4

I've noticed that fiber-optic SFP modules from Cisco and the like often specify a range of power outputs, such as -3 to -9 dBm. What determines the actual power output we should expect to see from such modules? Is it determined dynamically, or configurable statically, or is it just a manufacturing tolerance and we can't expect to know exactly how much power a given module will emit before we install it?

An example specification is here: http://www.cisco.com/en/US/prod/collateral/modules/ps5455/ps6577/product_data_sheet0900aecd8033f885.html - see "Table 2. Main Optical Parameters".

John Zwinck
  • 281
  • 2
  • 4
  • 17
  • What's your end-goal? What problem are you looking to solve? – ewwhite May 13 '13 at 04:03
  • The problem I'm trying to solve is that in some configurations I get too little light on the receiving side, so I want to know if this is something that software/firmware can resolve, or if the only way to fix it is by buying more/different hardware. – John Zwinck May 13 '13 at 04:31

2 Answers2

4

This is a manufacturing tolerance. It's not configurable or adjustable.

David Schwartz
  • 31,215
  • 2
  • 53
  • 82
  • OK, fair enough. Do you know then why the switch reports the current transmit power output as a specific number within the SFP's range for a particular port? For example, an SFP rated for -3 to -9 dBm shows in the management console as outputting -5 dBm. Does the SFP really know how much it is sending if it can't control it? – John Zwinck May 13 '13 at 10:12
  • @JohnZwinck: Yes. Per [SFF-8472](ftp://ftp.seagate.com/sff/SFF-8472.PDF) the actual output power is measured at the factory and programmed into the SFP. The device actually measures current and estimates its output power based on these measurements made at the factory. – David Schwartz May 13 '13 at 13:37
  • Updated link - SFF document has moved: https://members.snia.org/document/dl/25916 – Nathan Neulinger Sep 16 '19 at 18:01
2

Given the actual watts involved - by definition, negative dBm is < 1mW - the implementation details are probably not important to any problem you're trying to solve in a real-world scenario (unless you're installing tens of thousands of these in a hypersensitive environment).

Edit:

According to this Cisco forums thread this is not configurable via software; I'd suggest contacting your vendor if you have specific issues to see if adjustment is possible on the module itself, or whether different hardware is required.

Andrew
  • 7,772
  • 3
  • 34
  • 43
  • I'm not sure how to interpret your answer. To be clear, I'm not concerned about *saving* power here--I'm wondering if I can somehow make the transmitter use *more* power in configurations that need it (e.g. if the fiber is quite long or has a few joints, but not so long that I necessarily want to switch to long-haul SFPs). – John Zwinck May 13 '13 at 04:33
  • @JohnZwinck I've updated my answer based on that; you may wish to add your use-case to your initial question for clarity. – Andrew May 13 '13 at 04:44