Fiber loss budget for 1000SX Catalyst 6513 and others GBIC.

[bruce.porter]

I am having problems determining the fiber loss power budget for MM fiber running 1000SX on GBICs in a variety of Catalysts including 6513, 3550, 4509. Is the power budget a formula or is there simply a dB value for any end to end link? My network has one satellite patch panel going to a central patch rack and then going to another satellite patch. There are essentially 4 connectors and the mm fiber is 62.5. thx.

[Georg Pauwen]

Hi Bruce,

AFAIK, connectors, splices, and also simply distance affect attenuation. Check out this link for a document that describes these things in detail, hope that helps:

http://www.cisco.com/en/US/partner/tech/tk482/tk607/technologies_tech_note09186a00801492ce.shtml

Regards,

Georg

[bruce.porter]

Thanks, I had seen these pages before but, after looking at the table of device characteristics, I was able to come to a close number for Power Budget. The table gives specs for an HP GBIC.

http://www.cisco.com/en/US/partner/products/hw/routers/ps167/products_tech_note09186a0080093d83.shtml

Observation from table: If you had a worst case transmitter delivering –10dBm of power then you could budget 7dB of loss across the link and still get –17dBm at the receive end. You must, however, factor in approximately 2.5 dB of insertion loss which occurs at the insertion point where the fiber cable mates with the transmitter. The bottom line is that you should have less than 5 dB of loss (for worst case tollerance) on the link and no more than a max of 9 dB (best case transmitter power output tollerance).

[konigl]

Optical link power budget is the worst-case scenario that will still give you 100% reliable communications. The difference between the minimum transmitted power and minimum receive power (or sensitivity) of the GBIC gives you your link power budget.

If the end-to-end sum total dB loss in your fiber span due to connectors, splices, and fiber attenuation at your transmission wavelength is less than or equal to this number, then your link will be 100% reliable.

If your dB loss exceeds the worst-case loss budget, there's still a chance it may work. But the odds decrease as the distance and/or dB loss increase, out to the "best-case scenario" point. It depends on the individual GBICs used: some pairs may work, while others may not.

If you take the best-case figures for your GBIC, and calculate the difference between maximum transmission power and maximum receive power, then this dB loss value is your "best-case scenario" point, beyond which your link will definitely be 0% reliable (or 100% unreliable, for anyone who thought that was a typo).

Having said all that, nobody really wants to depend on a link that's less than 100% reliable, so plan on working with the worst-case scenario.

You can estimate the link loss of a given fiber span, and make your equipment purchases based on that. Or you can build out the fiber and then measure the loss with a power meter or OTDR, and then make your GBIC selections based on actual figures.

If equipment costs are a concern and you're maybe close enough for LX/LH GBICs vs. ZX GBICs, don't buy them until after you measure the loss. And remember to measure at both 1300nm wavelength for the LX/LH GBICs and 1550nm for the ZX GBICs.

I have worked with fiber spans in excess of 10km where the dB loss was still within the worst-case figures for LX/LH GBICs, so that's what we used (and saved several thousand dollars on equipment). No problems on those links.

Hope this helps.

[bruce.porter]

It turns out that the budget is 7.5 dBm for 1000SX GBICs. DBm is the same as dB without the 1 mw reference point of 0dBm. I would like to know whether or not the 7.5 dB, or dBm, value includes insertion loss or not. Insertion loss has been 2.5 dB in the past and is the loss experianced when you insert a fiber connector on the transmitter of the GBIC. (You apparantly loose some power at this point because the light is not all alligned with the fiber and some of it scatters????)

[konigl]

The Link Optical Power Budget is derived exclusively from the worst-case minimum transmit and receive power ratings of the active optical components being used. It does _not_ include any adjustments to take into account losses in the passive fiber span.

[bruce.porter]

It is true that the PB is determined by the bottom of the spec transmitter power output and the bottom of the spec receiver signal requirement but between transmitter and receiver you lose power and eventually you lose more than the power budget which is approximately 7.5 dBm for 1000SX GBICs. When you lose 8 dBm of power, you face the possiblilty of encountering a condition of a low end spec receiver and a low end spec transmitter which will mean - errors. Additionally you may need to subtract an insertion loss from the 7.5 dBm value. This loss is caused by the fact that not all the power measured at the transmitter will line up with the fiber and therefore you lose 2.5 dBm here.