Method enables cheaper pulse sources for OTDM systems


by Yvonne Carts-Powell

A group at the University of California Santa Barbara (UCSB) demonstrated a system that allows penalty-free optical time-division multiplexing (OTDM) to 40 Gbit/s, and demultiplexing back to 10 Gbit/s.1 The system uses self-phase modulation in a semiconductor optical amplifier (SOA) to improve pulse extinction ratio. The researchers believe this will allow cheaper pulse sources to be used with higher-bit-rate OTDM systems.

Optical time domain multiplexers and channel add/drop units interleave short data pulses in specific time spots. The interleaving channels must contain little energy outside their designated bit slots to avoid interference between channels. Therefore, the pulse extinction ratio (ER) must be large. (For this work, the pulse ER is defined as the ratio between the peak pulse power and the maximum power of the unwanted background signal between the pulses.)

A modelocked fiber-ring laser is the usual pulse source (which is then split into a number of channels and modulated with the data), because it can provide a good ER of about 40 dB. But a cheaper, compact source such as a lithium niobate modulator or electroabsorption modulator (EAM), would be welcome—if the ER were acceptable. Unfortunately, these less expensive options typically have ER values below 25 dB, which limits their use in OTDM systems.

Mads L. Nielsen, Bengt-Erik Olsson, and Daniel J. Blumenthal at UCSB have presented a new way of increasing the pulse ER, and consequently the system performance using self-phase modulation (SPM) in an SOA. Unlike methods that use a nonlinear fiber as the SPM element, an SOA is compact and can be integrated with other optical components.

They managed this improvement in the ER by using the self-phase modulation in the SOA. This causes in chirping of pulses, which produces spectral broadening, mostly to the long-wavelength side (see figure). The power-dependent spectral shift results in a separation of the high-intensity content of the pulses, corresponding to the peaks from the unwanted background power. The ER can be increased by filtering out the red-shifted part of the spectrum with a bandpass filter (BPF). Depending on the filtering, the spectral shift can also be accompanied by pulse compression, since the red-shifted peak is usually broader than the input spectrum.

The pulsewidth after the combination of the SOA and the BPF was approximately 9 ps (compared to 12 ps) at the input. After the filtering, the pulses were modulated with a pseudorandom bit sequence (PRBS) and multiplexed into a four-channel 40-Gbit/s signal. The signal was demultiplexed unto four 10-Gbit/s channels, using an EAM.

Without increasing the ER, the signal could not be demultiplexed. When the pulsewidth was merely narrowed to 8 ps, without the SOA and bandpass filter, the performance was not as good as with the ER-improving devices.

For more information contact, Mads L. Nielsen, now at the Research Center COM, Technical University of Denmark, at

1. M. L. Nielsen, B.-E. Olsson, D. J. Blumenthal, IEEE Photon. Tech. Lett. 14(2), 245 (February 2002).

More in Electronics