## Abstract

ABC (Auto Bias Control) technique for QAM (Quadrature Amplitude Modulation) transmitter is demonstrated. 16-QAM (10G baud) is generated and controlled using a single IQ modulator and asymmetric bias dithering technique. Measured penalty is 0.3dB.

© 2011 OSA

## 1. Introduction

Optical QAM and its variations are very attractive in improving the spectral efficiency of high speed transmission systems [1]. Several approaches can generate high order QAM signals [2]. The electrical approach using two electrical multi-level signals is quite simple and attractive, because it does not require multiple IQ modulators [1]. Most IQ modulators use a nested Mach-Zehnder (MZ) modulator consisting of two sub LiNbO_{3} phase modulators and an optical phase shifter [3]. The bias points of the phase modulators and the delay of the optical phase shifter are controlled by three DC bias voltages. Because the optimal bias voltages shift due to DC-drift, an ABC circuit with high sensitivity is required for stable long-term operation, especially for the phase shifter. We adopted asymmetric bias dithering to create a highly sensitive ABC technique for optical 4-QAM, i.e., QPSK systems [4, 5]. Several other approaches have since been proposed, monitoring the variance of signal power [6] or utilizing information of peak-to-average power ratio [7]. More recently, Choi et al. demonstrated an ABC for a 16-QAM signal that uses differential phasor monitoring [8]. However, to the best of our knowledge, no bit rate free ABC technique for optical QAM transmitters has been reported.

In this paper, we shows that our bit rate free ABC technique, based on asymmetric bias dithering, well supports 16-QAM transmitters that use a single IQ modulator. We also show the measured constellation of 16-QAM (10G baud) signals. Measured penalty induced by ABC was 0.3dB.

## 2. Construction of 16-QAM transmitter

Figure 1
shows the 16-QAM transmitter. The blue area in Fig. 1 represents the typical IQ modulator. A CW light is split and launched into two MZ phase modulators: PM _{I} and PM_{Q}. Two modulated lights (optical electric fields E_{I} and E_{Q}) are generated using two 4 level signals. The optical phase shifter sets the phase of optical carrier between E_{I} and E_{Q}, to ϕ. Coupling E_{I} and E_{Q} yields the 16-QAM signal. The best signal quality is achieved when ϕ = π/2. The DC bias voltage, BIAS_{k}, (k = I, Q) is set to the null transmission point. BIAS_{phase} is added to the optical phase shifter to control ϕ. The ideal constellation, schematically shown in Fig. 1, is symmetrical respect to the origin, then E_{k(L-1)} - E_{kL} = E_{kL} - E_{k(L + 1)} and E_{k1} = -E_{k4}, where L = 2 ~3.

Figure 2
schematically shows the transfer characteristic of PM_{k} (k = I, Q). V_{k1} ~V_{k4} correspond to the four levels of Signal_{k} in Fig. 1. When there is no bias drift (black symbols), E_{k} = C_{1}sin(C_{2}V_{k} + C_{3}BIAS_{k}), where C_{1} ~C_{3} are constants set from the optical power and half-wave voltage. Note that pre-emphasis based on an inverse sine function is required for V_{k1} ~V_{k4}, to achieve E_{k(L-1)} - E_{kL} = E_{kL} - E_{k(L + 1)} (For simplicity, we ignore the nonlinearity of the driver amplifier).

## 3. Effect of bias drifts in 16-QAM system

In the ideal case, total optical power of 16-QAM signal, P_{total}, is described as

_{IL}and E

_{QM}, P

_{total}can be affected by bias drift.

At first, let’s consider the drift of BIAS_{k} (k = I, Q). A 16-QAM signal exhibits a more complicated power change profile than a QPSK signal [4,5], because when |E_{k3}| is decreased, |E_{k2}| is increased. Now, we introduce parameter D, described as

_{π}is the half wave voltage. P

_{total}and V

_{drift}can be described as

_{drift}is the amount of bias drift. Figure 3 shows the calculated P

_{total}as a function of V

_{drift}/(2V

_{π}). Red line shows V

_{k1}-V

_{k4}= 0.80 x 2V

_{π}and D = −0.006. Blue line shows V

_{k1}-V

_{k4}= 0.78 x 2V

_{π}and D = + 0.036. When D is negative (positive), maximum (minimum) P

_{total}is achieved at V

_{drift}= 0. This means that the conventional bias dithering technique can be used to detect the drift of BIAS

_{k}[4, 5].

Next, let’s consider the drift of BIAS_{phase}. Parameter ϕ in Fig. 1 cannot remain at the optimum value, π/2. Total optical power, P_{total} is described as

_{k1}| = |E

_{k4}| and |E

_{k2}| = |E

_{k3}|, P

_{total}is independent of ϕ and BIAS

_{phase}[4, 5].

## 4. Asymmetric bias dithering

If the symmetry of the constellation is slightly broken, P_{total} depends on cos(ϕ), and the condition of BIAS_{phase} can be monitored by using a low speed optical power monitor. In Fig. 2, red and blue symbols schematically show the dithered E_{kL} (L = 1~4), with our asymmetric bias dithering [4, 5]. In Fig. 2, |V_{k1}-V_{k4}| is set slightly smaller than 2V_{π}. The dither signals for BIAS_{I} and BIAS_{Q} are orthogonal sine waves. Dithered V_{IL} and V_{QL} are given as V_{IL} + V_{dither} cos(ω_{d} t) and V_{QL} + V_{dither} sin(ω_{d} t), where ω_{d} is the angular frequency of the dithering signal and t is time. When ω_{d}t = π/4 (red symbol), |E_{k2}| > |E_{k3}| and |E_{k1}| > |E_{k4}|; when ω_{d}t = 5π/4 (blue symbol), the converse is true. The generated asymmetric constellation maps with ω_{d} t = π/4 and 5π/4 are schematically shown in Fig. 4
(red and blue stars). When cos(ϕ) > 0, maximum star shift appears at A or B, and maximum P_{total} is achieved. When cos(ϕ) < 0, maximum P_{total} is achieved with ω_{d} t = 3π/4 and 7π/4 (not shown). It means the sign of cos(ϕ) can be detected by the 2nd order lock-in amplifier and an optical power monitor [4, 5]. When detected cos(ϕ) is positive (negative), ϕ is smaller (larger) than the optimum value, π/2. As mentioned before, BIASI and BIASQ can be monitored by a conventional dithering technique using a 1st order lock-in amplifier with reference clocks cos(ω_{d} t) and sin (ω_{d} t) [4]. The detected bias drifts are fed back to the bias voltages which yields ABC. Here, the voltage of the feedback signal is decided by the cos(ϕ), V_{drift} and D.

The above discussion does not depend on the bit rate of the QAM signal. It means that this ABC technique is bit rate free.

## 5. Experimental setup and results

Figure 1 shows the experimental setup. AWG generated the 4 level Signal_{I} and Signal_{Q}, using PRBS patterns (PN15). Baud rate was 10G baud, and (V_{k1} - V_{k4}) was set to 80% of 2V_{π}. Parameter D is negative. Waveform of driver amplifier with pre-emphasis is also shown in Fig. 1. Inside the ABC circuit, the oscillator generated three sine waves (ωd). Two of them were used to dither BIASI and BIASQ. The phases of these dither signals were orthogonal. The third sine wave was used as the reference clock of the lock-in amplifier. In the modulator, the generated 16-QAM signal was tapped off and fed to the photo detector (PD) located in the modulator. Detected optical power was input to the lock-in amplifier. The lock-in amplifier detected the 1st order (ωd) error signal and 2nd order (2ωd) error signal. The bias controller in ABC circuit fed back the error signal for each bias voltage. An offset signal can be add to the 2nd order error signal, for fine tuning.

We measured the Q factor and constellation using a digital coherent receiver. Result is shown in Fig. 5 and 6 . In normal operation without offset signal, best constellation was achieved, and Q factor was 9.1 dB. When negative (positive) offset signal was added for comparison, ϕ was smaller (larger) than π/2 showing that the ABC circuit worked correctly. We also measured Q factor without using ABC circuit. When all biases were manually adjusted, maximum Q factor increased 0.3 dB. This means that the penalty induced by this ABC technique was 0.3 dB.

## 6. Conclusion

We show that our ABC technique based on asymmetric bias dithering well suits the 16-QAM (10G baud) transmitter constructed with a single IQ modulator. Measured penalty induced by ABC was 0.3dB.

## References and links

**1. **S. Yamanaka, T. Kobayashi, A. Sano, H. Masuda, E. Yoshida, Y. Miyamoto, T. Nakagawa, M. Nagatani, and H. Nosaka, “11x171Gb/s PDM 16-QAM transmission over 1440 km with a spectral efficiency of 6.4 b/s/Hz using high-speed DAC,” in Proceedings of ECOC 2010, We.8.C.

**2. **I. Morohashi, M. Sudo, T. Sakamoto, A. Kanno, A. Chiba, J. Ichikawa, and T. Kawanishi, “16 QAM synthesis by angular superposition of polarization using dual-polarization QPSK modulator,” in Proceedings of ECOC 2010, We.3.14.

**3. **T. Yamada, Y. Sakamaki, T. Saida, A. Kaneko, A. Sano, and Y. Miyamoto, “86-Gbit/s differential quadrature phase-shift-keying modulator using hybrid assembly technique with planar lightwave circuit and LiNbO 3 devices,” in Proceedings of LEOS 2006, ThDD4.

**4. **H. Kawakami, E. Yoshida, and Y. Miyamoto, “Asymmetric dithering technique for bias condition monitoring in optical QPSK modulator,” Electron. Lett. **46**(6), 430–431 (2010). [CrossRef]

**5. **H. Kawakami, E. Yoshida, and Y. Miyamoto, “Auto bias control technique for QPSK modulator with asymmetric bias dithering,” in Proceedings of OECC 2010, 8B2–4.

**6. **P. S. Cho and M. Nazarathy, “Bias control for optical OFDM transmitters,” IEEE Photon. Technol. Lett. **22**(14), 1030–1032 (2010). [CrossRef]

**7. **T. Yoshida, T. Sugihara, K. Uto, H. Bessho, K. Sawada, K. Ishida, K. Shimizu, and T. Mizuochi, “A study on automatic bias control for arbitrary optical signal generation by dual-parallel Mach-Zehnder modulator,” in Proceedings of ECOC 2010, Tu.3. A. 6.

**8. **H. G. Choi, Y. Takushima, H. Y. Choi, J. H. Chang, and Y. C. Chung, “Modulation-format-free bias control technique for MZ modulator based on differential phasor monitor,” in Proceedings of OFC/NFOEC 2011, JWA33.