• Abstract

# UWB LNAs for Ground Penetrating Radar

Two LNA topologies for use in UWB systems are examined and compared in this paper. Design specifications have been derived from a ground penetrating radar system which is the target application. Circuits have been designed to cover the frequency range from 3.1 GHz to 10.6 GHz. The SiGe:C technology SGB25V of IHP Microelectronics is applied to enable volume production at low prices. For measurement purposes, input and output are matched to 50 Ω.

SECTION I

## Introduction

DUE to the large bandwidth in ultra wide band (UWB) systems, high resolution can be achieved. As a large amount of information on the scene under test is gathered, object recognition is improved compared to small band systems. Furthermore, materials can be characterized by impedance spectroscopy using different relaxation phenomenons which occur at different frequencies in the UWB range. Thus, UWB systems open a wide field of applications, comprizing surface penetrating radar, industrial sensors, nondestructive testing in the food industry as well as in civil engineering, microwave imaging and others. Several techniques exist to realize a UWB system but for the named applications most often one of the following three techniques is chosen:

• The multisine technique uses a compound of sine signals at different frequencies, amplitudes and phases. They are numerically generated and converted by a DAC to an analog output signal. Thus, the operational speed of the DAC determines system speed in most of all cases. This limitation restricts the bandwidth to about 50 MHz (see [1]) and favors impedance spectroscopy as application.

• The pulse approach is advantageous in low power applications, which do not require digital post processing. This is due to the simple, less stable sampling control most often applied in pulse based systems. Saturation of system components is a further challenge in those systems often preventing monolithic integration.

• The maximum length binary sequence (M-sequence) approach is useful if stable data are required to enable sophisticated post processing. As low amplitude signals are used predominantly, monolithic integration is feasible. In conjunction with the simple control structure, this allows for volume production at low cost SiGe technologies.

A more detailed discussion of these aspects and further techniques is given in [1] and [2]. The following sections focus on design and testing of two UWB LNA topologies tailored for use in an M-sequence system. A basic block level representation of such a system is depicted in Fig. 1.

Fig. 1. Basic M-Sequence Concept (picture adapted from [3]).

A feedback shift register is clocked by a reference oscillator at a single tone fc to generate the M-sequence. More than 70% of signal power is concentrated at frequencies below . Thus, limiting operation to frequencies below this value does not impair the performance significantly. On the other hand, this restriction allows for sampling at fc without aliasing, as the Nyquist condition is kept. A/D conversion can be performed even at a lower rate, because M-sequences are periodic. A complete set of data samples can be gathered during successive signal periods as long as the scene under test can be assumed to be time invariant, i.e., TeX Source $$v_{max} < {c \over 4T_{m}B},$$see [2]. Within this formula, Tm is the observation time, B the bandwidth, vmax the maximum allowable speed of motion in the test scenario and c is the speed of light. Stable, time linear data acquisition is guaranteed by the use of a binary divider which is also driven by the reference oscillator. This is a prerequisite for further data processing, i.e., data reordering, correlation and potentially averaging in the digital domain. Further steps are possible. To enhance the overall bandwidth, system components are to be integrated jointly. Joint integration and the use of a low cost SiGe C1 technology also favor volume production, as system costs can be reduced significantly. Thus, different structures have been compared for LNA design and tied to the features of the low cost SiGe:C technology.

SECTION II

## Circuit Topologies

One challenge arising from the use of the low cost technology is to cope with small self resonance frequencies of the integrated inductors. Only small inductance values can be realized to provide a satisfying high frequency behavior. Additionally, since resonance effects achieved by inductors are limited to a small band of frequencies, reactive networks—or structures based on them—have to be used to cover a larger range of frequencies. This would counteract the target of integration to build low cost structures. To overcome these problems, inductorless designs have been preferred. In Fig. 2 schematics of used structures are presented.

Fig. 2. UWB LNA Topologies. (a) Active Feedback (AF) LNA, (b) Multiple Resistive Feedback (MFB) LNA.

The LNA structure presented in Fig. 2(a) employs a resistively degenerated common emitter amplifier (Q1) in its first stage. A second stage build by an emitter follower (Q3) is attached to drive a 50 Ω load. Input matching is achieved by feedback using a second emitter follower (Q2) as well as the resistors RF1 and RF2. By use of this arrangement, self biasing is achieved, too. The structure has been inspired by the circuit proposed in [4]. In the named publication, an inductor is used to improve the frequency response of the amplifier. For the SiGe:C technology applied here, it turned out to be inefficient and has been abscised from the design. Instead, high frequency behavior is improved by a capacitive peaking technique. Cp1 shortens the degeneration resistor RE at higher frequencies and extends the bandwidth thereby. Gain of this structure is, however, moderate due to its single stage character and resistive degeneration. It is further limited by the parasitic capacitances of the on chip coupling capacitors, which also affect the noise figure. Their influence can be illustrated by the noticeable deviation of post layout results from schematic simulations as shown in Fig. 3 for the active feedback structure.

Fig. 3. Schematic and Post Layout Simulation Results for the AF LNA Layout Version 2.

Owing to the simple structure, parasitic transistor parameters have a noticeable impact on the transfer characteristic as well as input/output matching. Hence, theoretic modeling could be applied for circuit design only qualitatively. It reveals a prevalent influence of RF1 on input matching.

Though facing the same limitations with respect to parasitic capacitances caused by coupling capacitors, better results are obtained by the circuit topology presented in Fig. 2(b), which is covered frequently in literature. Examinations can be found in [5], [6] and [7], for example. In [5] an extensive theoretical treaty of this circuit structure is given. In favor of the following aspects it won't be reproduced here. Unfortunately, sufficient output matching of the basic structure could not be achieved while retaining good values for the other parameters. Since, an emitter follower has been attached to drive the 50 Ω load. As with the Active Feedback LNA presented before, bandwith extension is achieved by capacitive peaking. According to [6], these capacitors might evoke stability problems. Thus, their influence has to be monitored carefully during circuit design.

SECTION III

## Chip Layout

Circuit layouts have been designed for production in the 0.25 μm SiGe:C technology SGB25V of IHP Microelectronics. In order to estimate the influence of layout on circuit performance, two layout versions of the Active Feedback (AF) LNA have been designed. In the first version, 90° lead corners are avoided. They are replaced by two consecutive 45° corners to smooth the change of direction. In consequence, the average lead length is increased. Thus, using 90° corners in the second version, a more compact layout has been created. However, as long as the signal path from the RF input to the first amplifier is kept comparably long, results do not differ significantly. Hence, only simulation and measurement results of the second layout version are presented in the measurement section. Motivated by this result, a compact layout has been chosen to realize the Multiple Feedback (MFB) LNA. In Fig. 4, a chip photograph is presented, in which the single layouts are grouped by the dimension arrows. Ground planes of adjacent LNAs are reused, as the LNAs do not operate simultaneously.

Fig. 4. Chip Photograph—From Top to Bottom: Active Feedback LNA Layout Version 1, Active Feedback LNA Layout Version 2, Multiple Resistive Feedback LNA.

Free spaces have been used to insert stabilizing capacities between the circuit individual Vcc's and ground. Furthermore, a good substrate contact has been achieved by placing PTAPs wherever possible. Neglecting DC-Pads, the circuit dimensions are 630 μm × 280 μm, 530 μm × 280 μm and 530 μm × 280 μm in the order given above.

SECTION IV

## Simulation and Measurement Results

For simulations, arrangements as given in Fig. 2 have been applied. By use of these, operational gain of the amplifiers has been determined. That means, the small signal voltage of the input ports has been set to 1 V and the small signal voltage at the output ports is displayed in dB normalized to 1 V. By use of this representation, results can be compared directly to the measurement results. For noise figure simulation, it is comfortable to use ports, anyway. The obtained results are shown in Figs. 5 and 6. Gain and noise figure have been measured simultaneously on wafer using the noise option of a spectrum analyser in conjunction with a noise source. A measurement setup as proposed in the Agilent application notes 57-1 ”Fundamentals of RF and Microwave Noise Figure Measurements” and 57-2 “Noise Figure Measurement Accuracy—The Y-Factor Method” has been applied. Special to on wafer measurement is the fact that signal attenuation in front of the device under test (DUT) can not be avoided due to the probe heads. Furthermore, it is recommended to add a 10 dB attenuator in front of the DUT to account for insufficient output matching of the noise source. As this would impair the measurement results, the spectrum analyser provides a method for correction. Input and output loss tables can be entered for this purpose. In order to obtain reliable results, attenuation caused by the transmission lines, the 10 dB attenuator, the probe heads and further contact elements has to be determined carefully and to be entered into the loss tables. Furthermore, it is important to shield the circuit during the measurement—especially from any light source, natural or artificial. By this precaution, undesired generation of charge carriers can be diminished. Respecting all these measures, good agreement between the post layout simulation results of the active feedback LNA and its measurement results can be confirmed, as can be seen in Fig. 5(a).

Fig. 5. Simulation and Measurement Results of the AF LNA. (a) Gain and Noise Figure of the AF LNA, (b) S11 and S22 of the AF LNA.
Fig. 6. Simulation and Measurement Results of the MFB LNA. (a) Gain and Noise Figure of the MFB LNA, (b) S11 and S22 of the MFB LNA.

Measured frequency points of minimal reflection coefficients deviate from their simulated values. This might be caused by the interaction of the probe heads with the DUT, but has not been examined yet. However, the reflection coefficients S11 and S22 almost entirely stay below −10 dB over the whole frequency range of interest, which is deemed to be sufficient for the presented application. Further performance characteristics are summarized in Table I.

For the multiple feedback arrangement, deviations are even larger. Additionally, peaking occurs in the gain curve progression of Fig. 6(a) around 8 GHz. Approximately at the same frequency point, the input reflection coefficient S11 in Fig. 6(b) reaches its minimum. Possibly, this is a systematic error due to interaction of the test setup with the DUT, since a similar peaking is observed in [6].

TABLE I Performance Characteristics

Compared to the active feedback implementation, gain is improved significantly. The reflection coefficients S11 and S22 stay well below −10 dB over the whole UWB frequency range. Also, the noise figure has been improved. Further performance characteristics are compared with those of the active feedback LNA in Table I. For comparison, performance characteristics of the structures presented in [4] and [6] are also given in this table.

SECTION V

## Conclusion

In the sections above, design and testing of two UWB LNA structures is presented. They are tied to the requirements of a ground penetrating radar application and produced in the low cost SiGe:C technology SGB25V of IHP Microelectronics. Thus joint integration with other system components to enable volume production becomes possible. Though a low cost technology has been used, good results are presented in the measurement section.

### Acknowledgment

The authors wish to thank the Deutsche Forschungsgemeinschaft (DFG) for supporting this work within the framework of UKoLoS-HaLoS as well as all project partners and subcontractors of the UKoLoS consortium who contributed to this work.

## Footnotes

Markus Robens, Ralf Wunderlich, and Stefan Heinen are with the RWTH Aachen University, Chair of Integrated Analog Circuits, Sommerfeldstr. 24, D-52074 Aachen. mailbox@ias.rwth-aachen.de.

1. A carbon (C) base implant is used in the technology to reduce undesired boron diffusion.

## References

1. Ultra Breitband-Sensoren: Anwendung und Prinzipien

J. Sachs, P. Peyerl

2003-09, 48. Internationales Wissenschaftliches Kolloquium, Technische Universität Ilmenau

2. Stimulation of UWB-sensors: Pulse or maximum sequence?

J. Sachs, P. Peyerl, R. Zetik

Oulu, Finland
International Workshop on UWB Systems, 2003-06, 5–9

J. Sachs, R. Herrmann, M. Kmec, P. Peyerl

Proceedings of German Microwave Conference, Karlsruhe, 2006-03

4. 15 GHz wideband amplifier with 2.8 dB noise figure in SiGe bipolar technology

H. Knapp, D. Zöschg, T. Meister, K. Aufinger, S. Boguth, L. Treitinger

2001 IEEE Radio Frequency Integrated Circuits Symposium, 2001, pp. 287–290, Infineon Technologies AG

5. Analysis, design, and optimization of InGaP-GaAs HBT matched-impedance wide-band amplifiers with multiple feedback loops

M.-C. Chiang, S.-S. Lu, C.-C. Meng, S.-A. Yu, S.-C. Yang, Y. J. Chan

IEEE J. Solid-State Circuits, vol. 37, issue (6), p. 694–701, 2002-06

6. A 3–10 GHz SiGe resistive feedback low noise amplifier for UWB applications

J. Lee, J. D. Cressler

2005 IEEE Radio Frequency Integrated Circuits Symposium, 2005-06, 545–548

7. Design of a SiGe low-noise amplifier for 3.1–10.6 GHz ultra-wideband radio

B. Shi, Michael, Y. W. Chia

Proceedings of the 2004 International Symposium on Circuits and Systems, 2004, I101–I104

## Cited By

No Citations Available

## Keywords

### INSPEC: Non-Controlled Indexing

No Keywords Available

### Authors Keywords

No Keywords Available

### More Keywords

No Keywords Available

No Corrections

## Media

No Content Available
This paper appears in:
International Symposium on Circuits and Systems
Issue Date:
2009
On page(s):
229 - 232
ISBN:
N/A
Print ISBN:
978-1-4244-3827-3
INSPEC Accession Number:
10760428
Digital Object Identifier:
10.1109/ISCAS.2009.5117727
Date of Current Version:
26 Jun, 2009