All-optical integrated system samples with 1 ps temporal resolution


A new device for characterizing very fast, repetitive, low-power pulse streams has been developed at the Swiss Federal Institute of Technology in Zurich (ETHZ). The new system, developed at the Institute of Quantum Electronics uses optical gating in a Mach-Zender interferometer switch to allow very short time-slices of an incoming signal—in the subpicosecond range if necessary—to be detected by a slow photodiode. By then sampling over a large enough number of repetitions, a picture of the entire waveform can be built up.

The system is an indium phosphide waveguide interferometer with semiconductor optical amplifiers (SOA) in each of the two main arms (see figure). Depending on the phase relationship between the two parts of the optical signal propagating through the device, the light will be diverted to just one of two output ports by constructive interference. What the phase relationship is, and therefore which port the signal will emerge from, is modulated by the SOAs.0701notes5

For this process to work, the main interferometer arms each have two inputs. The first is for the optical signal to be characterized, and the second is for the optical control pulse for the SOA. These control pulses induce phase shift in the SOA, thus causing the interferometer to switch. To perform the sampling, two pulses are sent into the device with a short delay, thus causing the optical signal to be switched very briefly to the sampling port where it can be detected. Crucially, the time gating is achieved by sending the control pulses into different arms of the interferometer and, therefore, into different SOAs. This is important because the SOA relaxation time is 80 ps—too long to allow it to provide the necessary resolution.

To test their system, the ETHZ team generated a 320-Gbit/s pulse stream, split it, and then recombined the two beams with a 3.125-ps delay. The signal was fed into the system, sampled in 1-ps chunks over time, and reconstructed successfully to show the original double-peaked waveform. Not only were the results good qualitatively, but the team was able to demonstrate that the sampling response was linear up to 300-nW optical power.

For more information, contact S. Fischer at

Sunny Bains


  1. S. Fischer et al., Optics Letters 26(9), 626 (May 1, 2001).

SUNNY BAINS is a scientist and journalist based in London, England;

More in Home