How to evaluate the performance of an amplifier predistorter (Part 2)

JasonYoo

How to evaluate the performance of an amplifier predistorter (Part 2) [Copy link]

The overcompensated feedforward (OCFF) method of amplifier linearization can reduce the size of the amplifier, save cost, and improve the overall efficiency of the power amplifier system.

The estimation of the quality factor of the OCFF method can be considered to be based on this assumption: the amplitudes of the distortion vectors are very consistent, but there is an error Θ between the phases. Figure 3(a) is the vector diagram of the FF method. For the initial distortion vector, the suppression of the distortion vector is obtained by calculating 20log[sin(Θ/2)] using elementary geometry knowledge. Figure 3(b) is the vector diagram of the OCFF method. All vectors are vectors of the reference point B in Figure 1(b). Due to the distortion caused by amplifier A1 and the shunt branch, the angle of the resultant vector is Φ. Assuming that the phase angle of the distortion vector caused by A1 is 0 degrees, for the initial distortion vector, the final suppression of the distortion vector is 20log[sin (Φ/2)]. From elementary geometry knowledge, it can be seen that

Figure 4 shows the suppression as a function of the phase error θ. Thus, the OCFF method is always inferior to the FF method in terms of quality factor, and when G _shunt is infinite, OCFF approaches the performance of FF. However, increasing G _shunt also means reducing the distortion caused by A1, thus making the goal of the OCFF method unattainable. For larger G _shunt , the noise of the system will also increase. However, in the OCFF method, the reduction in loss at the output of A1 is greater than the amount of loss compensated in the suppression.

If the imaginary part of the nonlinear transfer function of the active device is negligible, a distortion cancellation of about 20 dB is achievable. Amplifier A2 must be linear, but does not have to deliver high power compared to the conventional FF approach because it powers the distortion elements at the input stage rather than at the output stage.

For Class AB operation, fifth-order and higher IMD must be considered. Modeling becomes more complicated when the phase of the IMD is power dependent. Therefore, the problem can be reduced to matching the nonlinear transfer functions of A1 and A3 in a suitable way to obtain the optimal suppression of IMD (discussed further later). It should be noted that the method just mentioned is a special case of this approach for Class A operation. Intuitively, the OCFF is in optimal operation when A1 and A3 operate at the same conduction angle.

In both cases, assuming P2 _is the output power, the efficiencies of the FF and OCFF methods are similar. The relevant parameters are pure numbers (unitless) and can be expressed as follows:

GA1 , GA2 _, and GA3 _are the gains of amplifiers A1, A2, and A3, respectively _;

c1, c2 and c3 are the coupling coefficients of couplers C1, C2 and C3 respectively (c1, c2 and c3 are all less than 1); and l _m is the loss of the main path, and l _m = (1-c1)l _d (1-c2). Among them: l _d is the loss of the delay line of the main path.

When calculating the total efficiency, the power consumption in A3 has little effect as long as the gain GA1 _is large enough. Therefore, only the power consumption in A1 and A2 can be considered. If η1 is the efficiency of amplifier A1, the DC power consumption of A1 for the FF and OCFF cases is:

To calculate the DC power consumption of amplifier A2, two distinct cases should be considered. In the first case, the output of the distortion element H1 is several orders of magnitude less than the output of the distortion element of the main carrier, which is generally true for Class A operation. Therefore, A2 essentially determines the power of the suppressed carrier determined by the amount of cancellation of H1. The RF power at the output of A2 is P2sGA2/l m and P2sGA2/(GA1l m) for the FF and OCFF cases, respectively _{, where} s _is the _{carrier suppression} achieved at the _{output of} H1 .

Therefore, the DC power dissipation of A2 in the FF and OCFF states is:

A2 can control the input signal with a large peak-to-RMS ratio, so A2 is usually selected as a Class A amplifier. It can be seen that although the amplifier has greater energy in the FF state than in the OCFF state, it has the same efficiency η ₂ in both states .

Assuming that the total efficiency in the FF and OCFF methods is η _FF and η _OCFF respectively , it can be seen from equations 8(a), 8(b), 9(a) and 9(b) that:

When l _m G _A1 > 1 and l _m < 1, the efficiency of an OCFF system is often greater than that of a FF system.

In the second state efficiency estimation, the harmonic distortion of the output H1 is used as the main carrier frequency, which is often correct for class AB amplifiers. For an FF system, the RF power at the output of A2 is set to P2 ₍ δ/c2), where δ is the ratio of the distortion power to the carrier power before linearization. For an OCFF system, the RF power at the output of amplifier A2 is (P2 _/ GA1 ₎ (δ/c2)o _f , where o _f is the transition compensation coefficient (greater than 1) that can be used to account for the transition signal.

The first state is similar, the power ratio of the OCFF and FF systems is expressed as follows:

In practical applications, G _A1 > o _f and in state 1 the efficiency of the OCFF system is often greater than that of the FF system.

Based on the above analysis, it is clear that we need to improve the efficiency of the OCFF method compared to the traditional FF method, including eliminating the main channel loss and applying a smaller error amplifier (A2). The OCFF method can use a delay line with higher loss and lower cost than the FF method, such as a lumped element delay line. The lower power loss of A2 means lower cost and smaller overall system size.

The experiment used a class A amplifier to prove the OCFF theory. The third-order cutoff output power of the main amplifier A1 and the driver amplifier is +35.5 and +30.5dBm respectively. The main channel delay line has a loss of 2.5dB at 800MHz and a delay of about 10ns. The results are shown in Figures 5(a) and 5(b). Without linearization, the third-order intermodulation distortion index of amplifier A1 is: when the output power is +15dBm per tone, it is -40dBc at a bandwidth of 700MHz and -42dBc at a bandwidth of 800MHz. After applying the OCFF linearization circuit, the third-order intermodulation distortion suppression index is: greater than 18 and 15dB at a bandwidth of 700 and 800MHz, respectively.

Compared to FF technology, OCFF technology provides linear performance. The most attractive aspect of this technology is that any output power loss can be eliminated through the delay line and coupled FF configuration. Similarly, OCFF technology also improves the overall efficiency of the system. This technology can also be applied to circuits with essentially different inputs and outputs, such as in optoelectronic conversion circuits or linear up-conversion circuits near intermediate frequencies (IFs). The OCFF method allows the linearization of directly modulated lasers, where the distortion is mainly determined by the link distance caused by the interaction of the source modulation frequency chirp with the fiber propagation. The application of the OCFF method has no bandwidth restrictions and can also be well implemented in situations where both even and odd-order harmonic distortion are constrained. [ip]