Achieving precision in optical measurements
by Rich Racinskas
You have your favorite fiber test box and trust it for great measurements performed via more buttons and digits than you’ve ever used before. Are you getting the performance you paid for or even what the sales rep promised? Probably not.
Fiber is not just copper with a different color jacket. To obtain proper measurements, you must treat it with the utmost care within an almost clean-room environment when it comes to connections. Consistent, accurate, and repeatable measurements take knowledge, a few tools, and a lot of vigilance. This article reviews some of the primary sources of problems when it comes to obtaining precise measurements, matched with some suggestions about how to solve them.
The instrument starts at the far end of the patch cable. Your $50,000 box can behave like a $300 knockoff if you don’t have a proper cable. You wouldn’t put a $50 probe on a fancy 10-GHz scope, would you? The same principle applies here. A cheap, damaged, or dirty patch cable will throw power measurements way off. If you remate connectors several times and see a 0.5-to-1-dB reading bounce, you need to clean or replace them. If all is well, you should see 0.02-dB or less bounce each time. Since it is difficult to clean mating fiber adapters, one idea is to leave a 1-m patch cord on. Choose a reference or a calibrated grade and avoid generics.
Mating most any connector to a typical power meter should yield accurate readings due to the size of the photodetector. But source connections require precision mating to achieve maximum results. Remember you are lining up tiny lenses in three dimensions close to millionths of an inch. You’ll also need a good bulkhead adapter. A cheap, worn-out one won’t do. Pick up a nice ceramic adapter and keep it capped. For critical measurements, always clean the adapters first using only a fiber-grade stick cleaner. Clean transceivers as well, but be careful-too much force or excess cleaning liquid may damage the internal lens.
Laser sources and other similar gear can become unstable as the connector tip takes a beating. Often a factory repolish will restore performance. Adding a sacrificial patch cable is a cheaper solution. The idea is to reduce instrument matings and provide a connector that is easier to clean and inspect.
Don’t store cables tightly wound or take measurements that way. They will take that shape and cause nonlinear losses, especially at the 1,490/1,550-nm wavelengths. Verify cables by gently moving them during measurements. Replace them after a few hundred insertions. Always clean both ends!
What’s wrong with generic patch cables? Plenty. If you need precision in a lab or production environment, they may add 0-1 dB of unstable loss. The Purchasing Department may treat patch cables as cheap commodities, but they are not all the same. Only about one-third of generics are usable for lab purposes. So after cleaning and hand-testing each one, where is the savings?
For example, commodity ferrules are drilled out to larger holes for ease of assembly. This process causes the fiber to be off-center and creates excess loss and inconsistency. Some are not crimped properly, so twisting the connector will cause stress attenuation. The problem here is that remating them will cause inaccuracies due to the mechanical misalignment. You may find most patch cables are dirty right out of the bag, so check them. A better grade comes with factory test data. Remember the Reagan Cold War doctrine: “Trust but verify.”
You can buy a quality patch cable with very low insertion loss (IL) for about $20. For calibration labs and precise requirements, use a “tuned,” more expensive master-grade cable, which is available from a variety of suppliers. Label and treasure these cables, as they may get “borrowed” permanently to fix unexplainable problems elsewhere.
There are other potential sources of problems that deserve a quick review.
Caps: Caps are a major source of problems. They may be dirty and will outgas onto the surface. Clean them and store extras in a clean bag. Always cap fiber ends. Just the electrostatic charge on the ferrule alone will attract airborne contaminants.
Instrument adapters: After a number of insertions, instrument adapters can become a hotbed of contamination, so clean them. For lab use, I have gone as far as using an ultrasonic cleaner to get my repeatability back.
Optical inspection: The critical core area on a singlemode connector is 9 µm wide, or 0.0003 inches. One human hair is huge by contrast. A speck of dirt, residue, or even smoke you can’t see may cause serious attenuation, backreflection, and even permanent damage. Simply cleaning the connector may not be enough. Use a 400X scope with a clean adapter on it to inspect your work before mating. The connector may be scratched or even burned by EDFA power levels.
One important note is that negligible dirt may not be measurable for insertion loss, producing, let’s say, 0.5 dB. But it can reduce your backreflection by 10 to 30 dB. This is bad news for AM video systems or single-wavelength data. Even worse, you may change to passing data by simply remating the connectors once or twice. Your RF and reflection measurements may be compromised. Back at your customer’s site, the dirt is now optimized and your “FAIL” LED is just waiting for the tech to leave.
From the discussion so far, it’s clear that cleaning is an essential aspect of ensuring precision measurements. However, the methods (or lack of methods) for cleaning I have witnessed are amazing. I have seen canned air, shop air, window cleaner, rubbing alcohol, clothing, etc., used to clean the micro-miniature, highly precision-polished fiber ends. Would you do this to your expensive eyeglasses?
A simple, consistent (preferably dry) cleaning method is the key to low losses and accurate measurements. I don’t recommend most wet cleaners due to added complexity and because they require a dry wipe to remove film or water spots from evaporation. Using alcohol known as IPA is dangerous. The use of flux bottles is another common contamination problem because the alcohol readily absorbs water and after a while may look more like brandy due to dirt.
What’s the best cleaning method? It depends. In general, labs and production lines need a light cleaning for dust, but the field and service centers may need a heavier and repeated cleaning. Proper cleaning is worth another article in itself [which you can look for in June-Ed.]. For now, your best bet is to see the effects of incorrect cleaning for yourself with a 400X scope-then change that procedure or supplier.
If you haven’t established a serious fiber inspection and cleaning procedure yet, please start now. You have one for ESD, yet fiber cleanliness can be a much more critical and costly issue. Westover Scientific has released a free informative video on this topic called “Inspect Before You Connect!” (Download at www.westoverfiber.com/Support/downloads.php.) The major telecoms estimate that 65% to 75% of optical field issues are attributed to dirty fiber. So either inspect and clean proactively or hire more field staff.
It’s particularly tempting to skimp on cleaning in the field. A list of rationalizations might look like this:
- It was cleaned at the factory.
- I don’t need to clean it unless the link fails.
- I don’t do the fiber optics. I’m only an installer.
- My cleaner is empty right now.
- I only had a one-hour class on it.
- Look-it’s not dirty!
All of these excuses are dangerous to accept. Fibers in the field are the most susceptible to contaminants such as dirt, gels, oils, etc. Clean both ends prior to mating during an installation. One industry secret is how many boards are returned and tested as NTF (no trouble found) once the fiber is simply cleaned.
I can underscore the importance of cleaning in the field from my own painful experience with several FTTH service calls. The installers redid my coax needlessly, but I simply cleaned the fibers and my movie channels are HD-grade for the first time. They needed an optical power meter instead of the cable-TV one they used. I then ended up teaching a class on cleaning and inspection in my dining room.
Had you been in my dining room at the time, you might have seen something like Photo 1, which shows what a simple fingerprint (with clean hands, no less) looks like under 400X magnification. You can’t see this level of contamination with the naked eye. After cleaning, the core area is well defined and ready for low-loss connections.
These issues don’t stop with your own staff. Take Joe’s Discount Calibration, who comes to your door with offers of big cost savings. It takes real care and skill to do fiber calibrations. Do they clean the lens, adapters, etc.? Do they use reference-grade patch cables? Are they trained in fiber?
Some vendors will just “sticker” your box and “cal” the dirt out for you. Your readings will vary accordingly and may be worse than before. If you clean the lens now, your readings will be higher and “off.” Quality gear should not require regular tweaks if kept clean. Take before and after measurements. Visit the calibration house. Their gear should be better than yours-you may be surprised what some use. See if they know how to clean and inspect fiber per written procedure; it makes a huge difference. Have a third party or instrument handy for comparison. Stick to a reputable service or use factory calibration otherwise.
You say you’ve outsourced the fiber manufacturing and now you don’t have to worry about it? Here are a few points to consider.
If your vendor uses video inspection systems, it’s easy to tweak the brightness and contrast down to get the yield up. Critical core area damage or contamination may never be spotted. Are you really expecting production-level folks to measure a 27.2-µm blob in Region B as a pass/fail per your procedure? A preferred method is to use a computerized, nonsubjective USB-based system (see Photo 2). Besides offering good results, they’re great news for those involved with fiber inspection who don’t like grease pencils on screens.
There are other factors to consider when working with an outside vendor. Are they using a reputable calibration house that knows fiber? Are they replacing patch cords when they wear out? Are they cleaning each time? You will find that most contract manufacturers won’t spend the money on replacement patch cords or cleaners unless required. Proper long-term measurements dictate repeatable conditions. Only audits and enforcement will keep your measurements accurate.
I am amazed at how little proper hands-on training takes place. I have seen incredible yield changes-from 50% to more than 90%-by simply training employees on production lines. Showing a fingerprint on a large monitor leaves a lasting impression.
Even expensive equipment will give you erroneous results if not used correctly. Generally, engineers specify nice gear for measurements, but they trust someone else will write the procedures, do the proper training for the techs and installers, and enforce the process for long-term quality. But it just doesn’t happen often. We have ESD audits and procedures due to heightened industry awareness. It’s time for more fiber awareness from management on down.
Getting peak performance from your optical test gear requires knowledge and diligence. I hope I raised your awareness of what it takes to make accurate lab, field, and service measurements. The best part is that being meticulous in your measurements will increase your company’s quality and bottom line. Plus it gives you time to enjoy all those new movie channels fiber provides.
Rick Racinskas is a senior engineer at Tellabs (www.tellabs.com), where he supports global manufacturing for the company’s access products. Racinskas has more than 20 years of experience in fiber-optic design, test, and manufacturing. He holds five patents, a bachelor’s degree in engineering technology from the University of South Florida, and is a member of the international Electronics Manufacturing Initiative (iNEMI). He can be reached at firstname.lastname@example.org.