SFP 10G LR vs SFP 10G ER - Engineering Question

[anthonywilson5]

Greetings,

I have a few question as it pertain to SFPs and Distance. The current scenario is as follows:

We have switches at training locations that are approx. 20-30 Kilometers from our core. We currently use SFP-10-LRs for most of our applications and the rest are SFP-1G. After using an OTDR and determining that the entire path is within the margin of error. why would it be more efficient to use the SFP-10G-ER? I have looked at the specs for the LR and ER and it would seem that the ER would be the best in this scenario, since it is capable of 40 kilometers vs 10 kilometers with the SFP-10G-LR. If the fiber path is good, do you foresee any issues with using an SFP-10G-LR if it is currently working? What long term issues could happen if not using the correct SFP based on total distance? Is there somewhere within the Cisco Realm (in writing) that recommends using the correct SFP based on distance?  Thanks.

Respectfully,

Tony

[Philip D'Ath]

Of course Cisco are going to recommend using the correct SFP.  That is a given.  Lets put it another way.  Do you think Cisco would put in writing that you should use the wrong SFP?

If you have found it is already working then you might as well keep using it.

[Joanna -10Gtek]

Hello there,

I think you do not need to be tangled for this problem, only need to select the corresponding transceiver according to the distance.

Depending on your situation, you must use SFP-ER instead of SFP-LR for a distance of 20-30KM, which will cause the connection to be unavoidable.

Hope it's helpful for u.

Isabell

10Gtek

[fiberstorejames]

Cisco has the strict spec of all their devices: 10G LR is made for max distance of 10km, while ER is within 40km.

You may need SFP-10G-ER for your situation. 

[Bill Ferguson · Accepted solution]

If it's working with LR, and you're not RX'ing at -15 dBm on each side, don't worry about it.  The documented Rx minimum for 10GBASE-LR is -14.4 dBm.  So if you're at -13 dBm or higher, you're fine.  If you're at -15 dBm, then you should swap them for ER.

It's not an issue of efficiency.  It's about laser power and span loss.  ER transmits with more power, and its loss budget is better.  We budget to 0.35dB/km for 1310nm (LR), but only 0.25dB/km for 1550nm (ER and ZR).

In my experience, the conservative 10km and 40km (and 80km) limits are given for two reasons:

1. Dummy-proof the engineering.  We usually know the span lengths.  We don't always know the span loss.  This also lets us ignore chromatic dispersion.

2. Limitation of liability.  If it's 11km or 41km and it doesn't work, Cisco can always fall back on the specs.  That said, TAC never asks what the distance is.  They just check the RX levels.

Again, in my experience, span loss is what matters.  I have successfully used LR optics on clean fiber spans significantly exceeding 10km.  As long as you're above the minimum RX value, it will work.

Aside: 10km to 40km is a wide range.  I get a lot of success (and save a little money) in the middle of this range (15-30 km) by using ER on one side and LR on the other side.  You don't need to match the optics.  Again, being above the minimum RX threshold is what's important.

***Yes, I know that it's important to consider dispersion, but I've never seen CD be as significant as span loss until we exceed 80km on G.652 fiber.

[Marvin Rhoads]

wcferg  

Nicely written reply. 

[anthonywilson5]

Bill,

   Thank you very much for the valuable feedback. This is exaclty what I was looking for.