IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 11, NO. 6, JUNE 1999 659 Harmonic Signals from Electroabsorption Modulators for Bias Control G. L. Li, Y. Z. Liu, R. B. Welstand, C. K. Sun, W. X. Chen, J. T. Zhu, S. A. Pappert, and P. K. L. Yu Abstract— Harmonic signals reflected from electroabsorption modulators are measured and analyzed as a function of modula- tor bias. The second-harmonic signal exhibits a dip close to the bias where the maximum of RF link gain occurs, over a significant optical power range. Using an equivalent circuit analysis we show this is caused by the inherent electroabsorption effect. The second-harmonic signal can be exploited for dynamic self- bias control of electroabsorption modulators in analog fiber-optic links. Index Terms—Analog fiber link, bias control, electroabsorption modulator. S EMICONDUCTOR ELECTROABSORPTION modula- tors (EAM’s) are widely considered for use in analog fiber optic links due to their small size, large bandwidth [1], high efficiency and potential for monolithic integration with other semiconductor components [2]. For links employing an EAM, the RF efficiency and the multioctave spurious-free dynamic range (SFDR) can be optimized at the same modulator bias [3]. However, the optimum bias needs to be adjusted during operation, as the modulator transfer characteristics can change in response to changes in ambient temperature, polarization and optical power levels. A common approach for maintaining optimal bias is to insert a Y-branch coupler after the modulator and monitor a small portion of the modulated light [4]. This approach induces extra optical loss and requires a dedicated photodetector thereby increasing hardware complexity and component count. To simplify the optical configuration for EAM bias con- trol, we previously demonstrated a self-bias control approach based upon the correlation between the RF link gain and the modulator dc photocurrent [5]. In this work, we propose and demonstrate an alternative self-bias approach based upon the correlation between the RF link gain and the second-harmonic signal reflected from the EAM. From the measured RF link gain and the EAM harmonic signal as a function of modulator bias, we found that the second harmonic exhibited a dip at a bias close to that for maximum RF link gain, as the optical Manuscript received November 6, 1998; revised January 18, 1999. This work was supported in part by the Defense Advanced Research Projects Agency/Rome Laboratories, in part by the Office of Naval Research, in part by MICRO/Raytheon programs. G. L. Li, R. B. Welstand, W. X. Chen, J. T. Zhu, and P. K. L. Yu are with the Department of Electrical and Computer Engineering, University of California at San Diego, La Jolla, CA 92093-0407 USA. Y. Z. Liu is with Fermionics Laser Technologies; Simi Valley, CA 93063 USA. C. K. Sun and S. A. Pappert are with SPAWAR Systems Center, San Diego, CA 92152-5001 USA. Publisher Item Identifier S 1041-1135(99)04222-6. Fig. 1. Schematic diagram of measurement setup. Two spectrum analyzers are used for measuring the EAM second-order harmonic signals and the detector fundamental signals, respectively. power was varied from 8 to 14 dBm. These measurement results can be explained using an equivalent circuit model for the EAM. The observed correlation between the RF link gain and the second-harmonic signal is mainly caused by the inherent electroabsorption effect in the modulator. A fiber-packaged InGaAsP–InP Franz–Keldysh effect wave- guide modulator is used in this work. The waveguide is 2.5 m wide at the top and is 300 m long. It has a 0.3- m-thick intrinsic InGaAsP m) electroabsorption layer. To facilitate coupling to lensed fibers, a large optical cavity is incorporated in the waveguide. At 1.32- m wavelength, the packaged device has a fiber-to-fiber optical insertion loss of 10.7 dB at zero bias. Fig. 1 depicts the measurement setup, with arrows indicat- ing the direction of signal flow. Two RF spectrum analyzers are used to simultaneously measure and display the EAM second- harmonic signal, , and the photodetector fundamental signal, . The RF source is set at 1 GHz and 20 dBm. For simplicity, the biasing networks and the dc power supplies connected to the modulator and the photodetector are not shown in Fig. 1. With the EAM biased at 3.2 V, is measured as a function of optical input power . For small versus curve is a straight line on a logarithmic scale, with a slope of 2 dB/dB. When is increased to 14 dBm, is compressed by 1 dB from the line. Next, and are measured as a function of modulator bias with ranged from 0 to 14 dBm. Fig. 2 shows the measured and versus curves at different ’s. Within the measured range, always peaks at a certain voltage, denoted as - When is large enough, exhibits a dip at a voltage denoted as - Table I lists the and correspondingly measured - and - Within the range of 8–14 dBm, - and - are very close to each other, with - always slightly larger than - . This indicates that - can be tracked by checking - in this optical power range. 1041–1135/99$10.00  1999 IEEE