Absolute vs. Relative Group Delay on an OVA

The OVA shows you the test device data in both the Time Delay Domain and the Optical Frequency Domain; the data sets in the different domains are related by a Fourier Transform.   If we scanned a jumper in transmission and looked at the result in the Time Delay domain, we would expect to see a single narrow peak with a peak center along the x-axis equal to the average of the Group Delay spectral distribution.

In the OVA software, however, we are limited to looking at only a ~6.2 ns wide time delay window, although the center of the window can be moved over a ~150 ns delay range.  This limitation is a consequence of the method we use to obtain the full polarization response of the test device in a single sweep of the instrument’s tunable laser.  The Group Delay spectral results and the Time Domain Amplitude results are relative; the 0 delay value on the y-axis for GD as a function of spectrum or the x-axis for the Time Domain Amplitude plot corresponds to the beginning of the ~6.2 ns time delay domain window.  To get the absolute Group Delay value, we need to add back in terms related to the window width and how far the window was shifted out.

This time domain window extent limit isn’t present in the OFDR software package that can be run with the same instrument, although the OFDR results only include the response to a single polarization state.  So when we look at results for the same test device in the OFDR software, the GD correction isn’t needed.  The downside is that we are only looking a the device response to a single polarization state, so we don’t have enough information to calculate PMD or PDL.

The equation that describes the offset between relative and absolute Group Delay values is:

GD_a = GD_r – (t_win/2) + n_dut L_dut / c

Where:

GD_a  Absolute Group Delay.

GD_r  Relative Group Delay values as displayed in the Group Delay window in the OVA user interface.

t_win  Time window segment length, which you can get from the extent of the Time Domain amplitude plot.  This value is determined by the delay of a fiber path inside the instrument, and is slightly different for every instrument.

n_dut   Index of refraction for the device under test length calculation; this is set in the instrument software to 1.5 by default

L_dut  Length of device under test length as indicated on the software GUI front panel, or from a saved .bin file header.

c  Speed of light, 0.2998 m / ns.

Length of DUT and Time window segment length are also indicated in a details window that will appear when you click the Details button in the Matrix Memory Functions section of the OVA user interface:

Consider a simple example: the Group Delay for a ~ 8 m long single mode jumper.

In the OVA software, when we perform the “find DUT Length” operation and take a scan, and bring up the Time Domain Amplitude and the Group Delay plot we see:

In the Time Domain Amplitude plot we see that the center of the device time impulse response has a delay of 3.1321 ns; we see the same GD value near the center of the wavelength scan range.

However, because this value for Group Delay is relative to the start of the time domain window, we need to use the window information and the DUT Length values to calculate an absolute value for Group Delay:

GD_a = GD_r – (t_win/2) + n_dut L_dut / c = 

GD_a = 3.1321 – (6.259 / 2) + 1.5(7.995)/0.2998

GD_a = 40.0043 ns

This value would be for a wavelength near the center of the sweep wavelength range; as we see from the Group Delay spectral plot the delay does vary with wavelength.

If we scan the same test device in the OFDR software, we don’t have to compute a correction related to the OVA software’s time domain window size and location – we can see the scan results over the entire time domain range of the instrument in a single plot.

Here are the OFDR results with the entire time delay domain extent in the upper graph, and with the x-axis blown up around the time impulse response peak in the lower graph:

We see a peak center at 40.0044 ns.  The 0.1 ps discrepancy could come from rounding errors in our delay offset calculation, or from a slight temperature change in the test jumper in the ~ 20 min between the scans.

The OFDR software also can plot the Group Delay spectral curve in the lower graph:

Again, we see a slight slope to the GD curve, indicating a difference in delay through the test device with wavelength.

The OVA and OFDR Group Delay results are corrected for the standard dispersion value of -18 ps / (nm km) expected in telecom grade single mode fiber.  For an 8 m SMF jumper, we would expect a GD variation of - 18 ps / (nm km)  x 42 nm x 0.008 km = - 6 ps over the 42 nm scan range.  The small positive slope we observe indicates that the fiber under test had slightly less chromatic dispersion than the standard value.  The Group Delay slope is corrected for standard telecom single mode fiber dispersion so that time delay domain peaks aren’t unnecessarily broadened.  When examining devices with higher levels of chromatic dispersion that broaden out the time domain impulse response peaks, the user can alter instrument dispersion correction; see the “Pulse Compression” feature notes at the end of Chapter 4 of the OVA User Guide, or the “PulseCompression_OVA_EN-FY1501” application note.