I have a question regarding SFP pin requirements. The SFP pin descriptions can be found in the standard document found here.

I'm currently designing a high speed LED transmitter and PIN-TIA receiver with SFP interface. This is meant to be used with a typical simple SFP media converter like the one shown found here. I do not design the media converter, only the SFP-interface transceiver.

In theory, referring to the SFP pins on the SFP host side, I'm only planning to use the Power pins, Ground pins, TD+ and TD- pins, and RD+ and RD- pins.

I do not intend to use the following pins:

  • TX Fault
  • TX Disable
  • Rate Select
  • LOS

My question is, can I just leave the pins above floating? Examples of my concerns are:

  • TX Fault & Disable: If they are floating, would it somehow affect the output of the TD+ and TD- pins? These pins are normally connected to a typical laser driver chip, which I'm not using. I just have an LED (in-house developed) as s transmitter.
  • MOD-DEF: I understand this is meant to connect to an EEPROM for transceiver module details etc, which I'm not interested in in this simple media converter. Does that mean I can just leave it open.
  • LOS: This is normally connected to a LOS pin on the limiting amplifier on a ROSA. But my TIA has a built-in amplifier and does not have a LOS pin. Can i leave this pin floating?
  • Rate Select: Do I simply tie this high to get a 'full bandwidth'? What does 'reduced bandwidth' even mean? Is it 100Mbps vs 1000Mbps kinda thing?
  • \$\begingroup\$ Before you bark too far up this tree, are you aware that the primary reason gigabit SFPs (even short-range multimode ones) use lasers or VCSELs is that LEDs can't (or at least couldn't) manage the datarate beyond 100Mbit? Perhaps you've in-house developed one that can - otherwise, you may have issues with trying to operate an LED-based system that fast - the rest of the industry did... \$\endgroup\$
    – Ecnerwal
    Mar 5, 2014 at 4:46

1 Answer 1


I am answering this based on one particular SFP module's datasheet. I believe that everything I'll say applies in general, but I haven't checked it against the standard itself. I did find a copy of the standard online, but I'm not sure whether this version is current.

  • TX_FAULT: "This signal is an open collector output (pull-up required on the host board). A low signal indicates normal laser operation and a high signal indicates a fault." This implies you should provide a pull-down on this signal, otherwise the host might believe your module has had a fault and stop sending data.

  • TX_DISABLE: This is an input to the SFP module. You may ignore it if you like. If your source is not inherently eye safe, ignoring this input may disable some safety mechanisms (open fiber control), so would not be a good idea for a salable product.

  • MOD-DEF : these are "module-defined" pins, theoretically unique to each vendor, although there are some common uses for these pins that are consistent between vendors. You are not required to do anything in particular with these pins. Some host products, however, might refuse to work with your SFP module if it does not respond as expected to these pins. Mostly they expect an EEPROM to be present and to contain certain module-identification data. This is more likely to be a problem with an expensive router than with the device you linked to.

  • RX_LOS : This is an SFP module output signal. It is held low by the module to indicate a valid signal is received. Leaving it floating is not a good idea, because if it drifts high, the host sytem might stop accepting data from the receiver (assuming it is gibberish).

  • Rate Select : "This is an optional input used to control the receiver bandwidth for compatibility with multiple data rates". Setting this for low bandwidth allows the receiver to reduce its input bandwidth and improve it's sensitivity when a lower data rate input is expected. If your module only has one mode of operation you can ignore this input.

  • \$\begingroup\$ Giving an example for the usage of TX_FAULT and TX_DISABLE, an SFP connected to an FPGA was not emitting light from the TX port. When setting both TX_FAULT and TX_DISABLE signals to zero, The Tx Port on the SFP started emitting light. \$\endgroup\$ Nov 9, 2023 at 14:34
  • 1
    \$\begingroup\$ @KhaledIsmail, It's correct that when using a typical SFP module you must hold TX_DISABLE low in order to obtain an optical output. But this is a question about implementing an SFP module (for a bespoke application), not about making a system to interact with a commercially available module from another company. \$\endgroup\$
    – The Photon
    Nov 10, 2023 at 1:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.