On 20/09/2020 17:08, Paul Thornton wrote:
Oddly can't see Stephen's mail - it's not in my spam folders, so not sure what's happened there.On 20/09/2020 17:04, Stephen Wilcox wrote:I'm curious about the technical feedback.. single ended testing.. I presume that means a loop at the far end then running a test on each strand at the near end.. what artefacts would not be found?(Noting that since they have to go to both ends to deliver they may as well test both ends anyway but testing with a loop is best practice anyway to spot issues at the fibre end)I too wondered about this, especially as they sell this by the strand - so there is no guarantee that you've had a pair provisioned (unless they always put more than you've ordered in, just in case of failure or upgrade requests).
Optically, you can accurately measure loss on a fibre either through insertion loss measurement testing - which involves accurately-referenced kit at both ends, ideally physically referenced to the other end's kit, so often hard to do accurately - or bidirectional OTDR tests, which don't require referencing as such and so can be a bit easier to pull off.
One-way measurements of e.g. splice events can often miss or misinterpret things like mode field diameter mismatches (which can appear to "gain" light when measured one way, but will be an over-estimated "loss" in the other direction, but in true loss terms will actually be about the average of that gain and loss), and will be less accurate than a two-way measurement (which allows you to average out measurement errors and uncertainties).
Doing two-way measurements and combining the results will give you the most accurate measurement in practice. Though using an accurately referenced loss test set for insertion loss measurement will, on paper, give you a less /uncertain /answer, in practice it's rarely done correctly even by experienced operators, and is easy to /subtly/ get wrong. OTDR operator/kit errors are normally visible and obvious in the results (high noise floor from inadequate averaging, lack of launch/receive leads, etc).
In practice, single-ended OTDR measurement - if the OTDR measurement is done right, which will typically involve a few different traces at different wavelengths for long-haul links beyond a dozen km - will give you quite accurate results, typically within 0.5-1dB of true loss on most links, which is Good Enough for most applications. But where you're mixing and matching lots of old and new fibre or going through lots of splice closures etc on longer links, bidirectional testing is definitely sensible.
Testing with a loop is an option but not actually ideal - the best approach for accuracy is to use a launch lead and a receive lead while performing your test (launch typ. 150m, recv typ. 2-3km). This will let you measure the last and first connectors in your link accurately, proves continuity, and gives you the full dynamic range of the OTDR in each direction for your tests. If you have a tech at each end with an OTDR anyway, this is trivial to do, and doesn't require a pair to work (just make sure everyone's got inspection probes for connections onto your shiny new fibre, lest your first commissioning activity be polishing some ferrules or splicing on new connectors...). I've seen arguments in favour of loops because you only need one OTDR, but you still need someone with a microscope to connect the loop up cleanly, and so are generally sending a fibre tech who will have an OTDR anyway.
For any high-capacity links on our network (typically 20-60km hops) we do bidirectional tests w/ launch and receive leads, /and/ do insertion loss measurements as a sanity check (locally referenced with an OLTS at each end, or using a field-referenced light source/power meter pair), and take all that back along with microscope images to prove inspection of and cleanliness of connectors. Using good test equipment with intelligent automation (e.g. EXFO's iOLM, or VIAVI's FibreComplete stuff) helps a lot with variable fibre engineering skills in the field, but isn't a substitute for training and proper support/auditing.
I'll add to the DFX commentary - it does depend somewhat on your use case as to how useful it will be and what pitfalls you might come across. If you're trying to get fibre somewhere where you don't have exchange presence at either end (i.e. you'd take a Cablelink at each end to connect to outside plant, or are trying to explore fibre routes before establishing presence), you'll have to establish a rack before you can get a survey done to tell you if there's any fibre - which is quite a high cost survey!
Cheers, James Harrison
OpenPGP_0x2D36DC09F26C8E69_and_old_rev.asc
Description: application/pgp-keys
OpenPGP_signature
Description: OpenPGP digital signature
