Robustness investigations of proposed transmitter IQ skew measurement method . 37

37

modulator and the connection cables. Light source was a 1546.5nm continuous wave laser with 13dBm output power. An optical spectrum analyzer (OSA) with resolution bandwidth of 150MHz collected the main and image signal spectrum.

Fig 3. 8 Measured main signal and image signal spectrum with transmitter IQ skew 5.65ps One measured main signal and image signal spectra with the transmitter IQ skew is shown in Fig 3. 8 as an example. The dash line marks the wavelength of optical carrier. Fig 3. 8 a) shows the image spectrum with negative frequency components and Fig 3. 8 b) shows the image spectrum with positive frequency components, after these optical spectra move to the base frequency by subtracting the carrier frequency. The image spectrum shape is determined by the transmitter IQ skew as expressed in Eq 3.4. According to the equations in the previous sections, the transmitter IQ skew in this case is measured to be +5.65ps.

3.4. Robustness investigations of proposed transmitter IQ skew

set in the experiment by a special setting method. Firstly the bias phase of parent Mach-Zehnder structure was set 90 degree using the criterion that peak powers of positive image signal and negative image signal are same when sending the single sideband 12GHz single frequency signal.

Secondly bias I and Q of the child Mach-Zehnder structure were adjusted until output optical power at the carrier wavelength is minimum.

Deliberate bias deviations of parent Mach-Zehnder structure were added on the basis of the optimal bias point. The bias phase control voltage on single arm of parent Mach-Zehnder structure was adjusted in range 0V to 1.2V for the evaluation. According to the differential DC Vpi of 2.5V in the test report of the Indium Phosphide modulator, the corresponding bias deviations of parent Mach-Zehnder structure is of range 0 degree to 43.2 degree.

Fig 3. 9 Measured image signal spectrum denoted by IMSR under different bias phase control voltage of parent Mach-Zehnder structure

Fig 3. 10 Measured IQ phase at different frequencies under different bias phase control voltage of parent Mach-Zehnder structure

39

Fig 3. 11 Measured transmitter IQ skew error versus bias deviations of parent Mach-Zehnder structure with digital delay 0ps and 5ps

The measured results of image spectrum and IQ phase under different bias deviations of parent Mach-Zehnder structure are shown in Fig 3. 9 and Fig 3. 10 separately. As expected, the measured image spectrum and IQ phase curves shift along the frequency axis when increasing the control voltage of bias phase of parent Mach-Zehnder structure. Fig 3. 11 shows the measured transmitter IQ skew error as a function of bias deviations of parent Mach-Zehnder structure. Two transmitter IQ skew cases with additional digital delay 0ps and 5ps were considered. As shown in the figure, the proposed method is quite stable even when the bias significantly deviates from the optimum bias point. The standard deviation of the skew error is as small as 0.22ps. According to Eq 3.4, bias deviation of parent Mach-Zehnder structure shifts the image spectrum along the frequency axis and does not affect its shape. Thus, the measured skew is not affected.

Fig 3. 12 Measured image signal spectrum denoted by IMSR under different bias phase control voltage of child Mach-Zehnder structure

Fig 3. 13 Measured IQ phase at different frequencies under different bias phase control voltage of child Mach-Zehnder structure

Fig 3. 14 Measured transmitter IQ skew error versus bias deviations of child Mach-Zehnder structure with digital delay 0ps and 5ps

When changing the bias I or Q deviations, modulator output DC power changed as well. The measured results of image spectrum and IQ phase under different bias I deviations are shown in Fig and Fig 3. 13 separately. The measured image spectrum and IQ phase with different bias I deviations agree well with each other because bias I deviation did not affect the radio frequency signal with the small deviation assumption. Results of the skew error versus bias deviation of child Mach-Zehnder structure are shown in Fig 3. 14. The transmitter IQ skew error is rather stable considering the significant child modulator bias deviations and the identified skew error standard deviation is as small as 0.13ps.

41

Fig 3. 15 Measured image signal spectrum denoted by IMSR under different driving peak-to-peak voltage

Fig 3. 16 Measured IQ phase at different frequencies under different driving peak-to-peak voltage

Fig 3. 17 Measured relative skew error vs. driving peak-to-peak voltage with additional delay 0ps and 5ps.

Different driving voltages of the test signal were adjusted in range 0.9V to 2V in the

experiment to evaluate the impact to the proposed transmitter IQ skew measurement method.

The measured results of image spectrum and IQ phase under different driving voltages are shown in Fig 3. 15 and Fig 3. 16. The measured image spectrum and IQ phase with different driving voltages agree with each other in the trend. The measured skew errors under different driving voltages are shown in Fig 3. 17. The maximum skew error is 0.44ps and the skew error standard deviation for all driving voltages is identified to be 0.16ps. It tells that the proposed method is stable to the common-used driving voltages.

Fig 3. 18 Measured image signal spectrum denoted by IMSR under different phases for each frequencies of the SSB signal

Fig 3. 19 Measured IQ phase at different frequencies under different phases for each frequencies of the SSB signal

43

Fig 3. 20 Measured relative skew error vs. phases for each frequencies of the SSB signal with additional delay 0ps and 5ps.

The measured results of image spectrum and IQ phase under different SSB signal phases are shown in Fig 3. 18 and Fig 3. 19 separately. The waveform PAPRs corresponding to these SSB comb phases are designed in range of 4~6.25. For both the measured image spectrum and IQ phase, different SSB signal phases cause significant fluctuations in the frequency domain. It has been confirmed that the fluctuations are not from additive noise by checking the results of 10 repeat measurements by one same SSB signal phase. So it is an unexpected phenomenon that the image spectrum not only depends on the amplitude of test signal but also depends on the phase of test signal. Such phenomenon cannot be explained by the theoretical derivation and the transmitter impairment model in previous section. The possible mechanism for it is from the DAC imperfections that is from the amplitude and phase mismatch among the different sub-DACs in the current 64Gsps sampling rate DAC.

The measured skew error under different SSB signal phases are shown in Fig 3. 20. The maximum skew error is as large as 1.86ps and the skew error standard deviation for 10 SSB signal phases is identified to be 0.56ps. To further improve the accuracy, multi-phase average is used to reduce such measurement error. In order to verify this, more transmitter skew cases were considered in the experiment. By adding different digital delays in Q tributary in the experiment, the measured skew results using single SSB signal phase and 10 phase average are shown in Fig 3.

21. As can be seen, with the multiple phase average operation, the skew error standard deviation for all the additional delay cases is reduced from 0.60ps to 0.16ps. And for all the measurement cases, the measurement accuracy is identified as small as 0.33ps.

Fig 3. 21 Measured transmitter IQ skew versus preset additional digital delay

3.5. Short summary of this chapter

Measurement method of the in-phase/quadrature (IQ) skew in coherent optical transmitter based on analysing the shape of image spectrum when sending single sideband (SSB) comb signal is proposed and investigated in this chapter. The impairments model of coherent optical transmitter is also proposed to analyse the IQ skew character and accuracy of the proposed method. Experiments verified that the method has sub-picosecond accuracy and is robust under various child and parent modulator bias deviations and test signal magnitudes. It is unexpected to find that the measured IQ skew has low accuracy for different phases of SSB signal for testing. The occurrence of this phenomenon is suspected to be from the unknown DAC imperfections. Multiple phase average method is proposed and experimentally validated to further reduce the measurement error fluctuations possibly interfered from the DAC imperfections. As a result, the proposed method for transmitter IQ skew measurement in coherent optical system is validated feasible, accurate and robust by the experiment results.

45

4. Hardware efficient channel probing of