0

Let's say there is a no-name SFP transceiver without a label and without DDM(allows to see manufacturer programmed wavelength data on EEPROM from network device) support. Is it technically possible to detect on what wavelength this SFP transmits? I have used JDSU light-meters previously and they ask user to select wavelength(for example 1310nm, 1550nm), but regardless of the selected wavelength, they will show optical Rx power.

Martin
  • 269
  • 3
  • 11
  • defraction gratings of various sizes used in spectrometry or if you've got an easy way to see the light, you can just sent it through a diffraction grating of known spacing, and measure the spacing between the bright spots. Or, if you want to be more precise, you can use a Michelson interferometer (which is what's in a lot of commercial wavemeters). – Tony Stewart EE75 Oct 12 '16 at 12:30
  • Have you already ruled out the possibility the wavelength is 850 nm? – The Photon Oct 12 '16 at 15:42

3 Answers3

2

While Marcus Miller's answer is entirely correct, if you are driven to answer your question the instrument you want is an OSA (Optical Spectrum Analyzer). These are extremely high-resolution spectrometers, usually employing FTIR spectroscopy. Wavelength accuracies of about 1 ppm are available, but the best units will cost in the 5 figures. eBay is a good place to start looking, although you can start with companies such as HP (now Agilent Oops, now Keysight), Advantest, and Yokogawa. Or, if you're feeling really masochistic, Thorlabs has a very nice selection.

WhatRoughBeast
  • 60,521
  • 2
  • 37
  • 97
  • Agilent is now Keysight, although they don't seem to be making OSA's at the moment. They do have a wavemeter. – The Photon Oct 12 '16 at 15:59
2

For a (relatively) low cost measurement to distinguish crude differences in wavelength, I'd use a differential filter measurement.

You need two attenuating filters with different known attenuation vs. wavelength curves. Collimate the output of your source and input it to a power meter, first with one filter in front and then with the other. From the ratio of photocurrent between the two measurements you should be able to determine, roughly, the source wavelength. Possibly one filter is the "0-filter"; that is, no filter at all.

This should be much lower cost than an optical spectrum analyzer or wavemeter. But it likely won't be good enough to distinguish different WDM wavelengths near 1550 nm, for example.

Even cheaper would be a diffraction grating and an IR beamfinder card.

If you just want to distinguish 850 nm from 1310 and 1550 nm, you can try measuring it with a GaAs or silicon detector. 1310 and 1550 nm should get essentially no response from either of those detectors, while 850 nm should give a strong response. Unfortunately I don't know any readily available detector that responds to 1310 but not to 1550 nm (of course one could easily be made using an appropriate coating on an InGaAs detector).

The Photon
  • 129,671
  • 3
  • 164
  • 309
  • Hah, Diffraction grating and IR beam finder was an idea I had. (my IR viewer response dies at about 1.4 um) Then it struck me this morning. Get a 1.3 um and a 1.5 um LED and use those for a detector! (I think that should work but I've never tried it in the IR.) – George Herold Oct 13 '16 at 15:05
  • @GeorgeHerold, if OP has an IR card that responds to 1.3 um but not 1.5, they might have an even simpler solution in hand. – The Photon Oct 13 '16 at 15:37
0

Let's say there is a no-name SFP transceiver without a label and without DDM(allows to see manufacturer programmed wavelength data on EEPROM from network device) support.

Let's say you can get new SFP transceivers that are well-specced and do have support for this incredibly prolific feature for a couple of Euros and this question is really moot, especially considering used SFP transceivers on Ebay are essentially free.

Is it technically possible to detect on what wavelength this SFP transmits?

Yes, of course. Spectrometry, if you want to do this as a measurement, testing with cheap, used, ebay transceivers if you want to test operationality at the same time and have nearly no budget, self-built photodiode setups that, for example, use a prism to split the stream and where the angle of the highest intensity is related to the wavelength, etc.

In any case, you'd need to make your transceiver transmit for long enough to make a qualified measurement, which is non-trivial considering it won't be speaking to a communication partner. I'd say, no matter how I turn this, there's no way that solving this mystery is worth the cost of an SFP transceiver.

Marcus Müller
  • 94,373
  • 5
  • 139
  • 252