ADS simulation and design of broadband low noise amplifier

O Introduction

The low noise amplifier (LNA) is an important component of the front end of the RF receiver. Its main function is to amplify the weak received signal. The high enough gain overcomes the noise of the subsequent stages (such as the mixer) and reduces the interference of the additional noise as little as possible. The LNA is generally directly connected to the antenna or antenna filter through a transmission line. Since it is at the front end of the receiver, its ability to suppress noise is directly related to the performance of the entire receiving system. Therefore, the indicators of LNA are becoming more and more stringent. Not only a sufficiently small low noise figure is required, but also a sufficiently high power gain, a wide bandwidth, and good power gain flatness within the receiving bandwidth. This design uses the ADS software in the field of microwave design, combined with the low noise amplifier design theory, and uses S parameters to design a low noise amplifier with a simple and compact structure and good performance indicators.

1 Design indicators

The following are the indicators that need to be considered for the designed broadband low noise amplifier:

(1) Operating frequency band: 10-13 GHz. The operating frequency band refers only to the frequency band range in which the power gain meets the flatness requirement, and the noise figure must also meet the requirements in the entire frequency band.

(2) Noise figure: FN < 1.8 dB. FN represents the ratio of input signal-to-noise ratio to output signal-to-noise ratio. Ideally, the amplifier does not introduce noise, the input/output signal-to-noise ratio is equal, and FN = 0 dB. A lower FN can be obtained by matching the input to the best noise matching point and adjusting the static operating point of the transistor. Since it is a broadband amplifier, it is difficult to obtain a lower noise figure, which determines that the noise figure of the system will be relatively high.

(3) The gain is 25.4 dB. The LNA should have a high enough gain to suppress the impact of the subsequent stages on the system noise coefficient, but its gain should not be too large to avoid nonlinear distortion in the subsequent mixer.

(4) Gain flatness is 0.3 dB. It refers to the gain fluctuation within the operating frequency band. The low noise amplifier should maintain a relatively flat gain level. Since it is a broadband amplifier, the gain flatness is relatively small. The matching circuit should be used in the high frequency band to make the low end of the frequency band mismatched, thereby improving the gain flatness of the amplifier.

(5) Input/output matching. In order to achieve good noise performance, the input port is usually mismatched. At this time, the gain will decrease and the port standing wave ratio performance will deteriorate. In addition, since the gain of the microwave transistor itself decreases by about 6 dB per octave, in order to obtain good gain flatness within the operating frequency band, a certain port standing wave performance must be sacrificed.

(6) Stability. It is the basic condition to ensure the normal operation of the amplifier. When the reflection coefficient modulus of the input and output ends of the amplifier is less than 1 (i.e. |Γ1|<1, |Γ2|<1), the network is stable regardless of the source impedance and load impedance, which is called absolute stability; otherwise, it is called relative stability. For a conditionally stable amplifier, its load impedance and source impedance cannot be selected arbitrarily, but have a certain range, otherwise the amplifier cannot work stably.

According to the above discussion, the focus of this design is to ensure low noise figure and flat gain within a wide bandwidth. In order to ensure the realization of the above design indicators, a two-stage cascade design scheme is adopted: the first stage designs the input matching circuit according to the minimum noise to obtain an excellent noise figure; the second stage designs the output matching circuit according to the maximum power criterion to obtain the maximum amplification gain. Gallium arsenide field effect transistors (GaAsFET) are generally selected for designing LNA. Their advantages are high frequency, low noise, fast switching speed and good low temperature performance. This article uses NEC's gallium arsenide heterojunction field effect transistor NE3210S01.

2 Design Scheme

2.1 Stability Analysis

The conditions for determining the stability of the amplifier are as follows:

Where: △ = S11S12-S12S21; K is the stability factor. When the above three conditions are met at the same time, the amplifier is absolutely stable.

According to the S parameter model of NE3210S01, the amplifier is not absolutely stable in the full frequency band through software simulation calculation. The series resistor in the drain can effectively improve the stability without increasing the complexity of the design. In the design, a 10Ω resistor is connected in series with the drain of the first-stage amplifier to keep the amplifier absolutely stable in the full frequency band, while the effect on the gain is very small. There is internal feedback in the high-frequency amplifier tube. When the feedback reaches a certain intensity, it will cause the stability of the amplifier to deteriorate and lead to self-excitation. Therefore, the absolute stability of the amplifier must be guaranteed. If the amplifier does not meet the absolute stability conditions, some measures need to be taken to improve the stability of the amplifier. The main methods are: source series negative feedback; parallel negative feedback between the drain and the gate; drain series resistor and drain parallel resistor; inserting a ferrite isolator.

2.2 Input matching circuit

The block diagram of the two-port network of the microwave device is shown in Figure 1. Among them, Γ1 and Γ2 are the input and output reflection coefficients respectively; Γs and ΓL are the reflection coefficients of the source and load respectively.

The input matching circuit design in Figure 1 mainly considers the noise coefficient of the amplifier, which can be expressed as follows:

Where: FNmin is the optimal noise coefficient; Γs is the source reflection coefficient, Γopt is the optimal source reflection coefficient; RN is the equivalent noise resistance. When Γs=Γopt, the minimum noise coefficient FNmin can be obtained. However, the noise cancellation between the reactive components is achieved through the mismatch at the input end, so in general, the input standing wave ratio is relatively large, which will also reduce the gain of the amplifier. It is necessary to comprehensively consider the trade-off between the noise coefficient and the input standing wave ratio.

The matching circuit adopts the microstrip impedance transformation matching method, which is equivalent to several microstrip lines connected in series. In the matching process, the Smith chart can be used to obtain a suitable LC matching circuit, and then the microstrip line calculation tool included in ADS can be used to solve the equivalent microstrip line circuit form. The advantage of this matching method is that the size can be greatly reduced in the high frequency band, the circuit size will be more compact compared with the branch line matching, and it is suitable for constructing broadband matching. The number of series microstrip lines can be appropriately increased to ensure that a better gain flatness is achieved under broadband conditions.

2.3 Inter-stage matching circuit

Since the two-stage cascade design is adopted, a reasonable inter-stage matching circuit will have an important impact on the overall performance of the circuit. The purpose of the inter-stage circuit is to make the input impedance of the rear-stage microwave tube conjugately matched with the output impedance of the front-stage microwave tube to obtain the maximum gain while taking into account the requirements of output flatness. The inter-stage circuit uses a total of 4 microstrip lines, and the increased size parameters improve the output flatness. A DC blocking capacitor needs to be added between the two stages, but since it is difficult for the DC blocking capacitor to maintain good characteristics in the X-band, a λ/4 coupled microstrip line is used instead in the circuit. The coupling line width is taken as 0.2 mm and the coupling gap is 0.1 mm. The DC blocking effect is good and the transmission loss is small in a very wide frequency band.

2.4 Output matching circuit

As shown in Figure 1, the input matching circuit of the second-stage two-port network is actually an inter-stage matching circuit. The output matching circuit is designed according to the maximum power gain criterion, and the conjugate matching method is adopted. It is required that the output impedance of the inter-stage circuit is conjugate matched with the input impedance of the subsequent microwave tube, and the output impedance of the subsequent microwave tube is conjugate matched with the input impedance of the output matching circuit. The amplifier has the maximum power gain and the best port standing wave ratio performance. When the source and the load are both 50 Ω, the actual power gain of the amplifier is:

2.5 Bias circuit design

Since the noise coefficient is closely related to the static operating point of the transistor, a suitable bias circuit must be selected to allow the amplifier to work in the best state. The circuit is powered by dual power supplies. The so-called dual power supply means that the positive drain voltage and the negative source voltage are powered by two positive and negative power supplies respectively. In the preliminary circuit design, the static operating point of the amplifier can be set to VD=2 V and IDS=10 mA using a series voltage divider resistor according to the bias conditions provided by the S parameter model of the device. The feeding method selects a λ/4 high-resistance microstrip line terminated with a 70° fan-shaped line. The λ/4 high-resistance microstrip line is used to curb the influence of the AC signal on the DC power supply. The fan-shaped line is short-circuited at high frequencies and is equivalent to a capacitor. It can filter out power supply noise and is especially suitable for wide-band design. When in the low frequency band, attenuation is introduced to eliminate the gain peak and improve the gain flatness. In the subsequent adjustment and optimization process, the voltage divider resistor can be appropriately changed to pursue better overall performance.

3 Simulation and optimization

First, the dielectric parameters must be defined in ADS. This paper uses Rogers4003 dielectric plate. When performing ADS simulation, the dielectric constant εr=3.38 and the dielectric plate thickness h=0.5 mm need to be set.

Secondly, we need to build a model of the transistor core. There are two forms of models. One is the SP model: it is a small signal linear model. The model already has a certain DC operating point and S parameters within a certain range. Pay attention to the applicable scope when simulating. This model can only get preliminary results, which is sufficient for some applications. It cannot be used for large signal simulation or DC feed circuit design, and cannot directly generate layouts. The other is the SPice circuit parameter model of the transistor, which is generally provided by the chip company. You can install the Design Kit provided by NEC in ADS. The toolkit integrates FET, JBJT, HJ-FET of the NEC series of low-noise amplifiers. Select NE3210S01 in FET. Since the components in the Design Kit are already packaged transistors, their simulation results are more reliable than the transistor model using the S parameter model. In many cases, before simulating and designing the package model, the approximate indicators of the circuit can be obtained by simulating the SP model in advance.

Considering the parasitic effect of vias, the simulation effect of the circuit is greatly affected in the high frequency band, so a ground via is added between the transistor source and the ground. A step change joint or T-joint is used at the microstrip line connection to obtain a more accurate simulation model. At the front end of the input and output, a standard 50 Ω transmission line is used to connect to the λ/4 coupled microstrip line.

The above simulations are the approximate dimensions of the microstrip line obtained by simulation at a single frequency point of f=12 GHz. In order to keep the circuit at a bandwidth of 3 GHz, it is necessary to optimize the circuit. Before optimization, you can use the tuning tool to manually adjust the parameters of each component to observe which parameters are more sensitive to the performance of the circuit and which should be given priority in optimization.

Common optimization methods are divided into random optimization and gradient optimization. Random methods are usually used for large-scale searches, while gradient methods are used for local convergence. A small number of variable parameters can be set during optimization, and each index of the amplifier can be optimized in steps. First, use the random method for 100 to 200 steps, and then use the gradient method for 20 to 30 steps. Generally, the best result can be achieved.

Finally, perform the overall simulation to see if the index requirements are met. If the optimization result does not meet the requirements, it is generally necessary to reset the optimization range of the parameters, the optimization target, or consider changing the topology of the circuit, and then re-simulate the optimization. In the simulation, the accuracy and minimum size of the actual microstrip line processing should be taken into account. According to the processing accuracy, some lines are too thin and cannot be achieved. In addition, the pursuit of multiple digits of precision after the decimal point is meaningless. Generally, the width of the microstrip line should not be less than 0.2 mm, and 2 digits after the decimal point can be retained (unit: mm).

After repeated optimization simulations, the final parameters met the proposed design indicators: in the 10-13 GHz frequency band, the noise figure is less than 1.8 dB, the gain is 25.4 dB±0.3 dB, the output standing wave is less than 1.6, and the input standing wave is less than 2. The various parameter indicators after ADS optimization are shown in Figures 2 and 3.

4. Layout Design

Use AutoCAD to draw the final optimized result into a layout. Note that small islands should be added to the matching microstrip line for later debugging. The size of the microstrip line can be appropriately changed to obtain better performance. Copper is attached to a large area around the circuit, and denser metallized grounding vias are left to enhance the grounding performance of the circuit. Screw holes are left at the four corners to fix the circuit board in the metal shielding box. The final layout is shown in Figure 4.