Low Noise Amplier 2.45 GHz

Transcription

1 Electrical and Information Technology Radio Project ETI041 Low Noise Amplier 2.45 GHz Authors Robin S. Johansson Torbjörn E. Karlsson Supervisor: Göran Jönsson Abstract This report describes the design work of a low noise amplier at 2.45 GHz. Topics of interest are construction of matching network using micro strips, bias network and achieving high gain but still a low noise. Learning by doing mistakes will also be discussed. May 16, 2011

2 Contents 1 Preface 1 2 Design Specication Choice of transistor Transistor S-parameters Bias Matching PCB layout Measurements Gain Compression point Noise gure Third order interception point Conclusion 7 5 Acknowledgment 8 2

3 1 Preface When constructing an RF receiver it is important with low noise at the input stage according to Friis formula. Since the rst component is an antenna ampli- er one can't just strive for high gain but low noise is also important.since it's not possible to have the optimal values for both at the same time, a trade-o between the two is needed. Therefore the design of an LNA is one of the key components in order to build a receiver. In this project the operating frequency is chosen to be 2.45 GHz. For higher frequencies, than approximately 1 GHz, lumped components behaves dierently and thus the theory of transmission lines has to be used instead. 2 Design 2.1 Specication Operating frequency: 2.45 GHz Noise gure: F 3.1 db Gain: G t 11 db Transistor: Inneon BFP420 Quiescent point: V CE = 4 V, I C = 3 ma Source impedance: 50 Ω Load impedance: 50 Ω Matching network at the output: Conjugate match Matching network at the input: None 2.2 Choice of transistor In order to make an LNA the choice of transistor is crucial. For the frequency of interest the number of transistors are limited to a few. Comparison between several transistors led to the choice of BFP420 from Inneon. Figure 1 shows how the pins were congured. The values only dier slightly and the measured values were chosen for calculations. 2.3 Transistor S-parameters In order to use the transistor measurements of the s-parameters need to be done. In gure 2 the measured s-parameters is compared with the s-parameters that Inneon measured in the specication of the transistor.

4 Figure 1: Pin conguration of BFP420[1]. Parameter Measured Inneon S i i S i i S i i S i i Figure 2: The measured s-parameters compared with Inneon's measured s- parameters at 2.45 GHz, V CE = 4 V and I C = 0.3 ma [2]. 2.4 Bias Since low noise was crucial, gure 4 from the data sheet, was studied in order to nd a good value for the collector current. At 2.4 GHz a collector current around 3 ma gives the lowest value for the noise gure. However at V CE = 2 V, which is used in gure 4 the gain was not sucient. To achieve higher gain the voltage needs to increase. The maximum value according to the data sheet is 4.5 V and therefore V CE = 4 V was chosen to have a small margin of safety. According to Sundström, Jönsson and Börjesson it is practical to choose I D = I B β0 = I C / β 0 [3]. By using the values in gure 3 and V CC = 8 V one gets V CE = 4 V, I C = 2.99 ma and a loop gain of R B1 R B2 R B3 R C Values (kω) Figure 3: The chosen biasing resistor values. 2

5 Figure 4: Noise gure as a function of collector current for dierent frequencies with V CE = 2 V. Figure 5: The passive biasing network used in the amplier [3]. The biasing network shall not only feed the transistor with the correct qui- 3

6 escent values but also have good temperature independence and provide high loop gain. The passive network presented in gure 5 have those specications and is relatively simple and was therefore chosen. 2.5 Matching In gure 6 the stability-, noise- and gain circles are drawn. Since the stable input region, the noise circle and the gain circle surrounds the origin no matching network is needed at the input. At the output on the other hand a conjugate match will be used. Figure 6: The green circle is the input stability circle, the red circle is the output stability circle, the blue circle is the noise circle and the cyan circle is the gain circle. The matching network at the output is constructed with two transmission lines. From the transistor a transmission line is placed followed by a shortcircuited stub that will act capacitive. Start at the conjugate of S 22, which is the red dot in gure 6, follow the constant radius circle counter clockwise until the intersection with the help circle is reached. That gives the length of the transmission line. Then add a short-circuited stub of length enough to reach the origin of the smith chart which will fulll the matching condition. 2.6 PCB layout The designed PCB layout looks like gure 7. There is one error in the layout that made the output matching network useless. The error was noticed too late to be corrected before the fabrication of the PCB and that is the reason why some later measurements might not be as expected or even absent. The way it is done on the PCB the transmission line in the matching network will not have any eect since it is connected to the 50 Ω output. 4

7 Figure 7: The layout of the amplier. 3 Measurements 3.1 Gain In gure 8 the gain and noise gure is shown from 2 GHz to 3 GHz. There are three peaks in the gain measurement. Two of the peaks are outside the region of interest and the interesting peak reaches almost 10 db. In gure 10 only the interesting band is shown. Figure 8: Noise and gain level measurement. 5

8 3.2 Compression point The compression point was measured for two dierent frequencies. At 2.4 GHz the compression point was -7.2 dbm and at 2.45 GHz it was -8.6 dbm. The measurements are presented in g 9 Figure 9: Compression point measurements. 6

9 Figure 10: Noise and gain level measurements. 3.3 Noise gure The noise level for the LNA at the signal band GHz is also presented in gure 10. It is changing slightly from 2.1 db at the lower frequencies to 1.5 db in the upper part of the band. 3.4 Third order interception point The third order interception point was calculated with a handy formula [4]. IP 3 = P (1) out + P (1) out P (3) out = 2.4 dbm 2 P (1) out is the level of the signals and P (3) out is the level of the third order intermodulation. 4 Conclusion The amplier did not fulll all specications. The transistor had excellent noise performance and fullled the specication with good margin. The amplication however was not fullled most likely due to our incorrect matching network. As can be seen in the measurements we had several amplication peaks and our guess is that the peak around the interesting frequency would be higher with a correct matching network. The compression point and third order interception point is reasonable even though we had a large error in our design. We don't know if this would be the case with a correct matching network. 7

10 During the construction of the matching network we mixed the output re- ection coecient from the transistor with the output reection coecient from the amplier. This tricked us to do a reversed matching network that doesn't make any sense at all. 5 Acknowledgment We thank Göran Jönsson for an endless source of useful information and help in the lab. Andreas Axholt's help with the Eagle software was also of great importance. At last we thank our fellow students for the interesting conversations we've had during the project. References [1] Inneon Technologies AG, BFP420 data sheet, 2007 [2] Type=db3a b532e0124c9cec25d6360, 2011 [3] Radio Electronics, L. Sundström, G. Jönsson, H. Börjesson, Department of Electroscience, Lund University, 2004 [4] Göran Jönsson, Department of Electrical and Information Technology, Lund University,