Measurement of 100G OSNR

using the Signal On/Off Method

with various Test Set OSAs

This document describes how to measure OSNR of a 100G signal which is surrounded by

neighboring channels (100G or 10G) at 50 GHz spacing. One way to accurately measure OSNR in

this case is to first measure the Signal power, then turn off the Transmitter and measure the

ASE noise power in the absence of the Signal. This is referred to as the “On/Off Method” for

measuring OSNR. When using various field test set OSA modules, the proper method for this

measurement will differ slightly. This document gives instructions for how to perform this

measurement using two common test sets: JDSU T-BERD 8000 and EXFO FTB-400.

JDSU T-BERD 8000

First, set the wavelength measurement bounds to the area of interest for the channel you are testing. In

the example images shown, the test channel is ~1534.7 nm.

1. Set the Resolution Bandwidth (RBW) to 0.1 nm, then place Cursor A at the center of the test

channel, as shown below. Note that once the Cursor is placed, the power level at the selected

wavelength is shown in the area above the plotted spectrum.

2. Change the RBW to 0.5 nm (by widening the resolution, you are effectively “integrating” all the

signal power within the 50 GHz channel slot). Record the Signal power at Cursor A (without moving

it from its previous location), as shown below (-15.04 dBm). {Note: the spectrum will look quite

different with the wider RBW since the neighboring channels contain some overlapping spectral

content, which now appears as “peaks” between channels.}

3. Turn off (or block) the signal, and change the RBW to 0.1 nm. When you return to the plotted

spectrum there will be a gap where the signal was, now containing only ASE noise. Measure the

power again at Cursor A, as shown below (-29.84 dBm).

4. Use the two power levels recorded to calculate the OSNR. {Note: The first value measured was

actually Signal+Noise power (S+N), since the Signal was present and so was the Noise power

“underneath” the Signal. The second value measured was Noise power only (N). The equation

given below contains an adjustment factor since the Signal and Noise powers were taken at

different resolutions.} First, convert the measured power values from dBm to Linear (mW):

Then, calculate the OSNR using the following equation (using Linear values):

For the example images given above, (S+N) = -15.04 dBm and N = -29.84 dBm. Thus, the calculated

OSNR is 14.01 dB.

EXFO FTB-400

First, set the wavelength measurement bounds to the area of interest for the channel you are testing. In

the example images shown, the test channel is ~1534.7 nm.

1. Start sweeping, and go to the Measurement Tab. Set Markers A and B to ±0.2 nm from the

center of the test channel, as shown in the image below. These Markers should roughly line up

with the edges of the test channel (without bleeding into the neighboring channels), and the

distance between them should be ~0.4 nm, as shown in the image below. Marker C can be

ignored for this measurement.

2. Record the Integrated Power between Markers A & B, as shown in the image below (in this

example, the value is -20.48 dBm). This is the Signal + Noise power (S+N), since it includes the

integrated power of the test signal as well as the ASE noise power “underneath” the signal.

3. Turn off (or block) the test channel, and record the Integrated Power between Markers A & B

(you should see the test channel missing from the spectrum, and the new Integrated Power will

be lower.) This value is the Noise power (N).

Test Channel

4. Use the two power levels recorded to calculate the OSNR. {Note: The first value measured was

Signal+Noise power (S+N), and the second value measured was Noise power only (N). The

equation given below contains an adjustment factor since the Noise power was measured over

0.4 nm, while the standard definition of OSNR considers the noise within 0.1 nm resolution

bandwidth.} First, convert measured power values from dBm to Linear (mW):

Then, calculate the OSNR using the following equation (using Linear values):