Basic Knowledge of Transceivers

Basic Knowledge of Transceivers

I. What is a Transmitter

A Transmitter is a device that converts electrical signals into RF/microwave signals and radiates them into space through an antenna.

The baseband is the signal source generation part. The frequency of baseband data is relatively low, making it unsuitable for transmission, so modulation is required.

Common modulation methods include QPSK (Quadrature Phase Shift Keying), FSK (Frequency Shift Keying), QAM (Quadrature Amplitude Modulation), OFDM (Orthogonal Frequency Division Multiplexing), etc. During such modulation, a Local Oscillator (LO) is needed to provide the carrier frequency. Modulation also achieves frequency migration, which moves the baseband data from the low-frequency band to the high-frequency band.

Both the Driver Amplifier and Power Amplifier (PA) belong to the amplification module. The Driver Amplifier is part of the Power Amplifier and is used to drive it—sometimes a single-stage power amplification is insufficient, so an additional stage is needed to increase the gain and drive this part of the circuit.

Matching Circuit:

1. In RF communication systems, the impedance of each system module is designed to be 50 ohms, but the output impedance of the Power Amplifier transistor is generally not 50 ohms. To ensure the normal operation of the Power Amplifier, the output terminal must achieve a specific impedance. Therefore, a matching circuit needs to be designed to match it with 50 ohms without power loss, maximizing the transfer of the output power of the Power Amplifier to the antenna for radiation.
2. The matching circuit can also act as a filter to eliminate unwanted signals. Due to the nonlinear characteristics of system modules, the output terminal may generate second and third harmonics. As a filter, the matching circuit will attenuate these harmonics, leaving only the fundamental frequency at the output terminal of the transmitter.

II. Direct Conversion Transmitter

1. Overview of Direct Conversion Transmitters

Direct conversion refers to directly upconverting data from the baseband to the RF frequency, and such a transmitter is called a Direct Conversion Transmitter.

The term "direct conversion" does not mean that only single-carrier or quadrature upconversion can be used; the specific method depends on the modulation scheme. Here, we mainly focus on quadrature upconversion because it is commonly used in modern modulations such as QPSK and GMSK—most of the time, it is not single-carrier modulation, but it should be noted that quadrature upconversion is not used in all cases.

Refer to the figure below, which shows the binary sequence and baseband data (i.e., baseband or zero-IF low-frequency signals). The switch here is used to switch the sinusoidal carrier, so the output terminal is either the carrier (representing 1) or zero (representing 0), which is ASK (Amplitude Shift Keying) modulation. Essentially, this is direct conversion: converting the zero-IF baseband signal into an RF frequency, as the output data already has the characteristics of the RF frequency. The core idea of quadrature conversion is the same, which is detailed below.

Single-channel or quadrature upconverters are responsible for modulating baseband signals. For example, QPSK direct conversion transmitters adopt this structure. The specific configuration of the transmitter is determined by the modulation method—if QPSK or GMSK modulation is used, a quadrature structure may be required; if ASK/FSK modulation is used, a single carrier is sufficient.

Can a transmitter be built with only one mixer? Yes, depending on the modulation method. For example, ASK modulation can directly upconvert from zero-IF to RF. Taking a GMSK-modulated direct conversion transmitter as an example, GMSK is a digital modulation, and the modulation output in this case is as follows, with the signal generated by a signal processor. The output form varies with different modulations, but the core idea is the same. The spectrum of GMSK is also provided here for reference, and the structure described below can be used to generate such signals.

2. QPSK Transmitter Structure Example (Direct Conversion)

Let's look at another example of a QPSK transmitter structure. A quadrature structure is used here, but it is not mandatory. In this structure, quadrature upconversion is performed through two mixers, which is the direct conversion architecture of a QPSK transmitter. We upconvert the baseband data, which is provided by a serial-to-parallel converter operating at low frequency. Here, we mainly focus on the transmitter architecture. Therefore, during direct conversion, data is converted from the baseband to RF, and the output is as shown in the figure above. For reference, the spectrum of the QPSK transmitter is also provided.

III. Challenges in Designing Direct Conversion Transmitters

1. I/Q Mismatch

Let's take a QPSK system as an example. In the architecture shown in the figure below, the binary baseband data generates two channels of waveforms (A and B) through a Serial-to-Parallel (S/P) Converter, which are low-frequency pulse signals (1 or -1). Subsequently, quadrature upconversion is performed using a quadrature upconverter. This is the process of directly upconverting from the baseband to RF. Assuming the modulated signal is a QPSK signal, its expression is QPSK = A·Ac·cosωct + B·Ac·sinωct. What would the output signal look like in an ideal system without mismatch?

The above describes the ideal system, where data transmission is smooth. However, practical systems are not perfect, and mismatches are inevitable. There are two main types of mismatches: phase mismatch and amplitude mismatch.

In a quadrature upconverter, we always expect a 90° phase difference and equal amplitude between the two carriers, but there will always be errors in practice. We try to attribute the mismatch to a certain part of the system, which can be expressed as follows:

Therefore, after re-derivation, we obtain the normalized amplitude: if Δø (phase difference) and ΔAc (amplitude difference) are both 0, the result will be as shown in Table 1.

Sending data with errors to the receiver will cause problems—when the receiver receives and demodulates the data, bit errors will occur, which is one of the main challenges in design. Therefore, careful handling is necessary to ensure that the two oscillators in the in-phase (I) and quadrature (Q) parts are well-matched.

2. How to Eliminate I/Q Mismatch Through Calibration?

There is a method to detect and eliminate mismatches using tuning circuits, known as the calibration method. The first step of the calibration method is to attempt to identify phase mismatch, and the following structure can be used as an example:

(1) Eliminating Phase Mismatch

Apply a single sinusoidal wave to the two input terminals of the transmitter to expose the phase mismatch problem.

Next, calculate the average power of the waveform. Assume the amplitude error is ε and the phase error is Δø. Since ε is generally less than 1, the term 1+ε is ignored. During calibration, we assume sinΔø = 0, i.e., the phase error is 0. At this time, measure the output power Vo2, which is the power of the sinusoidal wave Vo·coswint. The principle here is: if a sinusoidal wave is applied to the input terminal of the transmitter and the phase error is 0, the output power should be the same as the input power. We can apply this power to the circuit and adjust the circuit parameters until the output power matches, which is an effective method to eliminate phase mismatch.

(2) Eliminating Gain Mismatch

Next is the method to eliminate gain mismatch. As shown in the figure, ground one side, input a cosine signal to the in-phase (I channel), and then calculate the output voltage Vout. According to the formula for Vout:

Then we calculate the average power of Vout1. Similarly, connect the Vo·coswint signal to the in-phase part and calculate the output power at this time. Essentially, when we obtain this power and calculate the difference between Vout1 and Vout2, the error should be zero in the ideal case, meaning the powers of the two channels should be completely consistent.

3. Carrier Leakage

(1) What is Carrier Leakage?

When the analog baseband circuit in the transmitter generates quadrature signals, DC offset may occur. Simply put, this is an unwanted DC component in the circuit that can interfere with normal signal transmission.

It mainly comes from Local Oscillator (LO) leakage. When the LO signal leaks into the baseband circuit, it will generate a large DC offset in the baseband, which may saturate the baseband circuit and affect the operation of the entire system.

There are two types, namely Vos1 and Vos2, which belong to the two types of DC offsets at the input port of the mixer, as shown in the figure below.

Since the baseband part includes analog circuits and mixers, DC offset will occur. This can be regarded as the input offset of the mixer, i.e., there is a DC voltage here. When this DC voltage is added, its impact will be manifested.

At the output terminal, in addition to the desired GMSK signal, two unwanted signals are generated, which is the problem. We can use "carrier leakage" to measure this phenomenon.

(2) What Problems Can Carrier Leakage Cause?

(1) Carrier leakage can distort the signal constellation diagram.

Taking QPSK modulation as an example:

The values of a1 and a2 are (11, -11, 1-1, -1-1). At this time, the signal needs to be multiplied by the amplitudes of cosine and sine, but additional components will be superimposed. These offsets introduce extra terms that change/distort the signal, leading to the deformation of the constellation diagram.

Ideally, the values of a1Vo and a2Vo should be obtained, but carrier leakage increases the amplitude, equivalent to introducing additional power, ultimately causing the constellation diagram to deviate from the ideal position.

(2) The base station measures the power level received from the device to adjust its transmission power.

Assume the schematic shows a mobile system, such as a transmitter like a GMSK transceiver. Each time a signal is sent to the base station, the base station adjusts its own transmission power according to the received power level to communicate with the mobile device. For example, when the mobile device is close to or far from the base station:

• If the base station receives a signal from a nearby mobile device, it can determine that the device is close. When sending signals to the device later, it will adjust the power level according to the distance;
• If the mobile device is far from the base station, the signal it sends will attenuate. The base station determines that it is far away through the received signal strength and sends a stronger signal than when the device is close. This is because if a low-amplitude signal is not sufficiently amplified, it may be excessively attenuated during transmission, making it impossible for the mobile device to detect the signal with an acceptable Signal-to-Noise Ratio (SNR).

However, if there is carrier leakage, the carrier power will dominate, making it more difficult to measure the actual signal power. As mentioned earlier, the amplitudes of the cosine and sine signals are ideally aVo, but carrier leakage introduces additional power. If the offset value is high, the carrier power accounts for a too large proportion, and the base station cannot accurately measure the true power due to the leakage problem.

(3) How to Reduce Carrier Leakage?

• Precision circuit design: Reduce leakage by optimizing the circuit structure;
• Increase the amplitude of the GMSK signal: Increasing the amplitude of the effective signal can relatively reduce the impact of leakage, but it should be noted that if the mixer enters the nonlinear operating region, the amplitude increase will be limited.

The core goal is always to reduce the relative carrier leakage index, but there is an upper limit to amplitude adjustment, which requires a balanced design within the linear operating range of the circuit.

Refer to the system block diagram above. When the system is operating but not transmitting any signals, the offsets Vos1 and Vos2 will be transmitted to the output terminal. At this time, the power detector will measure the power of these offsets. Since no new signals are being transmitted, the power detected by the power detector only comes from the offsets, which will then be digitized (sometimes stored in a memory because the offsets may change with the operating state of the transmitter, and the power values corresponding to different offsets are different, so dynamic adjustment is required instead of fixed processing).

After digitization, a negative offset is generated through the DAC. As shown in the figure, this negative offset can cancel out the original offset, ultimately zeroing the offset. This is an efficient and commonly used method.

IV. Linearity of Transmitters

1. What is Linearity?

Linearity is a characteristic of mathematical relationships or functions whose graphs can be represented as a straight line. A linear relationship is a straight line on a graph. We always hope that the system can output such results, but in practice, it is particularly difficult to construct a completely linear system, and nonlinear characteristics will inevitably occur.

In addition to the first-order term of a linear system, there are two other types of terms: second-order nonlinearity and third-order nonlinearity. However, nonlinear terms are not limited to these two types; there may be higher-order terms. Since terms above the third order are relatively small, they can generally be ignored, but they still need to be considered in high-sensitivity systems.

2. Linearity in Transmitters

When discussing linearity, special attention should be paid to two modules of the transmitter: the mixer and the Power Amplifier (PA). Excessive nonlinearity at the baseband port of the mixer can cause signal distortion or increase adjacent channel power.

3. What is an Adjacent Channel?

An adjacent channel refers to the channel or frequency immediately adjacent to the upper and lower sides of a specific channel or frequency. Taking the transmitter output as an example, assuming the center frequency is fc (such as in the case of GMSK modulation), the signal power should ideally be concentrated within the channel bandwidth. However, due to various circuit defects, power leakage outside the bandwidth may occur, as shown in the part marked "adj" in the figure below.

These adjacent channel interferences are undesirable because they will interfere with other channels and ultimately generate noise, which we need to avoid. Due to some errors inside the transmitter, interfering signals are generated, so we must reduce the power around the channel. Among these, the linearity of the mixer is a problem.

4. Mixer Nonlinearity in Transmitters

At the baseband port of the mixer, nonlinearity may occur due to the baseband circuit or the mixer itself. Referring to the GMSK modulation circuit of the transmitter mentioned above, we previously had a₁Acosφ and a₁Asinφ, but due to nonlinearity, a third component a₃A³cos3φ (additional component) will be generated. Let's see what happens after mixing; the following equation will be obtained at the output terminal:

The equation includes the GMSK modulation part, but additional components can be seen, which is an amplitude problem that can distort the signal and potentially reduce signal quality. In fact, we have generated two GMSK signals, and the second signal has a larger bandwidth. Ideally, we want to enhance the power around the target channel, but this will lead to an increase in adjacent channel power—and we do not allow any power outside the channel.

However, due to this high-bandwidth characteristic, additional power is generated, which is a challenge in terms of linearity. To avoid this problem, we must design a linear mixer.

5. Power Amplifier Nonlinearity in Transmitters

In terms of linearity, the last module of the cascaded system (excluding the matching circuit) is particularly important. We refer to each stage as Stage 1, Stage 2, and Stage 3. The signal swing at the input terminal of the Power Amplifier (PA) is relatively large, and the swing at the output terminal is even larger. Therefore, the PA output terminal has the maximum voltage swing. This stage determines the linearity and compression characteristics of the transmitter, so we must pay attention to the linearity of the PA. Because linearity is crucial here, we must consider the Compression Point (CP).

6. What is the Compression Point?

The Compression Point refers to the point in the graph where the output amplitude difference between a linear amplifier and a nonlinear amplifier at the fundamental frequency is 1 dB. It can also be understood as the point where the output of the nonlinear amplifier is 1 dB lower than that of the linear amplifier. This compression point is used to measure the degree of nonlinearity of the system. For more information about nonlinear systems, refer to "Gain Compression".

This means that if we want the system to maintain linearity, the Power Amplifier (PA) must operate before the compression point. The input power of the PA should not be increased beyond the value shown in the figure above—this value defines the linear operating range. If the input power exceeds this range, linearity problems will occur.

V. EVM Index

Error Vector Magnitude (EVM) is a key index that can reflect system linearity, noise floor, phase noise, and other information simultaneously. Therefore, EVM measurement can provide key references when designing and debugging RF links—almost all types of error sources, such as harmonic distortion, compression effects, noise floor, and phase noise, will be reflected in the measurable EVM value.

1. What is EVM?

EVM is a concise index that can quickly evaluate the overall performance of the system, generally used in communication systems with digital modulation. Qualitatively speaking, it is the distance between the received symbol position and the ideal symbol position in the constellation diagram.

Essentially, EVM measures the degree to which symbols deviate from their ideal positions on the constellation diagram. This deviation may be caused by various system problems, such as nonlinear effects, harmonic distortion, noise figure, and phase noise.

2. What Factors Affect EVM?

The following briefly discusses the error sources affecting EVM in communication systems:

(1) White Noise and EVM

White noise exists in all communication systems. Since white noise affects the Signal-to-Noise Ratio (SNR) of the system, EVM can be expressed by SNR as:

EVMWN = -SNR + PAPR + 3

where SNR is the system Signal-to-Noise Ratio (in dB), and PAPR is the Peak-to-Average Power Ratio (in dB). Note that PAPR is related to the modulated signal.

For ADCs and DACs, the above formula can be expressed by Noise Spectral Density (NSD) as:

EVMWN = NSD + 10logBW + PAPR + Pbackoff + 3

where NSD is the Noise Spectral Density of the converter (in dBFS/Hz), BW is the signal bandwidth (in Hz), and Pbackoff is the difference between the peak signal power and the full-scale range of the converter. The NSD of high-speed converters usually includes thermal noise and quantization noise, so the above formula reflects the combined effect of thermal noise and quantization noise on EVM.

These two formulas indicate that EVM is directly related to signal bandwidth, PAPR, and system thermal noise. Therefore, to pursue high data rates, wireless communication systems often adopt complex modulation schemes with high PAPR and wide bandwidth. However, as mentioned earlier, this comes at the cost of EVM degradation!

(2) Phase Noise and EVM

The phase noise of the system directly affects EVM. The overall phase noise is composed of multiple clock sources in the system, such as the sampling clock of the converter, the Local Oscillator (LO) for frequency conversion, and the reference clock. When calculating the EVM caused by phase noise, it is necessary to integrate the Single-Sideband (SSB) phase noise within the signal bandwidth. Therefore, systems operating at wider bandwidths and higher carrier frequencies are more likely to experience EVM degradation due to phase noise. For more details about phase noise, refer to here...

(3) Nonlinear Effects and EVM

System nonlinearity generates intermodulation products, especially third-order intermodulation products that may fall within the signal bandwidth. These intermodulation products interfere with the amplitude and phase of the carrier, leading to EVM distortion. The mathematical formula used to estimate the EVM caused by third-order products is as follows:

EVMlinearity = 2Prms – 2OIP3 + C

where Prms is the root-mean-square power of the signal, OIP3 is the Output Third-Order Intercept Point of the system, and C is a constant between 0 and 3 dB (depending on the modulation scheme). Obviously, a decrease in OIP3 leads to an increase in EVM, while a decrease in Prms can improve EVM—because lower signal power reduces the power of intermodulation products.

3. Optimizing System Performance Using EVM

System-level performance can be optimized through the EVM Bath-tub Curve. The figure below shows the relationship between EVM and system operating power: at low power, EVM is mainly dominated by the noise floor; at high power, EVM is determined by system nonlinearity. It can be seen from the bath-tub curve that the lower limit of the minimum EVM is jointly determined by all system parameters such as nonlinearity, noise floor, and phase noise. Therefore, EVM is an important tool that can simultaneously optimize all key parameters!