In the current field of electronic engineering, the continuous improvement of chip integration has made it normal to integrate one or more radio frequency systems on a tiny chip. This technological advancement has brought about architectural innovations, especially the widespread adoption of zero-IF and low-IF architectures. These architectures are favored for their simplicity and the elimination of the need for external filters of a superheterodyne receiver. However, although the RF part is thus simplified, the calibration of the digital processing part becomes more complex and important. This leads to a core question: What non-ideal characteristics in real devices affect the performance of RF systems?
The first thing we have to focus on is thermal noise and flicker noise. Any real electronic device will generate random noise due to the random movement of electrons, that is, thermal noise. For example, a passive resistor R at temperature T K will generate noise voltage. If the load of this resistor is considered to be equal to itself, the noise power input to the load is usually expressed as KTB. Without considering the system bandwidth, if the temperature T is 290K, then the noise power will be the well-known -174dBm/Hz. At the same time, the flicker noise (1/f noise) in active devices cannot be ignored. Because it is located near direct current (DC), the impact on zero-IF architectures is particularly significant, and the impact on low-IF architectures is slightly less.
The next consideration is the phase noise of the local oscillator (LO). The oscillator output under ideal conditions can be represented by a delta function in the frequency domain, but in actual situations phase noise often causes a skirt in the output signal spectrum. The impact of this phase noise on the transceiver is mainly manifested in two aspects: first, the increase in in-band noise caused by the multiplication of the local oscillator phase noise band and the signal; second, the in-band noise caused by the mixing of the interference signal and the local oscillator phase noise. Noise increases, known as reciprocal mixing.
In addition, sampling jitter is also a critical factor. Analog-to-digital converters (ADCs) and digital-to-analog converters (DACs) form the boundary between analog and digital in transceivers. In the conversion process between these two signal forms, a sampling clock is needed, which is essentially an oscillation signal. Since the actual oscillation signal will produce phase noise, which appears as jitter in the time domain, leading to sampling errors and further generating noise.
The next things to look at are carrier frequency offset (CFO) and sampling frequency offset (SFO). In communication systems, the carrier frequency is usually generated by a phase-locked loop. However, due to the slight difference in the carrier frequency of the transmitter (TX) and the receiver (RX), the frequency after conversion of the receiver will have a residual frequency error, that is, the carrier frequency offset (CFO). At the same time, there may also be a difference in the sampling frequency of the ADC and DAC, called sampling frequency offset (SFO), which will also have an impact on system performance.
When considering the performance of an RF system, one must also be aware of quantization noise and truncation of DACs and ADCs. When performing analog-to-digital conversion, these devices generate quantization noise, which in turn produces a limited signal-to-noise ratio (SNR). Therefore, when designing a receiver, it is usually necessary to provide sufficient gain in the ADC front-end to ensure that the noise level of the ADC itself is small enough to be ignored compared with its input thermal noise level (generated by the front-end circuit). The truncation effect of the ADC will limit the peak-to-average power ratio (PAPR) of the signal, thereby deteriorating the SNR of the signal.
Finally, there are quadrature imbalances and device nonlinearities to consider. During the upconversion or downconversion process, the quadrature mixer used may have gain mismatch and phase mismatch on the I and Q paths, which will affect the SNR of the signal or generate out-of-band noise. The nonlinearity of the device, especially the nonlinearity of the receiver, is mainly responsible for handling large signal interference, which is what we usually call intermodulation immunity. These factors jointly determine the performance of RF systems, and it is crucial for electronic engineers to understand and master these factors in order to make appropriate decisions when designing and optimizing RF systems.