High-speed Time-interleaved Digital-to-Analog Converter (TI-DAC) for Self-Interference Cancellation Applications

Nowadays, the need for higher data-rate is constantly growing to enhance the quality of the daily communication services. The full-duplex (FD) communication is exemplary method doubling the data-rate compared to half-duplex one. However, part of the strong output signal of the transmitter interferes to the receiver-side because they share the same antenna with limited attenuation and, as a result, the receiver’s performance is corrupted. Hence, it is critical to remove the leakage signal from the receiver’s path by designing another block called self-interference cancellation (SIC). The main goal of this dissertation is to develop the SIC block embedded in the current-mode FD receivers. To this end, the regenerated cancellation current signal is fed to the inputs of the base-band filter and after the mixer of a (direct-conversion) current-mode FD receiver. Since the pattern of the transmitter (the digital signal generated by DSP) is known, a high-speed digital-to-Analog converter (DAC) with medium-resolution can perfectly suppress main part of the leakage on the receiver path. A capacitive DAC (CDAC) is chosen among the available solutions because it is compatible with advanced CMOS technology for high-speed application and the medium-resolution designs. Although the main application of the design is to perform the cancellation, it can also be employed as a stand-alone DAC in the Analog (I/Q) transmitter. The SIC circuitry includes a trans-impedance amplifier (TIA), two DACs, high-speed digital circuits, and built-in-self-test section (BIST). According to the available specification for full-duplex communication system, the resolution and working frequency of the CDAC are calculated (designed) equal to 10-bit (3 binary+ 2 binary + 5 thermometric) and 1GHz, respectively. In order to relax the design of the TIA (settling time of the DAC), the CDAC implements using 2-way time-interleaved (TI) manner (the effective SIC frequency equals 2GHz) without using any calibration technique. The CDAC is also developed with the split-capacitor technique to lower the negative effects of the conventional binary-weighted DAC. By adding one extra capacitor on the left-side of the split-capacitor, LSB-side, the value of the split-capacitor can be chosen as an integer value of the unit capacitor. As a result, it largely enhances the linearity of the CADC and cancellation performance. If the block works as a stand-alone DAC with non-TI mode, the digital input code representing a Sinus waveform with an amplitude 1dB less than full-scale and output frequency around 10.74MHz, chosen by coherent sampling rule, then the ENOB, SINAD, SFDR, and output signal are 9.4-bit, 58.2 dB, 68.4dBc, and -9dBV. The simulated value of the |DNL| (static linearity) is also less than 0.7. The similar simulation was done in the SIC mode while the capacitive-array woks in the TI mode and cancellation current is set to the full-scale. Hence, the amount of cancelling the SI signal at the output of the TIA, SNDR, SFDR, SNDRequ. equals 51.3dB, 15.1 dB, 24dBc, 66.4 dB. The designed SIC cannot work as a closed-loop design. The layout was optimally drawn in order to minimize non-linearity, the power-consumption of the decoders, and reduce the complexity of the DAC. By distributing the thermometric cells across the array and using symmetrical switching scheme, the DAC is less subjected to the linear and gradient effect of the oxide. Based on the post-layout simulation results, the deviation of the design after drawing the layout is studied. To compare the results of the schematic and post-layout designs, the exact conditions of simulation above (schematic simulations) are used. When the block works as a stand-alone CDAC, the ENOB, SINAD, SFDR are 8.5-bit, 52.6 dB, 61.3 dBc. The simulated value of the |DNL| (static linearity) is also limited to 1.3. Likewise, the SI signal at the output of the TIA, SNDR, SFDR, SNDRequ. are equal to 44dB, 11.7 dB, 19 dBc, 55.7 dB.