by about 0.1 mm.

 

[ace1719]

I will say that the laser and the optical hardware are not connected electrically. I don't know much about optics, but I do know a lot about grounding, and we certainly aren't injecting noise from a ground loop or the like. The grounding is solid, and the power supplies are low noise.

 

[ace1719]

Oh, there is no way to block the optical laser beam and leave the laser drive running? That complicates the debug a bit...

I don't know whether this is relevant, but the lumen is made out of nylon-12, which has a refractive index of 1.525. Glass has a refractive index of 1.52. I don't know what the refractive index of the cladding and sheathing is, but they could be low enough to allow for evanescence.

 

[sophiecentaur]

The fibre works by Total Internal Reflection, which requires light to enter the end face of the fibre at an appropriate angle. For light hitting the outside of the glass cylinder from outside, some will always be admitted. I suspect that light nearly parallel to the axis of the fibre can enter and then, if there is a finite curvature, it could be captured and travel down the inside of the fibre.
The level of this effect would presumably depend on the quality of the fibre and how multimode it is (width and uniformity of fibre (=£££)). If the OPs fibre is not sold for high speed comms then it may just be that it's good enough for purpose. I've never bought fibre or looked too hard at its spec but that information (crosstalk etc) must be available . In the end, it's down to Signal to Noise or Signal to Interference ratio that's acceptable for the purpose.

Does this make sense?

 

[sophiecentaur]

Is the emitter amplitude continuous or modulated at some frequency?

When I was just a lad at university, I was told of the huge advantage of 'Synchronous Detection' in measurement. It reduces the problem of Low Pass signal filtering to much easier and well defined Band Pass filtering and eliminates all that DC level drift, hum and other low frequency interference.
If the OP can define his needs in real terms, it may be that the problem he's found is not actually a problem. Otherwise, the best way forward could well be to use a modulated system. Very little added hardware and the possibility of using lower power sources.

But I realize this is very much 'arm chair' designing.

 

[DaveE]

Unfortunately our laser can only run CW.

How about an external chopper? Be careful with reflections at 500mW!

 

[Tom.G]

If I understand the physical construction correctly, there are three fibers surrounding a 500mW source, the axes of the four devices are essentially parallel to each other.

The source is behind the plane of the fiber ends.

Some of the source radiation hits the cladding of the fibers, couples into the fibers, and causes a background signal.

If the above reflects your reality, how about a thin metal tube as a shield around the emerging source beam?

My original thought was aluminum foil, for no other reason than availability and workability for testing. Then I noticed your dimensions! Maybe a capillary tube of some sort. Anyhow, something to work as a light tunnel.

Better yet, can you move the emitter beyond the tips of the fibers? Or use a fiber on the emitter to extend past the receiving fibers?

Also, Nylon tends to be a translucent scatterer. If it is not already, try some Black (or Blue) Nylon to reduce translucency. And a matte finish may help on the holder and/or the fiber jackets.

Cheers,
Tom

Interesting problem, please keep us updated on what works or doesn't work!

 

[sophiecentaur]

As with any Engineering project, it’s essential to start off with the required specification.
How relevant is this break through of stray light to the level of the ‘wanted signal’?
How much in hand do you have?
Signal to noise ratio is a fundamental quantity for all measuring equipment and it’s always finite.
The fibre will have been produced for some target purpose. It will have a performance spec.. chasing screening problems may not be the best way forward. How many man-hours are worth expending when there are two available alternatives? Modulating the light source or using a different fibre could solve the ‘problem’ if it is actually a problem.