WO1996025791A1

WO1996025791A1 - Amplifiers

Info

Publication number: WO1996025791A1
Application number: PCT/GB1996/000343
Authority: WO
Inventors: Mohammad Reza Moazzam; Colin Stuart Aitchison
Original assignee: British Technology Group Limited
Priority date: 1995-02-15
Filing date: 1996-02-15
Publication date: 1996-08-22
Also published as: GB9502894D0; EP0809881A1; JPH11500276A

Abstract

The third order intermodulation product distortion of a high-frequency amplifier is reduced by means of a correction signal comprising the second order harmonic of the source signal, the amplitude and phasing of the correction signal being selected at least partially to compensate for distortion generated in the amplifier.

Description

Amplifiers

This relates to amplifiers and, in particular, to methods of reducing the third order intermodulation product levels in non-linear MESFET amplifiers. This method is based on the injection of a harmonic signal to the amplifier system and is applicable to other three-terminal active devices such as HEMTs and HBTs.

A common performance parameter of amplifiers used in communication systems, is the effect of nonlinearity on the amplifier performance. The intermodulation products are regarded as the most troublesome distortion in communication system amplifiers. Many attempts have been made to reduce intermodulation products levels. Beside linearisation techniques, other approaches such as optimising the load impedance or terminating the second harmonic voltage in a short circuit have been employed for this purpose. Applying the above techniques to improve intermodulation levels may prevent the designer from using the full capability of the active device or alternatively the required circuitry may be rather complex, expensive and large in size. For example, the use of negative feedback methods for reducing the intermodulation distortion in non-linear amplifiers causes a reduction in the amplifier gain, and its alternative, the feedforward technique requires a second high performance amplifier closely matched to the main amplifier. The third order intermodulation product is the most dominant intermodulation distortion in non-linear amplifiers which are employed in communication systems. We have devised a novel method for reducing the third order intermodulation product levels in non-linear amplifiers. With this method, not only there is no trade off between the gain and the levels of the third order intermodulation but also the required circuitry is simple, inexpensive and small in size.

According to the present invention there is provided an amplifier into which a correction signal comprising the second order harmonic of the source signal is fed, the amplitude and phasing of the correction signal being selected at least partially to compensate for distortion generated in said amplifier. The method is based on using non-linearity of the amplifier to cancel out the third order intermodulation product. In this method the second order harmonics of the source signals are injected into the amplifier as well as the fundamental signals. Non-linearity of the amplifier causes interaction between the source signals and their injected second order harmonics. This interaction results in additional signals at the output of the amplifier at the third order intermodulation frequencies. On the other hand there are components of the third order intermodulation product due to the interaction between the fundamental signals. By proper selection of phase and amplitude of the injected second harmonics, it is possible to make the third order intermodulation product produced by the second harmonics and the original third order product out of phase and equal in amplitude. As a consequence the third order intermodulation distortion is effectively eliminated.

The invention will now be particularly described with reference to the accompanying drawings, in which: —

Figure 1 is an amplifier circuit showing injection of the second harmonics;

Figures 2 to 4 and 6 are illustrative graphs, Figure 5 is a circuit diagram illustrating an alternative embodiment of the invention, and Figures 7 - 10 are oscilloscope traces of practical measurements. Referring now to the drawings, in order to analyse the technique mathematically, a simplified non-linear model is used for the MESFET transistor. In this model the transconductance is regarded as non-linear as it is the dominant nonlinearity in class B amplifiers. This nonlinearity can be approximated by a three-term power series expansion for the drain current, i_d , as

id = gml _in + gπϋ V_in ² + g_m3 V_in ³ ( 1 )

in which v_in. is the gate to source voltage. Writing the signals from two fundamental signal sources as A, cos(ω-t) and A^, cos(ω t) and the corresponding injected second harmonics as A,, cos(2ω,t + φ,) and A₂₂ cos(2ω₂t + Φ₂) ( φi and φ_> are the phases introduced at the second harmonics), the input signal to the amplifier is v_in = A, cos(ω-t) + A₂ cos(ω₂t) + A,, cos(2ω,t + φ,) + A₂₂ cos(2ω₂t + φ₂) (2)

Substitution of Eq. 2 into Eq. 1 gives all the components in the output spectrum. For brevity here, we only show the third order components. For the third order intermodulation product type (2ω₂ 1 - ω, t), we have A, A₂₂g_m2 cos(2ω₂t - ω,t+ φ₂)

+

(3) 4 2

The second term in Eq. 3 is the result of interaction between fundamental signals. The first and the third terms are the consequence of injection of the second harmonics into the amplifier. The third term in the above equation is small in comparison to the two other terms and can be ignored.

In order that the first term of Eq.3 cancels the second term, given total suppressing of (2ω₂t-ω,t) the following condition must be satisfied: A₂₂ = 2A₂ig„,₃- and |φ₂| = 180°

4g_m2 Similarly for suppression of the (2ω,t-ω₂t) term we must have

A„ = , g_m3. and |φ, | = 180° 4g_m2

These conditions would result in total cancellation of both terms of the third order intermodulation product.

The feasibility of the idea was confirmed by the CAD simulation. For this investigation the circuit structure of Figure 1 was used. This circuit has a wideband input configuration which consists of a T-section with its corresponding m-derived half T- sections to ensure a wideband match at the input. This configuration is used so that the input circuit bandwidth does not affect the predicted performance. A narrow-band parallel resonant circuit is used at the output together with a transformer to provide the optimum load for the amplifier. The drawing also shows the second harmonic frequency signal generators with phase adjuster also connected to the input of the amplifier (due to simulator limitations only one signal generator generating the second harmonic of one of the test two tones is used at a time). The transistor was modelled by its Curtice model (MESFET Model 2 in LIBRA). The fundamental input signals are arbitrarily chosen at frequencies 2.5 GHz and 2.51 GHz. The results of this investigation is shown in Figure 2 and 3. A typical fundamental and third order intermodulation product power versus frequency plot is drawn in Figure 2 when no external second harmonic is injected to the circuit. (This drawing shows a dip in the intermodulation curve. This dip is a result of cancellation between different nonlinearities namely transconductance and drain conductance). The simulated effect of the injected second harmonic level on the third order intermodulation at lower power levels (below the dip region in the third order intermodulation curve) is shown in Figure 3. As can be seen from this figure, there is no change in the fundamental power at the output while there is a reduction in the relevant third order intermodulation by about 30 dB which is clearly a substantial improvement.

It was observed that there are some modest discrepancies between required simulated and calculated phase and amplitude of the injected second harmonic which occurs because of the effects of the elements (especially feedback elements) of the transistor which are ignored in mathematical analysis. We have found that this technique can be employed at higher power levels than those considered above and even at IdB gain compression point (Figure 4 ). However, the required phase and amplitude of the injected harmonic are considerably different in the two regions. In addition, the dynamic range of the technique at higher power levels is relatively narrow, though it is useful for a number of communication systems. Figures 8 and 9 are photographs of the spectra in which both second harmonics were fed into the input via an atentuator and phase changer. This reduced the IM₃ level by better than 16dB on both intermodulation sidebands with the use of only one path to feed both second harmonics.

The mathematical analysis and CAD simulation showed that the fifth order intermodulation product remain almost unchanged using this technique. In addition, the technique is also applicable where more than two fundamental signals are involved.

The second harmonic signal may be injected at ei er the input or output of the amplifier. It is also possible to provide this harmonic at the input by feeding back the second harmonic generated by the nonlinearity of the amplifier from output to the input and the same benefit is obtained (Figure 5 and 6). This also illustrates that the second harmonic technique reduces both terms of the third order intermodulation product simultaneously. The mathematical analysis and CAD simulation of the performance of a MESFET non-linear amplifier reveal that the level of the third order intermodulation product can be reduced by generating and injecting the second harmonics of the input signals. The third order intermodulation levels at lower or higher power level or indeed at the IdB compression point can be reduced substantially by injecting the second harmonic signals with appropriate phase and amplitude into the amplifier. Non-linear simulation predicts a reduction of more than 30 dB in the third order intermodulation levels. The required second harmonic signal can be injected to either input or output of the amplifier. It is also possible to provide this signal through a feedback path which only feeds back the generated second harmonics from output to the input of the amplifier. The absence of trade-off between the gain and the third order intermodulation level (and higher efficiency as a result), simple circuitry and small size are among other advantages of this technique.

In an embodiment of the invention, two 500MHz signals separated by 1MHz were fed into an MESFET amplifier together with the appropriately phased second harmonic of one signal obtained using a diode doubler. Adjustment of the level of the second harmonic showed a reduction in the corresponding their order intermodulation sideband by 19dB without significant change in the fundamental levels and the other intermodulation sideband. Figures 7 and 8 are traces respectively showing the intermodulation spectrum with and without the technique applied.

Claims

1. An amplifier into which a correction signal is fed characterised in that said correction signal comprises the second order harmonic of the source signal.

2. An amplifier as claimed in claim 1 characterised in that third order intermodulation distortion is reduced by injecting the second order harmonic of the source signal.

3. An amplifier as claimed in claim 2 characterised in that said second order harmonic is injected at the input of said amplifier.

4. An amplifier as claimed in claim 2 characterised in that said second order harmonic is injected at the output of said amplifier.

5. An amplifier as claimed in claim 3 characterised in that said second order harmonic is obtained by feedback from the output of said amplifier and further characterised in that said feedback path is adapted to feed back substantially only said second order harmonic.