comparison between direct conversion and heterodyne receiver
It is more difficult to say which architecture, superheterodyne or direct conversion, leads to lower power consumption. In a direct-conversion receiver, there is no need to drive the signal off the chip at RF and IF. The interfaces to the off-chip filters are typically matched to quite low impedances, which are power-hungry. The quadrature downconversion at RF in a directconversion receiver consumes more power than the quadrature downconversion at IF in a superheterodyne solution. Since there is no passive channel-select filter before the baseband circuitry, the dynamic range requirement of the baseband part is considerably higher in a directconversion receiver. A sufficient dynamic range at the baseband may require a considerable amount of power. In the superheterodyne architecture, a passive IF filter attenuates the out-ofband signals, decreasing significantly the dynamic range requirement of the baseband circuit. In a direct-conversion receiver, two high-performance active filters are required. On the other hand, in a direct-conversion receiver, there is no IF circuitry at all. Signal processing at an IF consumes more power than at the baseband because of the higher operation frequency. Direct-conversion radio receivers have been used in commercial digital phones
The direct-conversion architecture suffers from several drawbacks, which make the design of a high-performance receiver a challenging task. These include DC offsets due to device mismatches and self-mixing of the LO, RF signal self-mixing as a result of leakage to the LO port, distortion due to even-order nonlinearities, flicker noise, and leakage of the LO signal out from the antenna .
The LO signal is in the passband of the LNA, mixers, and off-chip RF filters. The LO signal leaks to the LNA input and to the input of the down-conversion mixers. If the LO leaks to the LNA input, it is amplified with the LNA gain. The leaked LO signal is downconverted with itself, resulting in a constant DC offset. The level of the offset depends on the amount of leakage and the phase shift between the LO signal and leakage. The resulting DC offset at the mixer output can be orders of magnitude larger than the desired signal. If the RF gain is changed, the level of the LO leakage at the mixer input is altered, resulting in a change in the DC offset at the mixer output. A change in the phase is also possible. The DC offset change can be much higher than the wanted signal For example the LO-to-RF isolation is 65dB, while the LO power can be as high as 0dBm. This results in a leaked LO signal of –65dBm at the LNA input, which is 52dB higher than the wanted channel in the UTRA/FDD reference sensitivity test case . An effective method to mitigate the amount of LO signal at the LNA input is to use a double-frequency LO, from which the LO is generated on-chip using a divide-by-two circuit.
Since the LO signal is at the passband of the pre-selection filter, it can leak out from the antenna and reflect back. The LO signal leaking out from the antenna interferes with other receivers in the system. Wireless standards specify a maximum amount of spurious LO emission, which can range from –50dBm to –80dBm . In UTRA/FDD, the spurious LO emission of a cellular phone must not exceed –60dBm/3.84MHz at the antenna connector. The leaked LO signal may reflect back from external objects. Since the environment may change and may contain moving objects, the level and phase of the reflected LO signal can change. The result is a time-varying DC offset. The efficiency of the DC offset removal scheme in this case depends on the frequency content of the reflected LO signal. If the LO signal is reflected back from moving objects, the result is a Doppler shift in the frequency of the reflected signal. Therefore, the reflected LO signal is downconverted to a non-zero baseband frequency, which depends on the speed of the external moving object . The significance of the low-frequency spur generated due to the reflected LO signal depends on the amount of LO signal at the RF input. The use of a double-frequency LO and a divide-by-two circuit is an effective way to mitigate the LO signal at the RF input.