1
\$\begingroup\$

I recently bought a SFP fiber transceiver module to play around with. My idea was to connect the transmitter pins to a waveform generator, loop the optical output to the optical input with a fiber cable and then look at the receiver pins on an oscilloscope.

What I saw on the oscilloscope was an ugly 100MHz sine wave, but I only have 100MHz bandwidth on my scope so I assume the measurement is not valid at all.

Anyway, this got me thinking... Is it possible to transmit, let's say, a 1MHz, 50% duty cycle signal using a fiber transceiver rated at 1.063Gb/s, or do you have to transmit and receive at the rated value always?

\$\endgroup\$
6
  • \$\begingroup\$ You should be able to transmit at whatever speed you want. However SFP as far as I know is CML which ideally should be capacitively coupled, so you can't go too low (depends on the size of the caps). 1MHz should work. Try it and see is the simplest suggestion. \$\endgroup\$ Aug 11, 2015 at 22:15
  • \$\begingroup\$ @TomCarpenter Ok, great! That was what I was wondering about. I probably just have to hook everything up a bit better, but now I at least now that it should work at lower speeds. Thanks! \$\endgroup\$
    – sandberg
    Aug 11, 2015 at 22:24
  • \$\begingroup\$ Can you give an actual part number for what you bought? "SFP" just tells the physical form factor. SFP parts are available for a variety of signalling standards (Ethernet, Fibre Channel, ...), fiber types (single-mode and multimode), and link distances (anything from 100 m to 40 km). In particular, the signal standard your part was designed for will have a lot to do with the minimum frequency it'll operate at. \$\endgroup\$
    – The Photon
    Aug 12, 2015 at 0:54
  • \$\begingroup\$ @ThePhoton It's a FTLF8524P2BNL-MD \$\endgroup\$
    – sandberg
    Aug 12, 2015 at 6:17
  • \$\begingroup\$ That part is designed for Fibre Channel, which has very little frequency content below some fraction (1/10 or 1/100) of the baud rate. It's entirely possible it won't work at all at 1 MHz, due to internal ac-coupling. \$\endgroup\$
    – The Photon
    Aug 12, 2015 at 19:28

1 Answer 1

3
\$\begingroup\$

I set up the experiment again and got some positive result so I thought I would answer my own question.

Using both channels on my DG1032Z waveform generator I create a differential square wave signal that is fed into the transmitter side of the fiber module. The receiver side is connected to an oscilloscope.

The waveforms below do not look very clean but that's because I just soldered on some wires to the module. It's probably possible to get cleaner results by using a proper connector for the SFP module and do all the termination correct, but this was just a test to see what actually happens.

Here is the output waveform from a 10MHz input: 10MHz input

1MHz input: 1MHz input

200kHz input: 200kHz input

\$\endgroup\$

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.