Optical transceiver question

Mailman List Mirror (Read Only)
1

We recently purchased some generic optics from a reputable reseller that
were marketed to reach 60 km.

But what we found, based on the spec sheets, is that it could only reach
that distance if the optics were transmitting on the high side of the
transmit power range.

For example, if the TX range was 0 to +5 dBm and minimum RX power was -20
dB, the designed optical budget should be no more than 20 dB (0 - -20).
Based on the wavelength the appropriate loss would be 0.4 dB/km and results
in only 50 km, not 60 km. To get 60 km it would need 24 dB of link margin,
and that would only be attainable if it was transmitting on the high side,
at +4 dBm.

Is it an industry practice to market distance based on the hot optics, not
on the worst case, which is minimum TX power?

Frank

2

What you're saying is if you purchase ten identical optics with the same
SKU, and put them on a few hundred meters of coiled SC/UPC to SC/UPC
simplex fiber and an optical power meter on the other end, they're showing
varying real world Tx powers from between +0 to +5dBm?

That's not right at all, they're supposed to be sorted at the factory by
their actual optical power output before they have labels put on them.

3

The typical situation of the vendor is that the link-budget of the
transceiver considers the worst scenario for TX and loss of dBs generated
by time of operation of the laser, standard attenuation of the fiber, how
it changes in how old it is,... in other ways, the calc ispessimistic.

In my experience, the transceivers have a variation in TX power from
1-3 dBs in worst cases.
In short-haul links doesn't matter very much. In long-haul is something to
consider.

Nicolas./~

4

We have seen cases where the patches introduce enough loss to cause a lot of loss. Have you done an OTDR on each link?

Jared Mauch

5

Not buying fresh veggies here... All optics have about a 5 db range that the vendor will say it is good. The better venders stamp the output power on the optics but not all do this... What he said is to achieve the 60 Km selling point you would have to have all the optic be on the high side of the db TX power... I have never heard of a 60 Km rated optics and it would seem they should be saying 40 to 60 not just 60. It would be nice to say I only want the optics that have an output on the high side and will accept only a 1 db variance but have never seen that in reality. Most are in the middle..

Robert Jacobs | Network Director/Architect

Direct: 832-615-7742
Main: 832-615-8000
Fax: 713-510-1650

5959 Corporate Dr. Suite 3300; Houston, TX 77036

A Certified Woman-Owned Business

24x7x365 Customer Support: 832-615-8000 | support@pslightwave.com
This electronic message contains information from Phonoscope Lightwave which may be privileged and confidential. The information is intended to be for the use of individual(s) or entity named above. If you are not the intended recipient, any disclosure, copying, distribution or use of the contents of this information is prohibited. If you have received this electronic message in error, please notify me by telephone or e-mail immediately.

6

It's a bit like car fuel efficiency values, even with reputable brands :slight_smile:

In this industry, the number of kms for such optics is a best case approximation of the combination of (most notably) those elements:
worst case power budget, capability to deal with chromatic scattering on this length without compensation, with perfect fiber attenuation values without connectors/splicings (by the way, per km attenuation depends of the type of fiber ; so it's really possible to have less than 0.4dB/km even near 1310nm).

I confirm it's normal for several optics of the same model to have a large panel of Tx values as soon as they are within the guaranteed Tx range.
The actual value of Tx is only guaranteed to be somewhere within the Tx range shown in the specs (and it will vary/lower during the life of the optic). So the power budget of an optic is the value in the worst case.

If you cannot OTDR the link (or just put a known source at one side and a power meter at the other side), then you have to estimate: take the length and the attenuation per km for the wavelength, add attenuation values for connectors (crossconnects, patches) and current splicings and so on, take a margin (at least 3dB let's say ?) for later splicings or small defects, and then you obtain the minimum power budget.

10 years later, this page is still relevant:
http://www.cisco.com/c/en/us/support/docs/optical-networking/ons-15454-sonet-multiservice-provisioning-platform-mspp/27042-max-att-27042.html

Anyway, you must not use the km value for anything else than sorting the answers before looking at the specs :slight_smile:

7

There are a bunch of variables that impact actual power needed vs. road-miles, number of cross-connects, type of fiber, amount of slack, type of connectors, frequency, dispersion, etc. The km notation simplifies the naming convention. As a general rule 40Km budget 20db, 60km budget 24db, and 80km budget 28db. Best practice is design and deploy to stay a minimum of 4db above the Minimum Input Sensitivity, and 4db below the Max Receive Level to account for the change as the optic and glass as it ages. If you are engineering a 60km fiber length, you'd step up to the next higher optic to account for other losses. 60KM is just a rough number.

There is another value to worry about as well. It's the De-Assert Level; an optic will run all the way down to its minimum input sensitivity, sometimes a couple db below. But once it hit's the De-assert level, the optic will stop forwarding what it receives. So there is a little more fudge in the numbers if you're not playing by the rules.

8

Frank Bulk wrote:

We recently purchased some generic optics from a reputable reseller that
were marketed to reach 60 km.

transceivers don't work like that. They are sold with a specific
optical budget, normally rated in dB at a specific wavelength. The km
equivalent is usually based on G.652 fibre at an attenuation of 0.25dB
per km at 1550nm, or 0.35dB loss at 1330 nm.

This means that your optical link needs to fit within the optical budget
of your transceivers. If you put a pair of muxes in the middle, you
need to subtract the mux- and demux attenuation from your budget. Each
patch should be assumed to lose ~0.5dB, each splice should assume 0.1dB
and you need to add in a 2dB overhead for loss over time.

You need to feed all these figures into an optical budget calculator and
see what it comes out with. There are plenty online, some good, some not
so good.

If you're using a technology which is affect by dispersion, you may need
to subtract a couple more dB to compensate for this, or if the link is
too long or the technology requires it (10G dwdm or non coherent 100G),
you may need to implement optical dispersion compensation.

When dealing with optical links of this sort of size, the first and most
important thing to start out with is an accurate characterisation of
your optical link, end-to-end, at multiple wavelengths, but particularly
at 1550nm because that is the reference point for C band WDM. Your
provider should provide this as part of the hand-off process. Once you
have this data, you should be able to work out what you can support on
it. If you don't have this data, you cannot ever know for sure what
your optical link can support.

For more information, check out Richard Steenbergen's presentations on
optical design at https://www.nanog.org/resources/tutorials:

Optical Networking 101 (Feb 4 2013)
Everything You Always Wanted to Know About Optical Networking (June 13,
2016)

These are both excellent introductions to optical design, and it's worth
spending the time watching the youtube videos associated with these talks.

Nick

9

No. If this is 1310nm optics with 0.4dB/km budget, the budget figure should be end-of-life figure, ie worst case according to the specs.

I don't like the "kilometer" figures, that can be marketed with very optimistic figures. However, if the transceiver says 0 to -5 transmit, if it doesn't transmit 0 to -5 then it's out of spec.

I treat the kilometer figure as "marketing", and look only at the optical specifications. So using your figures, if the device doesn't have 0 to -5 out, and can receive error free at -20, then it's out of spec and it should be replaced free of charge.

However, if they market 1310nm with 15dB link budget at 60km reach, then I'd consder that false marketing.

10

In discussions with the reseller he admitted that they market the distance
based on average TX power and average link loss, so it is possible to
purchase optics that may not be able to attain certain necessary link
budgets and therefore distances.

There are 1270/1310 nm BiDi optics with a worst-case link margin of 19 dB.
Assuming a loss of 0.38 dB/km
(https://www.cisco.com/c/en/us/support/docs/optical-networking/ons-15454-son
et-multiservice-provisioning-platform-mspp/27042-max-att-27042.html) that's
just 50 km. Of course, with a link margin of 24 dB that would be 61.5 km.
So unless you assume best-cast scenarios, 60 km is a stretch.

Frank