Optimized (2nd Pass) Gallium Arsenide (GaAs) Integrated ...
Transcript of Optimized (2nd Pass) Gallium Arsenide (GaAs) Integrated ...
Optimized (2nd Pass) Gallium Arsenide (GaAs) Integrated
Circuit Radio Frequency (RF) Booster Designs for 425 MHz and Dual Band (425 and 900 MHz)
by John Penn
ARL-TR-5396 November 2010
Approved for public release; distribution unlimited.
NOTICES
Disclaimers The findings in this report are not to be construed as an official Department of the Army position unless so designated by other authorized documents. Citation of manufacturer’s or trade names does not constitute an official endorsement or approval of the use thereof. Destroy this report when it is no longer needed. Do not return it to the originator.
Army Research Laboratory Adelphi, MD 20783-1197
ARL-TR-5396 November 2010
Optimized (2nd Pass) Gallium Arsenide (GaAs) Integrated Circuit Radio Frequency (RF) Booster Designs for
425 MHz and Dual Band (425 and 900 MHz)
John Penn
Sensors and Electron Devices Directorate, ARL Approved for public release; distribution unlimited.
ii
REPORT DOCUMENTATION PAGE Form Approved OMB No. 0704-0188
Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing the burden, to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.
1. REPORT DATE (DD-MM-YYYY)
November 2010 2. REPORT TYPE
Final 3. DATES COVERED (From - To)
4. TITLE AND SUBTITLE
Optimized (2nd Pass) Gallium Arsenide (GaAs) Integrated Circuit Radio Frequency (RF) Booster Designs for 425 MHz and Dual Band (425 and 900 MHz)
5a. CONTRACT NUMBER
5b. GRANT NUMBER
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S)
John Penn 5d. PROJECT NUMBER
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)
U.S. Army Research Laboratory ATTN: RDRL-SER-E 2800 Powder Mill Road Adelphi, MD 20783-1197
8. PERFORMING ORGANIZATION REPORT NUMBER
ARL-TR-5396
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES)
10. SPONSOR/MONITOR’S ACRONYM(S)
11. SPONSOR/MONITOR'S REPORT NUMBER(S)
12. DISTRIBUTION/AVAILABILITY STATEMENT
Approved for public release; distribution unlimited.
13. SUPPLEMENTARY NOTES
14. ABSTRACT
High-performance microwave and radio frequency integrated circuits are of interest to the Army. Several monolithic microwave integrated circuits (MMICs) were designed to enhance the performance of commercial-off-the-shelf (COTS) RF integrated circuits (RFICs) used in many wireless systems. This report documents a set of MMIC designs optimized for the 400 to 450 MHz ultra-high frequency (UHF) band and a dual band design that also includes 850 to 950 MHz operation. Additional incorporation of discrete matching elements into a single integrated IC will improve size and weight of wireless systems. This is an optimized set of designs based on a previous 1st pass design effort. Ten separate MMIC designs were designed and fabricated.
15. SUBJECT TERMS
MMIC, RFIC
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF ABSTRACT
UU
18. NUMBER OF PAGES
96
19a. NAME OF RESPONSIBLE PERSON
John Penn a. REPORT
UNCLASSIFIED
b. ABSTRACT
UNCLASSIFIED
c. THIS PAGE
UNCLASSIFIED 19b. TELEPHONE NUMBER (Include area code)
(301) 394-0423 Standard Form 298 (Rev. 8/98)
Prescribed by ANSI Std. Z39.18
iii
Contents
List of Figures iv
List of Tables vii
Acknowledgments viii
1. Introduction 1
2. Lessons Learned from 1st Pass Design 1
3. Component Designs and Simulation Results 3
4. Active GaAs IC Designs 25
5. Integration of RFIC Matching Circuits 63
6. Design Rule Checking (DRC) 80
7. Layout versus Schematic (LVS) Checking 80
8. Tile Layout 81
9. Conclusion 83
10. References 84
List of Symbols, Abbreviations, and Acronyms 85
Distribution List 86
iv
List of Figures
Figure 1. Layout of 1st pass ARL08M450—BPSK modulator and power amplifier. ....................5
Figure 2. Comparison of performance of 1st pass ARL02M450 vs. ARL08M450. ........................6
Figure 3. Cripps load line for a 50-mW PA (6x80 µm PHEMT with a 2.8-V supply). ..................7
Figure 4. Cripps load line for a 100-mW PA (10x96 µm PHEMT with a 2.8-V supply). ..............8
Figure 5. Schematic of 100-mW, 2.8-V power amplifier (MWO). .................................................9
Figure 6. S-parameter simulation of 100-mW, 2.8-V power amplifier (MWO). ............................9
Figure 7. Power performance simulation of 100-mW, 2.8-V power amplifier (400 and 450 MHz). .......................................................................................................................................10
Figure 8. Dynamic load line simulation of 100-mW, 2.8-V power amplifier (MWO). ................10
Figure 9. Layout of 100-mW, 2.8-V power amplifier (1.15x1.15 mm). .......................................11
Figure 10. Schematic of 50-mW narrowband, 2.8-V power amplifier (MWO). ...........................12
Figure 11. S-parameter simulation of 50-mW narrowband, 2.8-V power amplifier (MWO). ......13
Figure 12. Power performance simulation of 50-mW narrowband, 2.8-V power amplifier (400 and 450 MHz). ................................................................................................................13
Figure 13. Layout of the 50-mW, 2.8-V narrowband power amplifier (1.15x1.2 mm). ...............14
Figure 14. Schematic of the 50-mW broadband, 2.8-V power amplifier (MWO). .......................15
Figure 15. S-parameter simulation of the 50-mW broadband, 2.8-V power amplifier (MWO). ...16
Figure 16. Power performance simulation of the 50-mW broadband, 2.8-V power amplifier (400 and 450 MHz). ................................................................................................................16
Figure 17. Layout of the 50-mW broadband, 2.8-V power amplifier (0.8x1.1 mm). ...................17
Figure 18. Schematic of the 425-MHz low noise amplifier (MWO). ............................................18
Figure 19. S-parameter simulation of the 425-MHz low noise amplifier (MWO). .......................19
Figure 20. Noise figure simulation of the 425-MHz low noise amplifier (MWO)........................19
Figure 21. Layout of the 425-MHz low noise amplifier (1.6x1.0 mm). ........................................20
Figure 22. Schematic of the broadband TR switch (MWO). .........................................................21
Figure 23. S-parameter simulation of the broadband TR switch (MWO). ....................................22
Figure 24. Power performance simulation of the broadband TR switch (MWO). ........................22
Figure 25. Layout of the broadband TR switch (probe-able test cell―0.7x1.1 mm). ...................23
Figure 26. Insertion loss simulation of the dual band BPSK modulator (ADS). ...........................24
Figure 27. Phase difference simulation of the dual band BPSK modulator (ADS). ......................24
Figure 28. Layout of the dual band BPSK modulator (1.3x1.3 mm). ............................................25
Figure 29. Layout of ARL21M425 (2.41x2.41 mm die). ..............................................................27
v
Figure 30. Wire-bond diagram of ARL21M425 in a 4x4 mm QFN package. ...............................28
Figure 31. Layout of ARL22M425 (2.41x2.41 mm die). ..............................................................29
Figure 32. Wire-bond diagram of ARL22M425 in a 4x4 mm QFN package. ...............................29
Figure 33. Layout of ARL23M425 (2.41x2.41 mm die). ..............................................................30
Figure 34. Wire-bond diagram of ARL23M425 in a 4x4 mm QFN package. ...............................31
Figure 35. Layout of ARL24DB (2.41x2.41 mm die). ..................................................................32
Figure 36. Wire-bond diagram of ARL24DB in a 4x4 mm QFN package. ...................................32
Figure 37. Layout of the ARL25 test cells (2.41x2.41 mm die). ...................................................33
Figure 38. Schematic of ARL21M425, 100 mW and 2.8 V (ADS). .............................................34
Figure 39. S-parameter simulation of ARL21M425 transmit (ADS). ...........................................35
Figure 40. S-parameter simulation of ARL21M425 receive (ADS). .............................................35
Figure 41. Schematic of ARL22M425 50 mW 2.8 V (ADS). .......................................................37
Figure 42. S-parameter simulation of ARL22M425 transmit (ADS). ...........................................38
Figure 43. S-parameter simulation of ARL22M425 receive (ADS). .............................................38
Figure 44. Schematic of ARL23M425, 50 mW 2.8/3.6 V (ADS). ................................................40
Figure 45. S-parameter simulation of ARL23M425 transmit (ADS). ...........................................41
Figure 46. S-parameter simulation of ARL23M425 receive (ADS). .............................................41
Figure 47. Schematic of ARL24DB, 425/900 MHz and 2.8/3.6 V (ADS). ...................................43
Figure 48. S-parameter simulation of ARL24DB transmit (ADS). ...............................................44
Figure 49. S-parameter simulation of ARL24DB receive (ADS). .................................................44
Figure 50. Schematic of the 425-MHz low noise amplifier, ARL25 (ADS). ................................47
Figure 51. S-parameter simulation of the 425-MHz low noise amplifier, ARL25 (ADS). ...........48
Figure 52. Noise figure simulation of the 425-MHz low noise amplifier, ARL25 (ADS). ...........48
Figure 53. Sonnet EM layout simulation of the 425-MHz low noise amplifier, ARL25 (ADS). ..49
Figure 54. Sonnet (dotted) vs. ADS S-parameter simulation of the 425-MHz low noise amplifier, ARL25. ....................................................................................................................49
Figure 55. Sonnet (dotted) vs. ADS noise figure simulation of the 425-MHz low noise amplifier, ARL25. ....................................................................................................................50
Figure 56. Schematic of the modified (8 mA) broadband low noise amplifier, ARL24DB (ADS). ......................................................................................................................................51
Figure 57. Simulation of the modified (8 mA) broadband low noise amplifier, ARL24DB (ADS). ......................................................................................................................................51
Figure 58. NF simulation of the modified (8 mA) broadband low noise amplifier, ARL24DB (ADS). ......................................................................................................................................52
Figure 59. Schematic of the 425-MHz, 100-mW power amplifier, ARL25 (ADS). .....................52
Figure 60. Simulation of the 425-MHz, 100-mW power amplifier, ARL25 (ADS). ....................53
vi
Figure 61. Power simulation of the 425-MHz, 100-mW power amplifier, ARL25 (ADS). .........53
Figure 62. Sonnet EM layout simulation of the 425-MHz, 100-mW power amplifier, ARL25 (ADS). ......................................................................................................................................54
Figure 63. Sonnet (dotted) vs. ADS S-parameter simulation of the narrowband 100-mW power amplifier, ARL25. .................................................................................................54
Figure 64. Schematic of the 425-MHz, 50-mW power amplifier, ARL25 (ADS). .......................55
Figure 65. Simulation of the 425-MHz, 50-mW power amplifier, ARL25 (ADS). ......................56
Figure 66. Power simulation of the 425-MHz, 50-mW power amplifier, ARL25 (ADS). ............56
Figure 67. Sonnet EM layout simulation of the 425-MHz, 50-mW power amplifier, ARL25 (ADS). ......................................................................................................................................57
Figure 68. Sonnet (dotted) vs. ADS S-parameter simulation of the narrowband 50-mW power amplifier, ARL25 ..........................................................................................................57
Figure 69. Schematic of the TR switch, ARL25 (ADS) ................................................................58
Figure 70. Simulation of the TR switch, ARL25 (ADS) ...............................................................59
Figure 71. Schematic of the 425-MHz BPSK modulator (ADS). ..................................................60
Figure 72. Insertion loss simulation of the 425-MHz BPSK modulator (ADS). ...........................60
Figure 73. Phase simulation of the 425-MHz BPSK modulator (ADS). .......................................61
Figure 74. Sonnet EM layout simulation of the 425-MHz BPSK modulator. ...............................61
Figure 75. Sonnet (dotted) vs. ADS S-parameter simulation of the narrowband BPSK modulator. ................................................................................................................................62
Figure 76. Sonnet (dotted) vs. ADS phase simulation of narrowband BPSK modulator. .............62
Figure 77. Concept of integrating RFIC matching elements into an IC for size, weight, and power (SWAP). ........................................................................................................................63
Figure 78. Layout of ARL26DB (2.41x2.41 mm die). ..................................................................66
Figure 79. Wire-bond diagram of ARL26DB in a 4x4 mm QFN package. ...................................67
Figure 80. Layout of ARL27M425 (2.41x1.27 mm die). ..............................................................68
Figure 81. Wire-bond diagram of ARL27M425 in a 4x4 mm QFN package. ...............................68
Figure 82. Layout of ARL28M900 (2.41x1.27 mm die). ..............................................................69
Figure 83. Wire-bond diagram of ARL28M900 in a 4x4 mm QFN package. ...............................70
Figure 84. Layout of ARL29M425 (2.41x1.27 mm die). ..............................................................71
Figure 85. Wire-bond diagram of ARL29M425 in a 4x4 mm QFN package. ...............................71
Figure 86. Layout of ARL29M425S (2.41x1.27 mm die). ............................................................72
Figure 87. Wire-bond diagram of ARL29M425S in a 4x4 mm QFN package. ............................73
Figure 88. Schematic of the CC1100 matching circuit for 425/433 MHz using TriQuint elements. ..................................................................................................................................74
Figure 89. Schematic of the CC1100 matching circuit for 900 MHz using TriQuint elements. ...75
vii
Figure 90. Simulation of the CC1100 matching circuit for 425/433 MHz (TQS and ideal). ........75
Figure 91. Simulation of the CC1100 matching circuit for 900 MHz (TQS and ideal). ...............76
Figure 92. Schematic of the CC1000 matching circuit for 433 MHz using ideal elements. .........77
Figure 93. Schematic of the CC1000 matching circuit for 900 MHz using ideal elements. .........77
Figure 94. Split of the CC1000 matching circuit into separate transmit and receive circuits. ......78
Figure 95. Simulation of the CC1000 matching circuits for 315, 433, and 900 MHz (ideal). .....78
Figure 96. Simulation of the CC1000 split matching circuits for 433 MHz (TQS and ideal). .....79
Figure 97. Schematic of the CC1100 dual band matching circuit (ARL26DB). ...........................79
Figure 98. Layout plot of ARLTILE2 (10 designs–4.8x9.8 mm) ..................................................82
List of Tables
Table 1. Measured power performance of 1st pass ARL02M450. ...................................................6
Table 2. Measured power performance of 1st pass ARL08M450. ...................................................7
Table 3. Simulated performance of 425-MHz, 100-mW power amplifier. ...................................11
Table 4. Simulated performance of the 425-MHz, 50-mW power amplifier. ...............................14
Table 5. Simulated performance of broadband PA. .......................................................................17
Table 6. Simulated performance of the 425-MHz low noise amplifier. ........................................20
Table 7. Simulated performance of transmit chain for ARL21M425. ...........................................36
Table 8. Simulated performance of receive chain for ARL21M425. ............................................36
Table 9. Simulated performance of transmit chain for ARL22M425. ...........................................39
Table 10. Simulated performance of receive chain for ARL22M425. ..........................................39
Table 11. Simulated performance of transmit chain for ARL23M425 (2.7 V). ............................42
Table 12. Simulated performance of transmit chain for ARL23M425 (3.6 V). ............................42
Table 13. Simulated performance of receive chain for ARL23M425. ..........................................42
Table 14. Simulated performance of transmit chain for ARL24DB (425 M/2.7 V). ....................45
Table 15. Simulated performance of transmit chain for ARL24DB (425 M/3.6 V). ....................45
Table 16. Simulated performance of transmit chain for ARL24DB (900 M/2.7 V). ....................45
Table 17. Simulated performance of transmit chain for ARL24DB (900 M/3.6 V). ....................46
Table 18. Simulated performance of receive chain for ARL24DB. ..............................................46
Table 19. Map of design layout within ARLTILE2. .....................................................................83
viii
Acknowledgments
I would especially like to acknowledge and thank Romeo Del Rosario, Paul Amirtharaj, and Eric Adler for ensuring that the design tapeout submission was on time and successful.
1
1. Introduction
Increased transmission range in low-power radio frequency (RF) applications is a common desire within the Army. The RF integrated circuit (RFIC) booster chip is intended to increase range between RF nodes for low-power wireless applications. The booster concept uses the high RF performance advantages of the gallium arsenide (GaAs) process to enhance the capabilities of systems using commercial RFIC transceivers. It can be inserted easily into systems to increase transmit power, receiver sensitivity, and efficiency or battery utilization. In this 2nd pass optimized booster design, lessons learned from a 1st pass design were used to optimize a design for a targeted system performance and battery voltage.
2. Lessons Learned from 1st Pass Design
The first booster chip integrated circuit (IC) consisted of RF circuits to enhance the performance of commercial RFIC transceiver circuits, which could be used in a variety of applications. Of particular interest was a wireless tag at ultra-high frequency (UHF) frequencies using a commercial RFIC transceiver. Lower UHF frequencies tend to propagate further in urban or canopy/forest environments, while higher frequencies provide advantages of smaller antennas and typically higher data rates. Since the multi-project low-cost prototype service provided by TriQuint Semiconductor allows one to customize a 5x10 mm quarter tile into multiple IC designs, many variations of an RF front-end booster design were fabricated for the 1st pass at frequencies of interest at 450, 900, and 2400 MHz with other circuit design variations. This is often called the “shotgun” approach by testing multiple design variations.
Due to the high performance of TriQuint’s 0.5 µm TQPED pseudomorphic high electron mobility transistor (PHEMT) process, low-cost prototyping service, high reliability process, quality control, excellent reputation, and available design models for Agilent’s Advanced Design System (ADS) and Applied Wave Research’s (AWR) Microwave Office (MWO), the TQPED process was used for the 1st pass and 2nd pass booster IC designs. While the 1st pass designs tried a number of circuits at different frequencies, the 2nd pass designs were mostly targeted at operation between 400 to 450 MHz, so one obvious change was to retune the original circuit designs from a center frequency of 450 MHz to the slightly lower 425 MHz. Another design variation targeted a dual band operation around 425 or 900 MHz.
A broadband power amplifier design in the 1st pass was very efficient (~50% power added efficiency [PAE]), had good gain, and had a good output power of 50 mW (17 dBm) with a 3.6-V battery supply. While a 3.6-V supply is a good choice when using a lithium ion rechargeable battery, nominally 3.9 V with regulation to 3.6 V, another popular battery is a non-
2
rechargeable 3.0-V battery, such as a small coin cell, which can be regulated to 2.7 or 2.8 V. A current mirror bias of the broadband power amplifier allowed good gain over a large supply voltage variation of 2.0 to 5.0 V with good efficiency, while the output power and DC power consumption varies proportionally with the supply voltage. There is a concern that the current mirror bias may slightly reduce the efficiency of the power amplifier over the simple but less robust resistor divider DC bias. Unfortunately, there were not enough data and design variations in the 1st pass to isolate an optimal DC bias choice between these two approaches relative to efficiency. While the 1st pass broadband amplifier design was a good robust stable design, the 2nd pass goal was to design an amplifier to have 50 or 100 mW of power (17, 20 dBm) optimized for a 2.8-V supply voltage. The power amplifiers in the 2nd pass designs were targeted for a narrowband 50-mW design at 2.8 V, a 100-mW design at 2.8 V, and an optimized variation of the 1st pass broadband design that would provide 50 mW for a 3.6-V supply yet would also operate with a 2.8-V supply at a lower output power but good efficiency.
There were two main variations of a low noise amplifier design in the 1st pass. One was a broadband low noise design with moderate gain but consumed about 16 mA of bias current compared to a narrowband approach with about 5 dB less gain that consumed a mere 3 mA of bias current. The low noise amplifiers in the 2nd pass designs were targeted for an ~2-dB noise figure with a narrowband 425-MHz design with 12-dB gain at 3 mA of current consumption and a second broadband amplifier modified from the 1st pass for 425 and 900 MHz operation with 15-dB gain at a moderate current bias of 8 to 10 mA. Unlike the power amplifiers whose efficiencies peak for a particular supply voltage, the performance of both low noise amplifiers varies little over a supply voltage of 2.0 to 5.0 V. The broadband low noise amplifier was used in a dual band 425- and 900-MHz booster IC design, while the narrowband low noise amplifier was used for the 425-MHz booster IC designs.
The positive voltage controlled binary phase shift key (BPSK) modulator from the 1st pass design was reused for the 2nd pass booster IC designs. A minor change was made to retune the filters in the narrowband switched design from a center frequency of 450 to 425 MHz to better suit a desired frequency of operation from 400 to 450 MHz.
The transmit/receive (TR) switch from the 1st pass design had more insertion loss than desired and was modified slightly to improve its operation at the slightly lower UHF frequency around 425 MHz. One reason for the higher than expected insertion loss in the original design was due to too much use of the higher loss metal0 layer for interconnects. The original simulations defaulted to use a single thick metal for all metal types, which did not correctly simulate the higher loss of the thin metal0 layer. While the thin metal0 layer is required for connecting the circuits, the 2nd pass TR switch redesign minimized the use of metal0 and transitioned to the lower loss thick metal1 or thick metal2 layers for less interconnect loss. Also, the PHEMT switches in the TR switch used negative depletion mode PHEMTs in order to handle the relatively high 100-mW power level from the power amplifier. While it would be easier to design a positive voltage controlled TR switch by replacing the depletion mode PHEMTs with
3
the enhancement mode PHEMTs, this swap only makes sense for the switches in the BPSK modulator circuit where the power handling requirements are much lower. To use positive control voltages for the TR switch with depletion mode PHEMTs requires several large capacitors and a positive reference voltage input. The capacitors need to be large enough to minimize the insertion loss at the lower 425-MHz operating frequency but small enough to fit in the available circuit size. Insertion loss for the 2nd pass TR switch is expected to achieve the 0.5-dB goal over a broad frequency range above 400 MHz, with approximately 0.25 dB of loss due to the PHEMT switches and 0.25 dB of loss due to the tradeoff of capacitor size.
One of the design variations (ARL08M450) from the 1st pass consisting of a BPSK modulator and power amplifier at 450 MHz and positive voltage control was integrated into an existing tag by a system integrator. Unfortunately, testing was still incomplete by the fabrication date for the 2nd pass designs. The status of the system level test is that data appears to be modulating OK, but the output power level is much lower than expected (~10 dB). It is not clear if there is a board problem, a software problem such as an incorrect drive level programmed into the RFIC, a matching problem at the input or output of the booster IC on the system board, or something else. Testing of this particular design was successful at the die level and the board level. The initial tests from a single tag that includes this booster IC do not correlate with previous tests of the design.
Since wire bond parasitics were not significant in the 1st pass designs at the lower UHF frequencies, they were not used for “tweaking” the 2nd pass designs for optimizing performance. Wire bond diagrams for all the designs in a 4x4 mm quad flat no lead (QFN) packages were created before sending the designs out to fabrication to verify the suitability of the input/output/DC pad layouts. Simulations of the overall designs including the TR switch were performed to verify amplifier performance in the top level designs as well as the individual performance as a standalone amplifier.
3. Component Designs and Simulation Results
The power amplifier for the 2nd pass designs were to be optimized for efficiency and a center frequency of 425 MHz, slightly retuned from the 450 MHz of the 1st pass designs. Also, the power output goals were 50 and 100 mW at 2.8 V, which exceed the original 1st pass design. Secondary goals were amplifiers of similar performance for 3.6-V operation. The original power amplifier was a stable broadband design, but a narrowband design optimized for the lower 2.8-V supply voltage with good output power was desired for the 2nd pass design.
An initial reanalysis of the 1st pass power amplifier was performed to look for ways to optimize the efficiency. There was a slight cost in efficiency with the feedback design due to 1 mA of current through a stabilizing resistor on the gate/input of the PHEMT. For the current mirror
4
bias, a small 5-µm PHEMT was used for a current reference. This was the smallest device size allowed by the ADS TriQuint library for simulation, but the foundry rules allow devices as small as 2 µm. A change in the current reference from 5 to 2 µm would reduce its current consumption from 1 to 0.4 mA. Reducing the parasitic DC current consumption of the current mirror bias and the stabilizing resistor should slightly improve the efficiency of the amplifier relative to the nominal 30-mA DC consumption of the 6- x 80-µm PHEMT of the amplifier.
There was a workaround to force the simulations to work with the small 2-µm PHEMT device, which had to be laid out manually using the Integrated Circuit Editor (ICED) computer-aided design (CAD) program. A similar current mirror circuit was designed by the author, and then fabricated, and tested as part of the Johns Hopkins University Monolithic Microwave Integrated Circuit (MMIC) Design Course in fall 2009. The simulations showed a minimal 0.4 mA of bias used by the current mirror. Unfortunately, there is the possibility that the power amplifier efficiency may be lower due to using the current mirror bias rather than the resistor divider bias. Reducing the DC 1-mA current in the stabilizing resistor requires a large capacitor to block the DC current while maintaining a low impedance RF connection. The cost is the area needed by the capacitor with minimal impact on the RF performance in order to increase the efficiency a few percent.
The first few designs in the 1st pass used negative control voltages for the depletion mode (dmode) PHEMT switches. Later, variations of the first designs were created using positive control voltages to drive enhancement mode (emode) PHEMT switches that replaced the dmode PHEMT switches. Figure 1 shows a plot of the paired designs ARL02M450 and ARL08M450, which consist of a BPSK modulator and a power amplifier for the 450-MHz band. The subtle differences between the two designs are the positive emode switches in the BPSK modulator of ARL08M450 and the current mirror robust DC bias in ARL08M450 versus a simple resistor divider gate bias in ARL02M450. The performance of the nearly identical second design was slightly less than the original design. Simulations along with measurements of some test circuits indicated that the emode PHEMT switches in the second design should not compress at the signal levels at the input of the modulator. This would seem to indicate that the lower performance was partly due to the current mirror. The output power and efficiency of the original resistor gate bias approached 50 mW and 50% PAE, while the current mirror bias peaked near 40 mW with less than 40% PAE. Figure 2 shows a plot of output power in mW (blue), PAE in percent (green), and gain in dB (red) versus current IDS in mA. Normally, performance plots are shown versus the input power but these were plotted versus current bias. Note that the resistor bias design (ARL02) acts like a typical Class AB amplifier, where the DC bias increases with increasing input power and similarly increased output power. The current mirror bias seems to saturate, limiting the output power and efficiency. However, note that the current range is narrower for the current mirror design. It is performing its expected function of biasing the amplifier at a controlled current. Tables 1 and 2 summarize the performance measurements for the two designs in a spreadsheet format. Some of the lower efficiency of the 2nd design can be
5
explained by the additional parasitic 1-mA current mirror, which will be reduced to 0.4 mA in the 2nd pass designs. Additional simulations of the amplifier using the current mirror bias were performed using TriQuint’s TOM3 model as well as the more advanced TOM4 model but gave no indication of a performance difference between a current mirror bias relative to a resistor divider gate bias. Given this unknown, the 2nd pass power amplifiers use the current mirror bias, but have a test pad connected through a large resistor to monitor the DC gate bias voltage, as well as providing a means to override the current mirror bias. One goal in testing the power amplifiers in the 2nd pass will be to determine which DC bias scheme provides the more efficient amplifier performance.
Figure 1. Layout of 1st pass ARL08M450—BPSK modulator and power amplifier.
6
0
5
10
15
20
25
30
35
40
45
50
26 27 28 29 30 31 32 33 34 35 36 37
IDS (mA)
ARL2 8 Meas 3/15/10Emode 450MHz 2.8V
Pout2A
Gain2A
PAE2A
Pout8A
Gain8A
PAE8A
Figure 2. Comparison of performance of 1st pass ARL02M450 vs. ARL08M450.
Table 1. Measured power performance of 1st pass ARL02M450.
BPSK + Power Amps--450 MHz at 2.8 V 0.20 - 0.25 dB loss on DUT QFN3x3 Thru 400-500M Meas New Boards measured at 450 MHz 3/15/2010450 MHz PKG#2A PA450MHz Emode ARL #8 Tile 1 TQPED 2.8V ; 27 mAPin(SG) Pout(PS) Pin(corr) Pout(corr)Gain I1(2.8V) PDC(mw) Pout(mw) Drn Eff PAE
-10.0 4.96 -10.10 5.06 15.16 26.7 74.7 3.21 4.3 4.2-8.0 6.92 -8.10 7.02 15.12 26.9 75.2 5.04 6.7 6.5-6.0 8.84 -6.10 8.94 15.04 27.2 76.2 7.83 10.3 10.0-4.0 10.66 -4.10 10.76 14.86 27.8 77.7 11.91 15.3 14.8-2.0 12.45 -2.10 12.55 14.65 28.8 80.7 17.99 22.3 21.50.0 14.11 -0.10 14.21 14.31 30.7 86.0 26.36 30.6 29.51.0 14.83 0.90 14.93 14.03 32.0 89.6 31.12 34.7 33.42.0 15.48 1.90 15.58 13.68 33.4 93.4 36.14 38.7 37.03.0 15.95 2.90 16.05 13.15 34.6 96.7 40.27 41.6 39.64.0 16.35 3.90 16.45 12.55 35.6 99.6 44.16 44.3 41.95.0 16.64 4.90 16.74 11.84 36.2 101.4 47.21 46.5 43.56.0 16.86 5.90 16.96 11.06 36.5 102.2 49.66 48.6 44.8
7
Table 2. Measured power performance of 1st pass ARL08M450.
450 MHz PKG#8A PA450MHz Emode ARL #8 Tile 1 TQPED 2.8V ; 26 mAPin(SG) Pout(PS) Pin(corr) Pout(corr)Gain I1(2.8V) PDC(mw) Pout(mw) Drn Eff PAE
-10.0 4.24 -10.10 4.34 14.44 26.4 73.8 2.72 3.7 3.5-8.0 6.18 -8.10 6.28 14.38 26.5 74.2 4.25 5.7 5.5-6.0 8.10 -6.10 8.20 14.30 26.8 74.9 6.61 8.8 8.5-4.0 9.91 -4.10 10.01 14.11 27.2 76.2 10.02 13.2 12.6-2.0 11.64 -2.10 11.74 13.84 27.9 78.1 14.93 19.1 18.30.0 13.14 -0.10 13.24 13.34 28.9 80.9 21.09 26.1 24.91.0 13.77 0.90 13.87 12.97 29.4 82.4 24.38 29.6 28.12.0 14.35 1.90 14.45 12.55 30.0 84.0 27.86 33.2 31.33.0 14.82 2.90 14.92 12.02 30.5 85.3 31.05 36.4 34.14.0 15.2 3.90 15.3 11.40 30.8 86.2 33.88 39.3 36.55.0 15.46 4.90 15.56 10.66 30.8 86.3 35.97 41.7 38.16.0 15.63 5.90 15.73 9.83 30.6 85.7 37.41 43.7 39.1
Using the Cripps method (from Steve Cripps), the first step in the amplifier design was to pick a PHEMT device size for 50 mW and 100 mW of RF output power biased with a 2.8-V supply. The 1st pass amplifier used a 0.5-mm PHEMT (6x80 µm), which should provide 50 mW for a 2.8-V supply (figure 3). Doubling the amplifier PHEMT size (10x96 µm) should increase the output power to the desired 100-mW goal for a 2.8-V supply (figure 4).
1 2 3 4 5 6 70 8
50
100
0
150
VGS=0.300VGS=0.350VGS=0.400VGS=0.450VGS=0.500VGS=0.550VGS=0.600VGS=0.650VGS=0.700VGS=0.750VGS=0.800VGS=0.850VGS=0.900
VGS=0.950VGS=1.000
VDS
IDS
.i, m
A
m1
m2
m1VDS=IDS.i=0.043VGS=0.650000
2.800m2VDS=IDS.i=0.091VGS=0.850000
0.500
Figure 3. Cripps load line for a 50-mW PA (6x80 µm PHEMT with a 2.8-V supply).
8
1 2 3 4 5 6 70 8
100
200
0
300
VGS=0.300VGS=0.350VGS=0.400VGS=0.450VGS=0.500VGS=0.550VGS=0.600VGS=0.650VGS=0.700VGS=0.750VGS=0.800VGS=0.850VGS=0.900VGS=0.950VGS=1.000
VDS
IDS
.i, m
A
m1
m2m1VDS=IDS.i=0.085VGS=0.650000
2.800
m2VDS=IDS.i=0.182VGS=0.850000
0.500
Figure 4. Cripps load line for a 100-mW PA (10x96 µm PHEMT with a 2.8-V supply).
AWR’s MWO was used to simulate the initial narrowband amplifier designs, with some additional simulations using Agilent’s ADS for comparison. Sonnet software was used to verify the layout using an electromagnetic (EM) simulation of the individual circuit layouts combined with transistor models of the PHEMTs. Figure 5 shows the schematic of the 100-mW (2.8-V), 425-MHz power amplifier from MWO. Simulations show that the amplifier provides 100 mW of output power with greater than 50% PAE at 400 and 450 MHz with 0-dBm input (1-mW) at 2.8 V and a small signal IDS of 78 mA. The amplifier has a lot of gain at low frequencies, in fact, the original stabilizing resistors were changed slightly after the design review to provide some additional margin, dropping the gain slightly (figure 6). Power performance was simulated and was shown to meet the goals of 100 mW and 50% PAE, as shown in figure 7. Figure 8 shows the dynamic load line of the amplifier PHEMT. Table 3 summarizes the 100-mW, 425-MHz power amplifier performance. A layout of the amplifier as a probable test cell included in the tile is shown in figure 9. There was an additional small pad and resistor added after the design review that is not shown in this plot that allows monitoring of the gate bias and also a means to bypass the current mirror circuit during testing.
9
Figure 5. Schematic of 100-mW, 2.8-V power amplifier (MWO).
Figure 6. S-parameter simulation of 100-mW, 2.8-V power amplifier (MWO).
10
Figure 7. Power performance simulation of 100-mW, 2.8-V power amplifier (400 and 450 MHz).
Figure 8. Dynamic load line simulation of 100-mW, 2.8-V power amplifier (MWO).
11
Table 3. Simulated performance of 425-MHz, 100-mW power amplifier.
Power (dBm) Pout 0.45 GHz PAE 0.45 GHz Pout 0.40 GHz PAE 0.4 GHz –10 14.3682 12.1759 15.0634 14.2445 –9 15.2921 15.0109 15.9557 17.4261 –8 16.1622 18.2815 16.7791 21.0045 –7 16.9601 21.9035 17.508 24.8799 –6 17.6728 25.7596 18.1425 28.8796 –5 18.2811 29.879 18.6907 32.9027 –4 18.789 34.3215 19.1363 37.3061 –3 19.2284 38.8689 19.4856 42.1762 –2 19.6223 42.9513 19.7429 46.7506 –1 19.9386 46.9051 19.8992 49.6577 0 20.1529 50.1867 19.9999 51.5825 1 20.2823 52.5203 20.0418 53.2254 2 20.3495 53.8208 20.0678 54.1865
Figure 9. Layout of 100-mW, 2.8-V power amplifier (1.15x1.15 mm).
Another narrowband power amplifier was also designed with a goal of 50-mW output power and 50% PAE using a 2.8-V supply for half the DC power consumption of the 100 mW PA. Figure 10 shows the schematic of the narrowband 50-mW (2.8-V), 425-MHz power amplifier from MWO. Simulations show that the amplifier provides more than 50 mW of output power with greater than 50% PAE at 400 and 450 MHz with 0-dBm input (1-mW) at 2.8 V and a small signal IDS of 39 mA. An S-parameter simulation of the design shows good gain and input return
12
loss (figure 11). Power performance was simulated and was shown to meet the goals of 50 mW and 50% PAE as shown in figure 12. Table 4 summarizes the 50-mW, 425-MHz power amplifier performance. A layout of the amplifier as a probe-able test cell included in the tile is shown in figure 13. There was an additional small pad and resistor added after the design review that is not shown in this plot that allows monitoring of the gate bias and also a means to bypass the current mirror circuit during testing.
Figure 10. Schematic of 50-mW narrowband, 2.8-V power amplifier (MWO).
13
Frequency (GHz)
PA_S2P_50mw_nb
-25
-20
-15
-10
-5
0
5
10
15
20
25
30
p54p53p52p51p50p49p48p47p46p45p44p43p42p41p40p39p38p37
p36p35p34p33p32p31p30p29p28p27p26p25p24p23p22p21p20p19
p18p17p16p15p14p13p12p11p10p9p8p7p6p5p4p3p2p1
0.4004 GHz-20.32 dB
0.418 GHz23.29 dB DB(|S(2,1)|)
PA450m_50mw_v28_nb
DB(|S(1,1)|)PA450m_50mw_v28_nb
DB(|S(2,2)|)PA450m_50mw_v28_nb
p1: Pwr = -15 dBm p2: Pwr = -14 dBm
p3: Pwr = -13 dBm p4: Pwr = -12 dBm
p5: Pwr = -11 dBm p6: Pwr = -10 dBm
p7: Pwr = -9 dBm p8: Pwr = -8 dBm
p9: Pwr = -7 dBm p10: Pwr = -6 dBm
p11: Pwr = -5 dBm p12: Pwr = -4 dBm
p13: Pwr = -3 dBm p14: Pwr = -2 dBm
p15: Pwr = -1 dBm p16: Pwr = 0 dBm
p17: Pwr = 1 dBm p18: Pwr = 2 dBm
p19: Pwr = -15 dBm p20: Pwr = -14 dBm
p21: Pwr = -13 dBm p22: Pwr = -12 dBm
p23: Pwr = -11 dBm p24: Pwr = -10 dBm
p25: Pwr = -9 dBm p26: Pwr = -8 dBm
p27: Pwr = -7 dBm p28: Pwr = -6 dBm
p29: Pwr = -5 dBm p30: Pwr = -4 dBm
p31: Pwr = -3 dBm p32: Pwr = -2 dBm
p33: Pwr = -1 dBm p34: Pwr = 0 dBm
p35: Pwr = 1 dBm p36: Pwr = 2 dBm
p37: Pwr = -15 dBm p38: Pwr = -14 dBm
p39: Pwr = -13 dBm p40: Pwr = -12 dBm
p41: Pwr = -11 dBm p42: Pwr = -10 dBm
p43: Pwr = -9 dBm p44: Pwr = -8 dBm
p45: Pwr = -7 dBm p46: Pwr = -6 dBm
p47: Pwr = -5 dBm p48: Pwr = -4 dBm
p49: Pwr = -3 dBm p50: Pwr = -2 dBm
p51: Pwr = -1 dBm p52: Pwr = 0 dBm
p53: Pwr = 1 dBm p54: Pwr = 2 dBm
Figure 11. S-parameter simulation of 50-mW narrowband, 2.8-V power amplifier (MWO).
Figure 12. Power performance simulation of 50-mW narrowband, 2.8-V power amplifier (400 and 450 MHz).
14
Table 4. Simulated performance of the 425-MHz, 50-mW power amplifier.
Power (dBm) Pout 0.45 GHz PAE 0.45 GHz Pout 0.40 GHz PAE 0.4 GHz –10 12.908 16.5759 13.2341 17.7262 –9 13.8551 20.3447 14.1645 21.6288 –8 14.7739 24.7309 15.0595 26.0825 –7 15.6435 29.6207 15.8954 30.9007 –6 16.4307 34.6991 16.6432 35.7074 –5 17.111 39.4916 17.2594 39.9942 –4 17.6743 43.607 17.7144 43.4018 –3 18.0763 46.8375 18.0745 46.1309 –2 18.3511 49.3081 18.3943 48.4815 –1 18.5574 51.2367 18.6227 50.3712 0 18.7138 52.8591 18.7428 51.6878 1 18.7881 54.2084 18.7925 52.6691 2 18.7931 54.93 18.7907 53.5416
Figure 13. Layout of the 50-mW, 2.8-V narrowband power amplifier (1.15x1.2 mm).
The power amplifier from the 1st pass was reused to provide a broadband design as well as to provide a 50-mW design for operation with a 3.6-V battery. This design was already shown to have excellent gain and efficiency but some minor improvements were made to increase the efficiency. The shunt stabilizing resistor was DC blocked with a large capacitor to improve the efficiency a few percent by removing the 1-mA parasitic current flow. Also, the current mirror
15
was reduced from a 5- to a 2-µm device reducing that parasitic current from 1 to 0.4 mA, which should add a couple of percent to the efficiency. Figure 14 shows the schematic of the broadband 50-mW (2.8/3.6-V) power amplifier from MWO. Simulations show that the amplifier provides almost 50 mW of output power with about 47% PAE at 400 and 450 MHz with 0-dBm input (1-mW) at 2.8 V. An S-parameter simulation of the design shows good gain over a broad frequency range (figure 15). Power performance was simulated and was shown to be close to the goals of 50 mW and 50% PAE as shown in figure 16. Table 5 summarizes the broadband PA performance. A layout of the amplifier is shown in figure 17, but was not included as a separate test cell since it is only slightly modified from the successful 1st pass design. There was an additional small pad and resistor added after the design review in the full chip design that allows monitoring of the gate bias and also a means to bypass the current mirror circuit during testing.
Figure 14. Schematic of the 50-mW broadband, 2.8-V power amplifier (MWO).
16
Figure 15. S-parameter simulation of the 50-mW broadband, 2.8-V power amplifier (MWO).
-15 -10 -5 0 4Power (dBm)
PAE_50mw
0
5
10
15
20
25
30
35
40
45
50
55
-0.0494 dBm16.8 dBm
0.005423 dBm47.57
DB(|Pcomp(PORT_2,1)|)[8,X] (dBm)pa450m_50mw_v28
PAE(PORT_1,PORT_2)[8,X]pa450m_50mw_v28
DB(|Pcomp(PORT_2,1)|)[7,X] (dBm)pa450m_50mw_v28
PAE(PORT_1,PORT_2)[7,X]pa450m_50mw_v28
p1: Freq = 0.4 GHz
p2: Freq = 0.45 GHz
p3: Freq = 0.4 GHz
p4: Freq = 0.45 GHz
Figure 16. Power performance simulation of the 50-mW broadband, 2.8-V power amplifier (400 and 450 MHz).
17
Table 5. Simulated performance of broadband PA.
Power (dBm) Pout 0.45 GHz PAE 0.45 GHz Pout 0.40 GHz PAE 0.4 GHz –10 9.47886 14.2564 9.45505 14.1451 –9 10.366 17.0402 10.3397 16.8885 –8 11.2249 20.1022 11.1965 19.9017 –7 12.0566 23.3937 12.027 23.1408 –6 12.8676 26.8906 12.8376 26.5813 –5 13.6657 30.6024 13.6359 30.2382 –4 14.4509 34.5239 14.422 34.1011 –3 15.2038 38.5338 15.1754 38.0493 –2 15.8872 42.3231 15.8435 41.7069 –1 16.4552 45.4821 16.4097 44.8043 0 16.8231 47.5654 16.8025 46.9706 1 17.05 48.9521 17.0187 48.321 2 17.1761 49.9858 17.116 49.2891
Figure 17. Layout of the 50-mW broadband, 2.8-V power amplifier (0.8x1.1 mm).
A low DC current consumption narrowband low noise amplifier was designed similar to the 1st pass design with a goal of a 2-dB noise figure and current consumption around 3 mA using the smaller 0.4-mA (2-µm PHEMT) current mirror for a very robust DC bias for supplies of 2 to 5 V. Figure 18 shows the schematic of the narrowband 425-MHz low noise amplifier from MWO. Simulations show that the amplifier provides about 11 dB of gain with an IDS of 3.2 mA
18
at 2.8 V (figure 19). Noise figure simulations show that the design should meet the goal of 2 dB as shown in figure 20. Table 6 summarizes the low noise amplifier performance. A layout of the low noise amplifier as a probe-able test cell is shown in figure 21. For the low noise amplifier, there was no need to add an extra test pad and resistor to bypass the current mirror circuit since the concern was only for large signal operation of the power amplifier. Table 6 summarizes the low noise amplifier performance.
Figure 18. Schematic of the 425-MHz low noise amplifier (MWO).
19
Figure 19. S-parameter simulation of the 425-MHz low noise amplifier (MWO).
Figure 20. Noise figure simulation of the 425-MHz low noise amplifier (MWO).
20
Table 6. Simulated performance of the 425-MHz low noise amplifier.
Frequency (GHz) DB(NF()) DB(|S(2,1)|)
0.3 3.69481 4.5861 0.35 2.32686 10.5096
0.4 1.83241 11.9386
0.45 1.68046 11.3902
0.5 1.65716 10.4841
0.55 1.68343 9.64657 0.6 1.72521 8.95129
0.65 1.76856 8.38857
0.7 1.80916 7.93259
0.75 1.84617 7.55922 0.8 1.87904 7.2503
0.85 1.90819 6.99148
0.9 1.9341 6.77191
0.95 1.95741 6.58272
1 1.97845 6.41824
Figure 21. Layout of the 425-MHz low noise amplifier (1.6x1.0 mm).
A broadband positive voltage control TR switch from the 1st pass was redesigned to reduce the higher than desired insertion loss with a goal of 0.5 dB. First, the PHEMT switches were doubled in size to reduce the inherent series resistance of the switches. Some test switch devices from the Johns Hopkins University MMIC Design Fall 2009 student designs showed that a switch using devices similar in size to those in ARL’s 1st pass had an insertion loss of 0.5 dB and using switches twice as big reduced the insertion loss to less than 0.2 dB as measured. Another source
21
of loss in the ARL 1st pass designs was too much use of the lossy thin metal0 layer, so care was taken in the 2nd pass design to transition from metal0 to the thicker metal1 or metal2 layers for connections. In order to make the dmode PHEMTs take positive voltage control signals, large capacitors are needed. The size of the capacitors is a tradeoff between size and insertion loss. Overall the insertion loss is expected to be close to 0.5 dB, about half due to the capacitor size tradeoff and half due to the lossy PHEMT switches. Figure 22 shows the schematic of the broadband TR switch from MWO. Simulations show that the insertion loss is around 0.5 dB (figure 23). Large signal power simulations show that the design should be able to handle 100 mW of power without compressing (figure 24). A layout of the broadband TR switch as a probe-able test cell is shown in figure 25. The two way switch either goes to a 50-ohm resistor to ground to measure isolation, or to a through path to measure insertion loss. A reference voltage must be supplied (typically 2.5 V) to use a positive control voltage for the dmode PHEMTs in the TR switch.
Figure 22. Schematic of the broadband TR switch (MWO).
22
Figure 23. S-parameter simulation of the broadband TR switch (MWO).
Figure 24. Power performance simulation of the broadband TR switch (MWO).
23
Figure 25. Layout of the broadband TR switch (probe-able test cell―0.7x1.1 mm).
The BPSK modulator from the 1st pass worked very well and was only slightly retuned for a center frequency of 425 MHz versus the previous 450 MHz. A second broader band design was created using 5 elements in the high pass and low pass networks that are switched to create the two modulation states. To minimize the number of inductors, the low pass network was modified from three series inductors with two shunt capacitors to two series inductors with three shunt capacitors to save space. Simulations of the dual band or broadband BPSK modulator are shown in figures 26 and 27. The layout of the dual band BPSK modulator included in a full IC design is shown in figure 28.
A broadband low noise amplifier was needed for the dual band design, so the high current high gain broadband low noise amplifier design from the 1st pass was modified for use with similar performance but less current consumption than the original. The current mirror was modified to reduce the current consumption from 16 mA to about 8 to 10 mA with a slight drop in gain to about 15 dB at 425 and 900 MHz. Since this design is so similar to the previous design only its simulation in the top level full IC designs will be shown.
24
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.00.0 1.1
-20
-15
-10
-5
-25
0
freq, GHz
dB(S
(2,1
))
m1m11
dB(S
(1,1
))
m6
dB(S
(2,2
))
m1freq=dB(S(2,1))=-0.850Vbias=0.000000
425.0MHzm11freq=dB(S(2,1))=-1.335Vbias=3.000000
425.0MHzm6freq=dB(S(1,1))=-23.491Vbias=0.000000
425.0MHz
Figure 26. Insertion loss simulation of the dual band BPSK modulator (ADS).
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.00.1 1.1
-180
-135
-90
-45
0
45
90
135
-225
180
freq, GHz
phas
e(S
(2,1
))
m2
m3
m9
m10
phas
e(S
(2,1
))[0
,::]
-pha
se(S
(2,1
))[1
,::]
m7 m8
m2freq=phase(S(2,1))=117.937Vbias=3.000000
425.0MHz
m3freq=phase(S(2,1))=-61.199Vbias=0.000000
425.0MHzm9freq=phase(S(2,1))=44.809Vbias=3.000000
900.0MHz
m10freq=phase(S(2,1))=-135.166Vbias=0.000000
900.0MHz
m7freq=phase(S(2,1))[0,::]-phase(S(2,1))[1,::]=-179.136
425.0MHz
m8freq=phase(S(2,1))[0,::]-phase(S(2,1))[1,::]=-179.97
900.0MHz
Figure 27. Phase difference simulation of the dual band BPSK modulator (ADS).
25
Figure 28. Layout of the dual band BPSK modulator (1.3x1.3 mm).
4. Active GaAs IC Designs
The focus of the 2nd pass was predominantly the 400 to 450 MHz frequency band, but there was also a desire to have a broadband or dual band application since the commercial RFICs are typically capable below 1 GHz with the proper external matching circuits, which then tend to narrowband the system design. As with the 1st pass tile, the designs were part of a multi-die fabrication so several active design variations were included to meet various design goals. Since the band of interest was lower in frequency and since inductors and capacitors grow in size as the frequency is lowered, the larger 4x4 mm QFN package was targeted for packaging the die. A 95x95 mil (2.41x2.41 mm) size was chosen. This could allow up to eight designs of this size to be included in a single quarter tile prototype fabrication. The previous 1st pass designs tried to maximize the number of design variations over many frequency bands and targeted either 3x3 or 4x4 mm QFN packages, such that only the BPSK modulator and power amplifier could be squeezed into the smaller die size at the lower frequency. The larger die size allowed for increasing the size of the lumped elements to retune designs to a slightly lower frequency
26
(425 MHz versus 450 MHz) and allowed for the inclusion of the modulator, power amplifier, TR switch, and low noise amplifier in a single die. There are four active design variations targeting 50 mW at 2.8 V, 100 mW at 2.8 V, at least 50 mW for either 2.8 V or 3.6 V, and a dual band design at 425 or 900 MHz using the broadband power amplifier design with a newly designed broadband BPSK modulator design. The fifth active design is a test circuit with the narrowband 50-mW power amplifier, the narrowband 100-mW power amplifier, the narrowband 3-mA low noise amplifier, and the broadband TR switch design for individual probe testing. Later, passive GaAs IC designs to improve the system level performance by integrating the external RFIC matching circuits into a single IC are discussed. Following is a brief description of each active design in the 2nd pass ARL tile following a similar naming convention to the 1st pass designs. The numbers start with 21 and go to 29 with some additional letters to designate frequency or other variations.
The following are descriptions of the active designs:
• ARL21M425―This 425-MHz design contains a BPSK modulator, a 100-mW power amplifier for 2.8 V, a narrowband low DC power consumption low noise amplifier, and a TR switch using positive voltage control inputs with negative threshold depletion PHEMTs. It is a 95x95 mil die.
• ARL22M425―This 425-MHz design contains a BPSK modulator, a 50-mW power amplifier for 2.8 V, a narrowband low DC power consumption low noise amplifier, and a TR switch using positive voltage control inputs with negative threshold depletion PHEMTs. It is a 95x95 mil die.
• ARL23M425―This 425-MHz design contains a BPSK modulator, a 50/75-mW power amplifier for 2.8/3.6 V, a narrowband low DC power consumption low noise amplifier, and a TR switch using positive voltage control inputs with negative threshold depletion PHEMTs. This design is intended to operate over a broader range of battery voltages. It is a 95x95 mil die.
• ARL24DB―This dual band design contains a BPSK modulator for 425 or 900 MHz, the broadband 50/75-mW power amplifier for 2.8/3.6 V, a broadband moderate DC power consumption low noise amplifier, and a TR switch using positive voltage control inputs with negative threshold depletion PHEMTs. This design is intended to operate at either the 425 or 900 MHz frequency bands. It is a 95x95 mil die.
• ARL25―This is a test circuit of the individual subcircuits from the previous four design variations. It contains a 100-mW power amplifier for 2.8 V, a 50-mW power amplifier for 2.8 V, a narrowband low DC power consumption low noise amplifier, and a TR switch. It is a 95x95 mil die.
27
The full designs were initially simulated in MWO, but then were re-simulated with ADS. It was easy to reuse the ADS templates from the 1st pass IC designs. As expected, results of the full IC with power amplifier, TR switch, BPSK modulator, and low noise amplifier are comparable to the combined individual simulations of the subcircuits.
All layouts were designed to a 2.41x2.41 mm die size to fit in the cavity of a 4x4 mm QFN package. RF input to the transmit chain is placed in the upper left of each layout. The RF output of the receive chain or low noise amplifier, is placed in the lower left of each layout. A common RF connection to the antenna is placed in the lower right of each layout. One intent of this layout is to separate the RF front end into a selectable module for a higher level package assembly. While the designs have similar layouts, one could interchange the designs depending on battery supply voltage (typically, 2.8 or 3.6 V), output power desired (currently 50 or 100 mW), and possibly frequency of operation (425 MHz, or both 425 and 900 MHz). Another goal of this design effort was to integrate the many lumped element matching circuits required as board level components for an RFIC transceiver into a single small IC package. The matching circuit portion of the tile of IC designs are discussed in a following section.
The design designated as ARL21M425 contains a BPSK modulator in the upper left of the layout, a power amplifier optimized for 100 mW at 2.8 V in the upper right of the layout, a TR switch in the lower right of the layout, and a low current (3-mA) low noise amplifier in the lower left of the layout (figure 29). A preliminary wire bond diagram of the die in a 4x4 mm SEMPAC, 20-pin, 4x4 mm QFN package is shown in figure 30.
Figure 29. Layout of ARL21M425 (2.41x2.41 mm die).
28
RFIN
RFOUT ANT
VA VB P28
L28 TA TB
Figure 30. Wire-bond diagram of ARL21M425 in a 4x4 mm QFN package.
The design designated as ARL22M425 contains a BPSK modulator in the upper left of the layout, a power amplifier optimized for 50 mW at 2.8 V in the upper right of the layout, a TR switch in the lower right of the layout, and a low current (3-mA) low noise amplifier in the lower left of the layout (figure 31). A preliminary wire bond diagram of the die in a 4x4 mm SEMPAC, 20-pin, 4x4 mm QFN package is shown in figure 32. This design is identical to ARL21M425 except for the power amplifier which is targeted for half as much RF and DC power.
29
Figure 31. Layout of ARL22M425 (2.41x2.41 mm die).
RFIN
RFOUT ANT
VA VB P28
L28 TA TB
Figure 32. Wire-bond diagram of ARL22M425 in a 4x4 mm QFN package.
30
The design designated as ARL23M425 contains a BPSK modulator in the upper left of the layout, a broadband power amplifier optimized for 50 mW at 3.6 V in the upper right of the layout, a TR switch in the lower right of the layout, and a low current (3-mA) low noise amplifier in the lower left of the layout (figure 33). A preliminary wire bond diagram of the die in a 4x4 mm SEMPAC, 20-pin, 4x4 mm QFN package is shown in figure 34. This design is identical to ARL21M425 and ARL22M425 except for the broadband power amplifier whose design was used in the 1st pass designs. The power amplifier has good efficiency over a range of supply voltages but is expected to have less than the desired 50 mW of output power with a 2.8-V supply. If a 3.6-V supply is used, the output power should be greater than 50 mW with good efficiency.
Figure 33. Layout of ARL23M425 (2.41x2.41 mm die).
31
RFIN
RFOUT ANT
VA VB P28
L28 TA TB
Figure 34. Wire-bond diagram of ARL23M425 in a 4x4 mm QFN package.
The design designated as ARL24DB contains a dual band 425/900-MHz BPSK modulator in the upper left of the layout, a broadband power amplifier optimized for 50 mW at 3.6 V in the upper right of the layout, a TR switch in the lower right of the layout, and a moderate current (9-mA) broadband low noise amplifier in the lower left of the layout (figure 35). A preliminary wire bond diagram of the die in a 4x4 mm SEMPAC, 20-pin, 4x4 mm QFN package is shown in figure 36. This design is similar to ARL23M425 except for the dual band BPSK modulator and the broadband higher current low noise amplifier, whose design was slightly modified from the broadband low noise amplifier in the 1st pass designs. As with ARL23M425, the power amplifier has good efficiency over a range of supply voltages but is expected to have less than the desired 50 mW of output power with a 2.8-V supply. If a 3.6-V supply is used, the output power should be greater than 50 mW with good efficiency. The broadband low noise amplifier uses more current than the narrowband 425-MHz design in the other three designs, but it is also expected to have 3 dB more of gain.
32
Figure 35. Layout of ARL24DB (2.41x2.41 mm die).
RFIN
RFOUT ANT
VA VB P28
L28 TA TB
Figure 36. Wire-bond diagram of ARL24DB in a 4x4 mm QFN package.
33
The design designated as ARL25 has individual subcircuits for test. It contains a power amplifier optimized for 50 mW at 2.8 V (ARL22M425) in the upper left of the layout, a broadband power amplifier optimized for 100 mW at 2.8 V (ARL21M425) in the upper right of the layout, a TR switch in the lower right of the layout, and a low current (3-mA) low noise amplifier in the lower left of the layout (figure 37). Since this is a test circuit, a wire-bond diagram is not shown, but it could be packaged for individual subcircuit testing with wire bonds. Note that the layout of ARL24DB has a probe testable layout of the moderate current (9-mA) broadband low noise amplifier. The only subcircuits not directly testable as individual elements are the narrowband and dual band BPSK modulators.
Figure 37. Layout of the ARL25 test cells (2.41x2.41 mm die).
The full IC designs were simulated to include the performance of the combined subcircuits, including the insertion loss affects of the TR switch. Figure 38 shows the full schematic of ARL21M425. A simulation of the transmit chain including the BPSK modulator, power amplifier, and TR switch is shown in figure 39. The receive chain simulation including the low noise amplifier and TR switch is shown in figure 40. Performance of the transmit and receive modes of ARL21M425 are shown in tables 7 and 8. As expected the output power, efficiency, and noise figure are slightly degraded by the insertion loss of the TR switch.
VD
S
VG
S
Use
500
ohm
s, 2
400
ohm
s fo
r 3.3
-3.6
VU
se 5
00 o
hms,
190
0 oh
ms
for 2
.7V
VA
RV
AR
1
Lh18
0=17
.7Ll
180=
17.7
Cl1
80=7
.1C
h180
=7.1
WF=
100
Vbi
asN
=3 -
Vbi
asV
bias
=0N
F=6
Eqn
Var
VA
RV
AR
14
VbA
N=2
.7 -
VbA
Vof
f=2.
7V
bA=2
.7
Eqn
Var
VA
RV
AR
13
VbA
N=2
.7 -
VbA
Vof
f=2.
7V
bA=0
Eqn
Var
Par
amS
wee
pS
wee
p3
Ste
p=2.
7S
top=
0S
tart=
-2.7
Sim
Inst
ance
Nam
e[6]
=S
imIn
stan
ceN
ame[
5]=
Sim
Inst
ance
Nam
e[4]
=S
imIn
stan
ceN
ame[
3]=
Sim
Inst
ance
Nam
e[2]
=S
imIn
stan
ceN
ame[
1]="
SP
1"S
wee
pVar
="V
bA"
PA
RA
ME
TER
SW
EE
P
Par
amS
wee
pS
wee
p1
Ste
p=2.
8S
top=
2.8
Sta
rt=0
Sim
Inst
ance
Nam
e[6]
=S
imIn
stan
ceN
ame[
5]=
Sim
Inst
ance
Nam
e[4]
=S
imIn
stan
ceN
ame[
3]=
Sim
Inst
ance
Nam
e[2]
=S
imIn
stan
ceN
ame[
1]="
SP
1"S
wee
pVar
="V
bias
"
PA
RA
ME
TER
SW
EE
P
S_P
aram
SP
1
Ste
p=0.
025
GH
zS
top=
2 G
Hz
Sta
rt=0.
1 G
Hz
S-P
AR
AM
ETE
RS
tqpe
d_ca
pC
12c=
62.5
pF
tqpe
d_ca
pC
1c=
106
pF
tqpe
d_ca
pC
4c=
117
pF
tqpe
d_re
sR
12
w=4
um
R=4
33 O
hm
tqpe
d_re
sR
13
w=3
5 um
R=3
Ohm
tqpe
d_re
sR
14
w=3
5 um
R=3
Ohm
Atte
nuat
orA
TTE
N1
VS
WR
=1Lo
ss=1
dB
tqpe
d_ph
ssQ
7
Ng=
1W
=wd2
um
Por
tP
10N
um=6
tqpe
d_pa
dP
2
tqpe
d_ca
pC
20c=
23.2
pF
tqpe
d_ca
pC
13c=
31.2
pF
tqpe
d_eh
ssQ
17
Ng=
1W
=6 u
mtqpe
d_re
sR
28
w=4
um
R=1
00 O
hm
tqpe
d_eh
ssQ
18
Ng=
10W
=96
um
tqpe
d_ca
pC
14c=
17.7
pF
tqpe
d_re
sR
29
w=4
um
R=1
35 O
hm
tqpe
d_re
sR
30
w=5
um
R=3
0 O
hm
Por
tP
46N
um=4
6
VA
RV
AR
15w
d2=2
Eqn
Var
V_D
CS
RC
11V
dc=2
.8 V
tqpe
d_ca
pC
6c=
4.65
pF
t
tqpe
d_ph
ssQ
14
Ng=
1W
=wd2
um
tqpe
d_eh
ssQ
15
Ng=
1 t
W=1
2 um
t
tqpe
d_m
rind
L13
LVS
_Ind
="LV
S_V
alue
"n=
40s=
10 u
mw
=10
um
tqpe
d_ca
pC
18c=
38.5
pF
t
tqpe
d_ca
pC
5c=
4.25
pF
t
tqpe
d_m
rind
L12
LVS
_Ind
="LV
S_V
alue
"n=
40s=
10 u
mw
=16
umtq
ped_
res
R27
w=3
um
R=3
kO
hm
tqpe
d_ca
pC
19c=
39 p
Ftq
ped_
res
R10
w=3
um
R=2
kO
hm
tqpe
d_re
sR
11
w=2
um
R=3
50 O
hmtq
ped_
ehss
Q16
Ng=
4 t
W=1
5 um
t
Por
tP
43N
um=4
3
Por
tP
44N
um=4
4
tqpe
d_pa
dP
45
tqpe
d_sv
iaV
7
L L26
R=
L=10
0000
0 nH
I_P
robe
I_P
robe
1
tqpe
d_re
sR
26
w=3
um
R=4
000
Ohm
tqpe
d_re
sR
25
w=3
um
R=4
000
Ohm
V_D
CS
RC
10V
dc=V
off V
tqpe
d_pa
dP
41
L L3 R=
L=10
00 n
H
Por
tP
42N
um=4
2
tqpe
d_re
sR
18
w=3
um
R=2
000
Ohm
tqpe
d_re
sR
19
w=3
um
R=2
000
Ohm
tqpe
d_re
sR
17
w=3
um
R=2
000
Ohm
tqpe
d_re
sR
16
w=3
um
R=2
000
Ohm
tqpe
d_ph
ssQ
8
Ng=
6W
=100
um
tqpe
d_ph
ssQ
10
Ng=
6W
=100
um
tqpe
d_ph
ssQ
1
Ng=
8W
=150
um
tqpe
d_ph
ssQ
9
Ng=
8W
=150
um
tqpe
d_m
rind
L25
LVS
_Ind
="LV
S_V
alue
"l2
=445
um
l1=5
40 u
mn=
28
tqpe
d_m
rind
L24
LVS
_Ind
="LV
S_V
alue
"l2
=445
um
l1=5
40 u
mn=
28
tqpe
d_m
rind
L14
LVS
_Ind
="LV
S_V
alue
"l2
=500
um
l1=4
80 u
mn=
28s=
10 u
m
tqpe
d_ca
pC
16c=
7.52
pF
tqpe
d_ca
pC
17c=
7.52
pF
tqpe
d_ca
pC
15c=
7.52
pF
tqpe
d_re
sR
15
w=3
um
R=8
000
Ohm
tqpe
d_re
sR
5
w=3
um
R=4
000
Ohm
tqpe
d_re
sR
2
w=3
um
R=4
000
Ohm
tqpe
d_de
ml
D3
w=1
0 um
tqpe
d_de
ml
D2
w=1
0 um
tqpe
d_re
sR
3
w=3
um
R=4
000
Ohm
tqpe
d_re
sR
4
w=3
um
R=4
000
Ohm
tqpe
d_re
sR
8
w=3
um
R=8
000
Ohm
tqpe
d_re
sR
6
w=3
um
R=4
000
Ohm
tqpe
d_eh
ssQ
6
Ng=
6W
=133
um
tqpe
d_eh
ssQ
3
Ng=
6W
=133
um
tqpe
d_eh
ssQ
5
Ng=
6W
=133
um
tqpe
d_eh
ssQ
4
Ng=
6W
=133
um
V_D
CS
RC
3V
dc=V
DS
V
Term
Term
3
Z=50
Ohm
Num
=3
tqpe
d_re
sR
1
w=2
um
R=2
000
Ohm
Por
tP
9N
um=5
Por
tP
32N
um=9
Por
tP
15N
um=1
5
Por
tP
8N
um=4
tqpe
d_sv
iaV
6Te
rmTe
rm2
Z=50
Ohm
Num
=2
tqpe
d_pa
dP
29
L L5 R=
L=10
00 n
H
L L4 R=
L=10
00 n
H
Por
tP
7N
um=3
Por
tP
5N
um=1
tqpe
d_sv
iaV
3
tqpe
d_sv
iaV
2
tqpe
d_pa
dP
1
Term
Term
1
Z=50
Ohm
Num
=1
tqpe
d_pa
dP
3
tqpe
d_pa
dP
4
V_D
CS
RC
1V
dc=V
bias
N V
V_D
CS
RC
2V
dc=V
bias
V
Por
tP
6N
um=2
Por
tP
33N
um=1
0
Por
tP
20N
um=2
0
Por
tP
13N
um=1
3
Por
tP
18N
um=1
8
tqpe
d_pa
dP
23
tqpe
d_pa
dP
31
L L18
R=
L=10
00 n
H
tqpe
d_pa
dP
22
tqpe
d_pa
dP
30
L L17
R=
L=10
00 n
H
V_D
CS
RC
6V
dc=V
bA V
V_D
CS
RC
5V
dc=V
bAN
V
L L6 R=
L=10
0000
0 nH
VA
RV
AR
4
VG
S=0
.52
VD
S=2
.7
Eqn
Var
VA
RV
AR
3
VG
S=0
.52
VD
S=3
.6
Eqn
Var
tqpe
d_in
clud
eN
ET1
capm
od=1
Sta
tistic
al_I
nfo=
On
Sta
tistic
al_A
naly
sis=
Off
Gat
e_Le
akag
e=N
omin
alta
u_gd
=Slo
wkE
GC
S=1
.0kR
hv=1
.0kM
IM=1
.0kI
s=1.
0kR
ni=1
.0kR
sh=1
.0kV
pe=0
kVpd
=0pH
EM
T_M
odel
=TO
M3
TQP
ED
Net
list I
nclu
de
Fig
ure
38. S
chem
atic
of
AR
L21
M42
5, 1
00 m
W a
nd 2
.8 V
(A
DS)
.
34
35
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.0 1.0
-25
-20
-15
-10
-5
0
5
10
15
20
-30
25
freq, GHz
dB(S
(2,1
))
m1
dB(S
(1,1
))
m6
dB(S
(2,2
))
m1freq=dB(S(2,1))=22.667Vbias=0.000000
425.0MHzm6freq=dB(S(1,1))=-18.947Vbias=0.000000
425.0MHz
Figure 39. S-parameter simulation of ARL21M425 transmit (ADS).
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.80.0 2.0
-20
-15
-10
-5
0
5
10
-25
15
freq, GHz
dB(S
(3,2
))
m7
dB(S
(3,3
))
m8
dB(S
(2,2
))
m7freq=dB(S(3,2))=12.288Vbias=2.800000
425.0MHzm8freq=dB(S(3,3))=-20.390Vbias=2.800000
425.0MHz
Figure 40. S-parameter simulation of ARL21M425 receive (ADS).
36
Table 7. Simulated performance of transmit chain for ARL21M425.
Power (dBm) Pout 0.425 GHz PAE 0.425 GHz Gain 0.425 GHz –16 6.66 2.36 22.66 –14 8.65 3.72 22.65 –12 10.64 5.86 22.64 –10 12.60 9.17 22.60 –8 14.51 14.12 22.51 –6 16.25 20.92 22.25 –4 17.66 28.89 21.66 –2 18.64 37.34 20.64 0 19.39 44.22 19.39 2 19.81 48.94 17.81
Table 8. Simulated performance of receive chain for ARL21M425.
Frequency (GHz) DB(NF()) DB(|S(2,1)|)
0.3 5.21 3.54 0.35 3.34 9.96
0.4 2.65 12.25
0.45 2.49 11.98
0.5 2.54 11.04
0.55 2.65 10.09 0.6 2.77 9.28
0.65 2.89 8.61
0.7 3.01 8.06
0.75 3.12 7.60 0.8 3.23 7.21
0.85 3.32 6.88
0.9 3.40 6.59
0.95 3.47 6.33
1 3.53 6.10
Figure 41 shows the full schematic of ARL22M425. A simulation of the transmit chain including the BPSK modulator, power amplifier, and TR switch is shown in figure 42. The receive chain simulation including the low noise amplifier and TR switch is shown in figure 43. Performance of the transmit and receive modes of ARL22M425 are shown in tables 9 and 10. As expected the output power, efficiency, and noise figure are slightly degraded by the insertion loss of the TR switch.
37
VG
S
VD
S
Use
500
ohm
s, 2
400
ohm
s fo
r 3.3
-3.6
VU
se 5
00 o
hms,
190
0 oh
ms
for 2
.7V
tqpe
d_ca
pC
13c=
31.2
pF
tqpe
d_ca
pC
14c=
17.7
pF
tqpe
d_ca
pC
12c=
62.5
pF
tqpe
d_re
sR
30
w=5
um
R=3
0 O
hmtq
ped_
cap
C22
c=1.
5 pF
tqpe
d_ca
pC
21c=
30.9
pF
tqpe
d_eh
ssQ
17
Ng=
1W
=6 u
m
tqpe
d_ph
ssQ
7
Ng=
1W
=wd2
um
VA
RV
AR
14
VbA
N=2
.7 -
VbA
Vof
f=2.
7V
bA=2
.7
Eqn
Var
VA
RV
AR
13
VbA
N=2
.7 -
VbA
Vof
f=2.
7V
bA=0
Eqn
Var
tqpe
d_eh
ssQ
18
Ng=
6W
=80
um
Por
tP
42N
um=4
2
tqpe
d_re
sR
12
w=4
um
R=4
33 O
hm
tqpe
d_re
sR
1
w=3
um
R=4
000
Ohm
Por
tP
47N
um=4
7tqpe
d_ca
pC
17c=
7.52
pF
Por
tP
7N
um=3
tqpe
d_sv
iaV
8
Por
tP
9N
um=5
tqpe
d_re
sR
26
w=3
um
R=4
000
Ohm
Por
tP
33N
um=1
0
Por
tP
8N
um=4
tqpe
d_m
rind
L13
LVS
_Ind
="LV
S_V
alue
"n=
40s=
10 u
mw
=10
um
Por
tP
43N
um=4
3
Por
tP
44N
um=4
4
Por
tP
46N
um=4
6
VA
RV
AR
1
Lh18
0=17
.7Ll
180=
17.7
Cl1
80=7
.1C
h180
=7.1
WF=
100
Vbi
asN
=3 -
Vbi
asV
bias
=0N
F=6
Eqn
Var
Par
amS
wee
pS
wee
p3
Ste
p=2.
7S
top=
0S
tart=
-2.7
Sim
Inst
ance
Nam
e[6]
=S
imIn
stan
ceN
ame[
5]=
Sim
Inst
ance
Nam
e[4]
=S
imIn
stan
ceN
ame[
3]=
Sim
Inst
ance
Nam
e[2]
=S
imIn
stan
ceN
ame[
1]="
SP
1"S
wee
pVar
="V
bA"
PA
RA
ME
TER
SW
EE
P
Par
amS
wee
pS
wee
p1
Ste
p=2.
8S
top=
2.8
Sta
rt=0
Sim
Inst
ance
Nam
e[6]
=S
imIn
stan
ceN
ame[
5]=
Sim
Inst
ance
Nam
e[4]
=S
imIn
stan
ceN
ame[
3]=
Sim
Inst
ance
Nam
e[2]
=S
imIn
stan
ceN
ame[
1]="
SP
1"S
wee
pVar
="V
bias
"
PA
RA
ME
TER
SW
EE
P
S_P
aram
SP
1
Ste
p=0.
025
GH
zS
top=
2 G
Hz
Sta
rt=0.
1 G
Hz
S-P
AR
AM
ETE
RS
tqpe
d_ca
pC
1c=
106
pF
tqpe
d_ca
pC
4c=
117
pF
tqpe
d_re
sR
13
w=3
5 um
R=3
Ohm
tqpe
d_re
sR
14
w=3
5 um
R=3
Ohm
Atte
nuat
orA
TTE
N1
VS
WR
=1Lo
ss=1
dB
Por
tP
10N
um=6
tqpe
d_pa
dP
2
tqpe
d_re
sR
28
w=4
um
R=1
00 O
hm tqpe
d_re
sR
29
w=4
um
R=1
35 O
hm
VA
RV
AR
15w
d2=2
Eqn
Var
V_D
CS
RC
11V
dc=2
.8 V
tqpe
d_ca
pC
6c=
4.65
pF
t
tqpe
d_ph
ssQ
14
Ng=
1W
=wd2
um
tqpe
d_eh
ssQ
15
Ng=
1 t
W=1
2 um
t
tqpe
d_ca
pC
18c=
38.5
pF
t
tqpe
d_ca
pC
5c=
4.25
pF
t
tqpe
d_m
rind
L12
LVS
_Ind
="LV
S_V
alue
"n=
40s=
10 u
mw
=16
umtq
ped_
res
R27
w=3
um
R=3
kO
hm
tqpe
d_ca
pC
19c=
39 p
Ftq
ped_
res
R10
w=3
um
R=2
kO
hm
tqpe
d_re
sR
11
w=2
um
R=3
50 O
hmtq
ped_
ehss
Q16
Ng=
4 t
W=1
5 um
t
tqpe
d_pa
dP
45
tqpe
d_sv
iaV
7
L L26
R=
L=10
0000
0 nH
I_P
robe
I_P
robe
1
tqpe
d_re
sR
25
w=3
um
R=4
000
Ohm
V_D
CS
RC
10V
dc=V
off V
tqpe
d_pa
dP
41
L L3 R=
L=10
00 n
H
tqpe
d_re
sR
18
w=3
um
R=2
000
Ohm
tqpe
d_re
sR
19
w=3
um
R=2
000
Ohm
tqpe
d_re
sR
17
w=3
um
R=2
000
Ohm
tqpe
d_re
sR
16
w=3
um
R=2
000
Ohm
tqpe
d_ph
ssQ
8
Ng=
6W
=100
um
tqpe
d_ph
ssQ
10
Ng=
6W
=100
um
tqpe
d_ph
ssQ
1
Ng=
8W
=150
um
tqpe
d_ph
ssQ
9
Ng=
8W
=150
um
tqpe
d_m
rind
L25
LVS
_Ind
="LV
S_V
alue
"l2
=445
um
l1=5
40 u
mn=
28
tqpe
d_m
rind
L24
LVS
_Ind
="LV
S_V
alue
"l2
=445
um
l1=5
40 u
mn=
28
tqpe
d_m
rind
L14
LVS
_Ind
="LV
S_V
alue
"l2
=500
um
l1=4
80 u
mn=
28s=
10 u
m
tqpe
d_ca
pC
16c=
7.52
pF
tqpe
d_ca
pC
15c=
7.52
pF
tqpe
d_re
sR
15
w=3
um
R=8
000
Ohm
tqpe
d_re
sR
5
w=3
um
R=4
000
Ohm
tqpe
d_re
sR
2
w=3
um
R=4
000
Ohm
tqpe
d_de
ml
D3
w=1
0 um
tqpe
d_de
ml
D2
w=1
0 um
tqpe
d_re
sR
3
w=3
um
R=4
000
Ohm
tqpe
d_re
sR
4
w=3
um
R=4
000
Ohm
tqpe
d_re
sR
8
w=3
um
R=8
000
Ohm
tqpe
d_re
sR
6
w=3
um
R=4
000
Ohm
tqpe
d_eh
ssQ
6
Ng=
6W
=133
um
tqpe
d_eh
ssQ
3
Ng=
6W
=133
um
tqpe
d_eh
ssQ
5
Ng=
6W
=133
um
tqpe
d_eh
ssQ
4
Ng=
6W
=133
um
V_D
CS
RC
3V
dc=V
DS
V
Term
Term
3
Z=50
Ohm
Num
=3P
ort
P32
Num
=9
Por
tP
15N
um=1
5tqp
ed_s
via
V6
Term
Term
2
Z=50
Ohm
Num
=2
tqpe
d_pa
dP
29
L L5 R=
L=10
00 n
H
L L4 R=
L=10
00 n
H
Por
tP
5N
um=1
tqpe
d_sv
iaV
3
tqpe
d_sv
iaV
2
tqpe
d_pa
dP
1
Term
Term
1
Z=50
Ohm
Num
=1
tqpe
d_pa
dP
3
tqpe
d_pa
dP
4
V_D
CS
RC
1V
dc=V
bias
N V
V_D
CS
RC
2V
dc=V
bias
V
Por
tP
6N
um=2
Por
tP
20N
um=2
0
Por
tP
13N
um=1
3
Por
tP
18N
um=1
8
tqpe
d_pa
dP
23
tqpe
d_pa
dP
31
L L18
R=
L=10
00 n
H
tqpe
d_pa
dP
22
tqpe
d_pa
dP
30
L L17
R=
L=10
00 n
H
V_D
CS
RC
6V
dc=V
bA V
V_D
CS
RC
5V
dc=V
bAN
V
L L6 R=
L=10
0000
0 nH
VA
RV
AR
4
VG
S=0
.52
VD
S=2
.7
Eqn
Var
VA
RV
AR
3
VG
S=0
.52
VD
S=3
.6
Eqn
Var
tqpe
d_in
clud
eN
ET1
capm
od=1
Sta
tistic
al_I
nfo=
On
Sta
tistic
al_A
naly
sis=
Off
Gat
e_Le
akag
e=N
omin
alta
u_gd
=Slo
wkE
GC
S=1
.0kR
hv=1
.0kM
IM=1
.0kI
s=1.
0kR
ni=1
.0kR
sh=1
.0kV
pe=0
kVpd
=0pH
EM
T_M
odel
=TO
M3
TQP
ED
Net
list I
nclu
de
Fig
ure
41. S
chem
atic
of
AR
L22
M42
5 50
mW
2.8
V (
AD
S).
38
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.0 1.0
-25
-20
-15
-10
-5
0
5
10
15
20
-30
25
freq, GHz
dB(S
(2,1
))
m1
dB(S
(1,1
))
m6
dB(S
(2,2
))
m1freq=dB(S(2,1))=20.672Vbias=0.000000
425.0MHzm6freq=dB(S(1,1))=-15.269Vbias=0.000000
425.0MHz
Figure 42. S-parameter simulation of ARL22M425 transmit (ADS).
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.80.0 2.0
-20
-15
-10
-5
0
5
10
-25
15
freq, GHz
dB(S
(3,2
))
m7
dB(S
(3,3
))
m8
dB(S
(2,2
))
m7freq=dB(S(3,2))=12.288Vbias=2.800000
425.0MHzm8freq=dB(S(3,3))=-20.390Vbias=2.800000
425.0MHz
Figure 43. S-parameter simulation of ARL22M425 receive (ADS).
39
Table 9. Simulated performance of transmit chain for ARL22M425.
Power (dBm) Pout 0.425 GHz PAE 0.425 GHz Gain 0.425 GHz –16 4.64 2.81 20.64 –14 6.62 4.40 20.62 –12 8.58 6.85 20.58 –10 10.52 10.55 20.52 –8 12.41 15.92 20.41 –6 14.20 23.11 20.20 –4 15.75 31.16 19.75 –2 16.93 38.06 18.93 0 17.76 43.19 17.76 2 18.21 46.66 16.21
Table 10. Simulated performance of receive chain for ARL22M425.
Frequency (GHz) DB(NF()) DB(|S(2,1)|)
0.3 5.19 3.54 0.35 3.31 9.96
0.4 2.61 12.25
0.45 2.45 11.98
0.5 2.50 11.04
0.55 2.62 10.09 0.6 2.74 9.28
0.65 2.87 8.61
0.7 2.98 8.06
0.75 3.09 7.60 0.8 3.19 7.21
0.85 3.27 6.88
0.9 3.34 6.59
0.95 3.41 6.33
1 3.47 6.10
Figure 44 shows the full schematic of ARL23M425. A simulation of the transmit chain including the BPSK modulator, power amplifier, and TR switch is shown in figure 45. The receive chain simulation including the low noise amplifier and TR switch is shown in figure 46. Performance of the transmit and receive modes of ARL23M425 are shown in tables 11, 12, and 13. Output power and efficiency are shown at both 2.7 and 3.6 V. The results shown at 2.7 V, should be slightly better using the expected 2.8-V supply. As expected the output power, efficiency, and noise figure are slightly degraded by the insertion loss of the TR switch.
40
Vds
Vgs
Use
500
ohm
s, 2
400
ohm
s fo
r 3.3
-3.6
VU
se 5
00 o
hms,
190
0 oh
ms
for 2
.7V
tqpe
d_eh
ssQ
2
Ng=
6 t
W=8
0 um
t
tqpe
d_ph
ssQ
7
Ng=
1W
=wd2
um
tqpe
d_eh
ssQ
17
Ng=
1W
=6 u
m
Por
tP
14N
um=1
4P
ort
P11
Num
=11
tqpe
d_ph
ssQ
10
Ng=
6W
=100
um
Por
tP
44N
um=4
4P
ort
P43
Num
=43
tqpe
d_re
sR
9
w=4
um
R=1
900
Ohm
t
tqpe
d_ca
pC
8c=
50.4
pF
t
tqpe
d_m
rind
L9 LVS
_Ind
="LV
S_V
alue
"n=
5s=
10 u
mw
=10
um
tqpe
d_ca
pC
7c=
31.1
pF
t
Por
tP
12N
um=1
2
tqpe
d_m
rind
L11
LVS
_Ind
="LV
S_V
alue
"n=
36s=
10 u
mw
=10
um
tqpe
d_re
sR
28
w=4
um
R=1
00 O
hm
tqpe
d_re
sR
7
w=4
um
R=5
00 O
hmtq
ped_
cap
C23
c=48
.0 p
F t
tqpe
d_ca
pC
10c=
54.6
pF
t
VA
RV
AR
13
VbA
N=2
.7 -
VbA
Vof
f=2.
7V
bA=0
Eqn
Var
VA
RV
AR
14
VbA
N=2
.7 -
VbA
Vof
f=2.
7V
bA=2
.7
Eqn
Var
V_D
CS
RC
3V
dc=2
.8 V
L L6 R=
L=10
0000
0 nH
Por
tP
49N
um=4
9
tqpe
d_pa
dP
48
tqpe
d_sv
iaV
9
tqpe
d_sv
iaV
4
tqpe
d_pa
dP
16
tqpe
d_re
sR
14
w=3
5 um
R=3
Ohm
Por
tP
42N
um=4
2
tqpe
d_re
sR
12
w=4
um
R=4
33 O
hm
tqpe
d_re
sR
1
w=3
um
R=4
000
Ohm
Por
tP
47N
um=4
7tqpe
d_ca
pC
17c=
7.52
pF
Por
tP
7N
um=3
tqpe
d_sv
iaV
8
Por
tP
9N
um=5
tqpe
d_re
sR
26
w=3
um
R=4
000
Ohm
Por
tP
33N
um=1
0
Por
tP
8N
um=4
tqpe
d_m
rind
L13
LVS
_Ind
="LV
S_V
alue
"n=
40s=
10 u
mw
=10
um
VA
RV
AR
1
Lh18
0=17
.7Ll
180=
17.7
Cl1
80=7
.1C
h180
=7.1
WF=
100
Vbi
asN
=3 -
Vbi
asV
bias
=0N
F=6
Eqn
Var
Par
amS
wee
pS
wee
p3
Ste
p=2.
7S
top=
0S
tart=
-2.7
Sim
Inst
ance
Nam
e[6]
=S
imIn
stan
ceN
ame[
5]=
Sim
Inst
ance
Nam
e[4]
=S
imIn
stan
ceN
ame[
3]=
Sim
Inst
ance
Nam
e[2]
=S
imIn
stan
ceN
ame[
1]="
SP
1"S
wee
pVar
="V
bA"
PA
RA
ME
TER
SW
EE
P
Par
amS
wee
pS
wee
p1
Ste
p=2.
8S
top=
2.8
Sta
rt=0
Sim
Inst
ance
Nam
e[6]
=S
imIn
stan
ceN
ame[
5]=
Sim
Inst
ance
Nam
e[4]
=S
imIn
stan
ceN
ame[
3]=
Sim
Inst
ance
Nam
e[2]
=S
imIn
stan
ceN
ame[
1]="
SP
1"S
wee
pVar
="V
bias
"
PA
RA
ME
TER
SW
EE
P
S_P
aram
SP
1
Ste
p=0.
025
GH
zS
top=
2 G
Hz
Sta
rt=0.
1 G
Hz
S-P
AR
AM
ETE
RS
tqpe
d_ca
pC
1c=
106
pF
tqpe
d_ca
pC
4c=
117
pF
tqpe
d_re
sR
13
w=3
5 um
R=3
Ohm
Atte
nuat
orA
TTE
N1
VS
WR
=1Lo
ss=1
dB
VA
RV
AR
15w
d2=2
Eqn
Var
V_D
CS
RC
11V
dc=2
.8 V
tqpe
d_ca
pC
6c=
4.65
pF
t
tqpe
d_ph
ssQ
14
Ng=
1W
=wd2
um
tqpe
d_eh
ssQ
15
Ng=
1 t
W=1
2 um
t
tqpe
d_ca
pC
18c=
38.5
pF
t
tqpe
d_ca
pC
5c=
4.25
pF
t
tqpe
d_m
rind
L12
LVS
_Ind
="LV
S_V
alue
"n=
40s=
10 u
mw
=16
umtq
ped_
res
R27
w=3
um
R=3
kO
hm
tqpe
d_ca
pC
19c=
39 p
Ftq
ped_
res
R10
w=3
um
R=2
kO
hm
tqpe
d_re
sR
11
w=2
um
R=3
50 O
hmtq
ped_
ehss
Q16
Ng=
4 t
W=1
5 um
t
tqpe
d_pa
dP
45
tqpe
d_sv
iaV
7
L L26
R=
L=10
0000
0 nH
I_P
robe
I_P
robe
1
tqpe
d_re
sR
25
w=3
um
R=4
000
Ohm
V_D
CS
RC
10V
dc=V
off V
tqpe
d_pa
dP
41
L L3 R=
L=10
00 n
H
tqpe
d_re
sR
18
w=3
um
R=2
000
Ohm
tqpe
d_re
sR
19
w=3
um
R=2
000
Ohm
tqpe
d_re
sR
17
w=3
um
R=2
000
Ohm
tqpe
d_re
sR
16
w=3
um
R=2
000
Ohm
tqpe
d_ph
ssQ
8
Ng=
6W
=100
um
tqpe
d_ph
ssQ
1
Ng=
8W
=150
um
tqpe
d_ph
ssQ
9
Ng=
8W
=150
um
tqpe
d_m
rind
L25
LVS
_Ind
="LV
S_V
alue
"l2
=445
um
l1=5
40 u
mn=
28
tqpe
d_m
rind
L24
LVS
_Ind
="LV
S_V
alue
"l2
=445
um
l1=5
40 u
mn=
28
tqpe
d_m
rind
L14
LVS
_Ind
="LV
S_V
alue
"l2
=500
um
l1=4
80 u
mn=
28s=
10 u
m
tqpe
d_ca
pC
16c=
7.52
pF
tqpe
d_ca
pC
15c=
7.52
pF
tqpe
d_re
sR
15
w=3
um
R=8
000
Ohm
tqpe
d_re
sR
5
w=3
um
R=4
000
Ohm
tqpe
d_re
sR
2
w=3
um
R=4
000
Ohm
tqpe
d_de
ml
D3
w=1
0 um
tqpe
d_de
ml
D2
w=1
0 um
tqpe
d_re
sR
3
w=3
um
R=4
000
Ohm
tqpe
d_re
sR
4
w=3
um
R=4
000
Ohm
tqpe
d_re
sR
8
w=3
um
R=8
000
Ohm
tqpe
d_re
sR
6
w=3
um
R=4
000
Ohm
tqpe
d_eh
ssQ
6
Ng=
6W
=133
um
tqpe
d_eh
ssQ
3
Ng=
6W
=133
um
tqpe
d_eh
ssQ
5
Ng=
6W
=133
um
tqpe
d_eh
ssQ
4
Ng=
6W
=133
um
Term
Term
3
Z=50
Ohm
Num
=3P
ort
P32
Num
=9
Por
tP
15N
um=1
5tqpe
d_sv
iaV
6Te
rmTe
rm2
Z=50
Ohm
Num
=2
tqpe
d_pa
dP
29
L L5 R=
L=10
00 n
H
L L4 R=
L=10
00 n
H
Por
tP
5N
um=1
tqpe
d_sv
iaV
3
tqpe
d_sv
iaV
2
tqpe
d_pa
dP
1
Term
Term
1
Z=50
Ohm
Num
=1
tqpe
d_pa
dP
3
tqpe
d_pa
dP
4
V_D
CS
RC
1V
dc=V
bias
N V
V_D
CS
RC
2V
dc=V
bias
V
Por
tP
6N
um=2
Por
tP
20N
um=2
0
Por
tP
13N
um=1
3
Por
tP
18N
um=1
8
tqpe
d_pa
dP
23
tqpe
d_pa
dP
31
L L18
R=
L=10
00 n
H
tqpe
d_pa
dP
22
tqpe
d_pa
dP
30
L L17
R=
L=10
00 n
H
V_D
CS
RC
6V
dc=V
bA V
V_D
CS
RC
5V
dc=V
bAN
V
VA
RV
AR
3
VG
S=0
.52
VD
S=3
.6
Eqn
Var
tqpe
d_in
clud
eN
ET1
capm
od=1
Sta
tistic
al_I
nfo=
On
Sta
tistic
al_A
naly
sis=
Off
Gat
e_Le
akag
e=N
omin
alta
u_gd
=Slo
wkE
GC
S=1
.0kR
hv=1
.0kM
IM=1
.0kI
s=1.
0kR
ni=1
.0kR
sh=1
.0kV
pe=0
kVpd
=0pH
EM
T_M
odel
=TO
M3
TQP
ED
Net
list I
nclu
de
Fig
ure
44. S
chem
atic
of
AR
L23
M42
5, 5
0 m
W 2
.8/3
.6 V
(A
DS
).
41
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.0 1.0
-25
-20
-15
-10
-5
0
5
10
15
20
-30
25
freq, GHz
dB(S
(2,1
))
m1
dB(S
(1,1
))
m6
dB(S
(2,2
))
m1freq=dB(S(2,1))=18.003Vbias=0.000000
425.0MHzm6freq=dB(S(1,1))=-8.351Vbias=0.000000
425.0MHz
Figure 45. S-parameter simulation of ARL23M425 transmit (ADS).
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.80.0 2.0
-20
-15
-10
-5
0
5
10
-25
15
freq, GHz
dB(S
(3,2
))
m7
dB(S
(3,3
))
m8
dB(S
(2,2
))
m7freq=dB(S(3,2))=12.288Vbias=2.800000
425.0MHzm8freq=dB(S(3,3))=-20.390Vbias=2.800000
425.0MHz
Figure 46. S-parameter simulation of ARL23M425 receive (ADS).
42
Table 11. Simulated performance of transmit chain for ARL23M425 (2.7 V).
Power (dBm) Pout 0.425 GHz PAE 0.425 GHz Gain 0.425 GHz –16 1.99 1.62 17.99 –14 3.98 2.57 17.98 –12 5.96 4.05 17.96 –10 7.93 6.39 17.93 –8 9.87 10.02 17.87 –6 11.71 15.63 17.71 –4 13.31 23.93 17.31 –2 14.41 33.61 16.41 0 14.78 39.55 14.78 2 14.74 41.75 12.74
Table 12. Simulated performance of transmit chain for ARL23M425 (3.6 V).
Power (dBm) Pout 0.425 GHz PAE 0.425 GHz Gain 0.425 GHz –16 1.99 1.12 17.99 –14 3.98 1.76 17.98 –12 5.97 2.79 17.97 –10 7.95 4.40 17.95 –8 9.92 6.92 17.92 –6 11.86 10.81 17.86 –4 13.74 16.73 17.74 –2 15.42 25.14 17.42 0 16.41 33.61 16.41 2 16.69 38.15 14.69
Table 13. Simulated performance of receive chain for ARL23M425.
Frequency (GHz) DB(NF()) DB(|S(2,1)|)
0.3 5.17 3.54 0.35 3.28 9.96
0.4 2.58 12.25
0.45 2.43 11.98
0.5 2.48 11.04
0.55 2.59 10.09 0.6 2.72 9.28
0.65 2.85 8.61
0.7 2.96 8.06
0.75 3.06 7.60 0.8 3.15 7.21
0.85 3.24 6.88
0.9 3.31 6.58
0.95 3.38 6.33
1 3.45 6.10
43
Figure 47 shows the full schematic of ARL24DB. A simulation of the transmit chain including the BPSK modulator, power amplifier, and TR switch is shown in figure 48. The receive chain simulation including the low noise amplifier and TR switch is shown in figure 49. Performance of the transmit and receive modes of ARL24DB are shown in tables 14 through 18. Output power and efficiency are shown for the frequencies 425 and 900 MHz as well as supply voltages of 2.7 and 3.6 V. As expected the output power, efficiency, and noise figure are slightly degraded by the insertion loss of the TR switch.
Vds
Vgs
Us e 500 ohm s , 2400 ohm s for 3 .3-3 .6VUs e 500 ohm s , 1900 ohm s for 2 .7V
VARVAR14
VbAN=2.7 - VbAVoff=2.7VbA=2.7
EqnVar
VARVAR13
VbAN=2.7 - VbAVoff=2.7VbA=0
EqnVar
PortP9Num =5
PortP8Num =4
tqped_dem lD3w=10 um
tqped_c apC15c =9 pF t
tqped_m rindL21
LVS_Ind="LVS_Value"l2=540 uml1=620 um tn=32
tqped_m rindL20
LVS_Ind="LVS_Value"l2=540 uml1=620 um tn=32
tqped_m rindL29
LVS_Ind="LVS_Value"l2=380 uml1=400 um tn=24s =10 um
tqped_c apC27c =9 pF t
tqped_res
R3
w=3 umR=4000 Ohm
tqped_resR4
w=3 umR=4000 Ohm
tqped_resR5
w=3 umR=4000 Ohm
tqped_c apC2c =9 pF t
tqped_c apC3c =3.4 pF t
tqped_m rindL1
LVS_Ind="LVS_Value"l2=380 um
l1=400 um tn=24s =10 um
tqped_c apC26c =2.1 pF t
tqped_resR2
w=3 umR=4000 Ohm
tqped_resR1
w=3 umR=4000 Ohm
tqped_phs sQ18
Ng=4W=10 umPort
P53Num =3
Port
P42Num =42
tqped_resR25
w=3 umR=4000 Ohm
PortP13Num =13
PortP33Num =10
tqped_resR6
w=3 umR=4000 Ohm
tqped_resR8
w=3 umR=8000 Ohm
PortP11Num =11
PortP12Num =12
PortP2Num =2
S_ParamSP1
Step=0.025 GHzStop=2 GHzStart=0.1 GHz
S-PARAMETERS
Param SweepSweep1
Step=2.8Stop=2.8Start=0Sim Ins tanc eNam e[6]=Sim Ins tanc eNam e[5]=Sim Ins tanc eNam e[4]=Sim Ins tanc eNam e[3]=Sim Ins tanc eNam e[2]=Sim Ins tanc eNam e[1]="SP1"SweepVar="Vb ias "
PARAMETER SWEEP
tqped_c apC28c =2.1 pF t
PortP51Num =51
tqped_s v iaV7
tqped_s v iaV8
PortP52Num =52
tqped_dem lD2w=10 um
TermTerm 1
Z=50 OhmNum =1
tqped_padP1
PortP5Num =1
tqped_ehs sQ4
Ng=6W=133 um
tqped_ehs sQ5
Ng=6W=133 um
VARVAR2
VGS=0.52RB=360wd2=2VDS=2.8
EqnVar
PortP18Num =18TermTerm 3
Z=50 OhmNum =3
tqped_padP22
tqped_resR10
w=4 umR=RB Ohm
tqped_ehs sQ19
Ng=6 tW=75 um t
tqped_resR29
w=4 umR=1100 Ohm t
S2PSNP1Fi le="E450V3I21.S2P"
21
R e f
V_DCSRC11Vdc =VDS V
tqped_c apC25c =30 pF t
tqped_c apC6c =25 pF t
tqped_c apC5c =25 pF t
tqped_m rindL26
LVS_Ind="LVS_Value"n=5s =10 umw=10 um
tqped_padP50
tqped_s v iaV1
LL28
R=L=1000000 nH
I_ProbeI_Probe1
Param SweepSweep3
Step=2.7Stop=0Start=-2.7Sim Ins tanc eNam e[6]=Sim Ins tanc eNam e[5]=Sim Ins tanc eNam e[4]=Sim Ins tanc eNam e[3]=Sim Ins tanc eNam e[2]=Sim Ins tanc eNam e[1]="SP1"SweepVar="VbA"
PARAMETER SWEEP
tqped_ehs s
Q2
Ng=6 tW=80 um t
tqped_phs sQ7
Ng=1W=wd2 um
tqped_ehs sQ17
Ng=1W=6 um
PortP14Num =14
tqped_phs sQ10
Ng=6W=100 um
tqped_resR9
w=4 umR=1900 Ohm t
tqped_c apC8c =50.4 pF t
tqped_m rindL9
LVS_Ind="LVS_Value"n=5s =10 umw=10 um
tqped_c apC7c =31.1 pF t
tqped_m rindL11
LVS_Ind="LVS_Value"n=36s =10 umw=10 um
tqped_resR28
w=4 umR=100 Ohm
tqped_resR7
w=4 umR=500 Ohmtqped_c apC23c =48.0 pF t
tqped_c apC10c =54.6 pF t
V_DCSRC3Vdc =2.8 V
LL6
R=L=1000000 nH
PortP49Num =49
tqped_padP48
tqped_s v iaV9
tqped_s v iaV4
tqped_padP16
tqped_resR14
w=35 umR=3 Ohm
tqped_resR12
w=4 umR=433 Ohm
tqped_resR26
w=3 um
R=4000 Ohm
VARVAR1
Lh180=17.7L l180=17.7Cl180=7.1Ch180=7.1WF=100Vbias N=3 - VbiasVbias =0NF=6
EqnVar
tqped_c apC1c =106 pF
tqped_c apC4c =117 pF
tqped_resR13
w=35 umR=3 Ohm
AttenuatorATTEN1
VSWR=1Los s =1 dB
V_DCSRC10Vdc =Voff V
tqped_padP41
LL3
R=L=1000 nH
tqped_resR18
w=3 umR=2000 Ohm
tqped_resR19
w=3 umR=2000 Ohm
tqped_resR17
w=3 umR=2000 Ohm
tqped_resR16
w=3 umR=2000 Ohmtqped_phs s
Q8
Ng=6W=100 um
tqped_phs sQ1
Ng=8W=150 um
tqped_phs sQ9
Ng=8W=150 um
tqped_resR15
w=3 umR=8000 Ohm
tqped_ehs sQ6
Ng=6W=133 um
tqped_ehs sQ3
Ng=6W=133 um
Port
P32Num =9
PortP15Num =15tqped_s v ia
V6TermTerm 2
Z=50 OhmNum =2
tqped_padP29
LL5
R=L=1000 nH
LL4
R=L=1000 nH
tqped_padP3
tqped_padP4
V_DCSRC1Vdc =Vbias N V
V_DCSRC2Vdc =Vbias V
PortP20Num =20
tqped_padP23
tqped_padP31
LL18
R=L=1000 nH
tqped_padP30
LL17
R=L=1000 nH
V_DCSRC6Vdc =VbA V
V_DCSRC5Vdc =VbAN V
VARVAR3
VGS=0.52VDS=3.6
EqnVar
tqped_inc ludeNET1
c apm od=1Statis tic a l_ In fo=OnStatis tic a l_Analy s is =OffGate_Leak age=Nom ina ltau_gd=Slowk EGCS=1.0k Rhv =1.0k M IM =1.0k Is =1.0k Rni=1.0k Rs h=1.0k Vpe=0k Vpd=0pHEM T_M odel=TOM 3
TQPEDNetl is t Inc lude
Figure 47. Schematic of ARL24DB, 425/900 MHz and 2.8/3.6 V (ADS).
44
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.00.0 1.1
-10
-5
0
5
10
15
-15
20
freq, GHz
dB(S
(2,1
))
m1m10 m11m12
dB(S
(1,1
))
m6 m13
dB(S
(2,2
))
m1freq=dB(S(2,1))=17.783Vbias=2.800000
425.0MHz
m6freq=dB(S(1,1))=-10.185Vbias=0.000000
425.0MHz
m10freq=dB(S(2,1))=18.665Vbias=0.000000
425.0MHzm11freq=dB(S(2,1))=17.941Vbias=2.800000
900.0MHzm12freq=dB(S(2,1))=17.627Vbias=0.000000
900.0MHz
m13freq=dB(S(1,1))=-9.215Vbias=2.800000
900.0MHz
Figure 48. S-parameter simulation of ARL24DB transmit (ADS).
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.80.0 2.0
-10
-5
0
5
10
15
-15
20
freq, GHz
dB(S
(3,2
))
m7
dB(S
(3,3
))
m8
dB(S
(2,2
))
m7freq=dB(S(3,2))=15.355Vbias=2.800000
425.0MHzm8freq=dB(S(3,3))=-8.189Vbias=2.800000
425.0MHz
Figure 49. S-parameter simulation of ARL24DB receive (ADS).
45
Table 14. Simulated performance of transmit chain for ARL24DB (425 M/2.7 V).
Power (dBm) Pout 0.425 GHz PAE 0.425 GHz Gain 0.425 GHz –16 2.58 1.86 18.58 –14 4.57 2.94 18.57 –12 6.55 4.65 18.55 –10 8.51 7.31 18.51 –8 10.43 11.44 18.43 –6 12.22 17.75 18.22 –4 13.71 26.63 17.71 –2 14.58 35.08 16.58 0 14.83 39.66 14.83 2 14.74 41.21 12.74
Table 15. Simulated performance of transmit chain for ARL24DB (425 M/3.6 V).
Power (dBm) Pout 0.425 GHz PAE 0.425 GHz Gain 0.425 GHz –16 2.57 1.28 18.57 –14 4.56 2.02 18.56 –12 6.55 3.19 18.55 –10 8.53 5.03 18.53 –8 10.49 7.89 18.49 –6 12.41 12.30 18.41 –4 14.25 18.87 18.25 –2 15.76 27.44 17.76 0 16.50 34.24 16.50 2 16.73 37.94 14.73
Table 16. Simulated performance of transmit chain for ARL24DB (900 M/2.7 V).
Power (dBm) Pout 0.90 GHz PAE 0.90 GHz Gain 0.90 GHz –16 1.63 1.49 17.63 –14 3.63 2.36 17.63 –12 5.61 3.74 17.61 –10 7.59 5.90 17.59 –8 9.55 9.28 17.55 –6 11.45 14.56 17.45 –4 13.17 22.68 17.17 –2 14.50 33.46 16.50 0 15.06 40.93 15.06 2 15.16 44.03 13.16
46
Table 17. Simulated performance of transmit chain for ARL24DB (900 M/3.6 V).
Power (dBm) Pout 0.90 GHz PAE 0.90 GHz Gain 0.90 GHz –16 1.58 1.02 17.58 –14 3.58 1.61 17.58 –12 5.57 2.54 17.57 –10 7.56 4.01 17.56 –8 9.54 6.33 17.54 –6 11.50 9.94 17.50 –4 13.43 15.50 17.43 –2 15.24 23.81 17.24 0 16.54 33.79 16.54 2 17.00 39.59 15.00
Table 18. Simulated performance of receive chain for ARL24DB.
Frequency (GHz) DB(NF()) DB(|S(2,1)|)
0.3 2.31 15.44 0.35 2.25 15.41
0.4 2.21 15.38
0.45 2.19 15.33
0.5 2.18 15.28
0.55 2.17 15.22 0.6 2.16 15.15
0.65 2.16 15.08
0.7 2.16 15.01
0.75 2.16 14.93 0.8 2.16 14.84
0.85 2.17 14.75
0.9 2.17 14.66
0.95 2.18 14.56
1 2.19 14.46
The circuits that make up the full RFIC were initially simulated with MWO. They were later simulated with ADS with similar results. Most of these subcircuits were included on the test chip ARL25, so those individual layouts were simulated with Sonnet’s 2.5-D Simulator. This helps to verify the actual layout connections, which are also verified with a layout versus schematic check (see section 7). Also, this checks the layout for unsimulated coupling, which caused a stability issue with the 2.4-GHz low noise amplifier from the 1st pass designs. Since ADS and MWO do not simulate the full coupling of the actual layout, only the Sonnet EM simulator showed an indication of the stability problem that occurred in that particular amplifier design. The Sonnet EM simulations assumes lossless 2-D metallization to reduce simulation times, so the losses will be underestimated, but unsimulated coupling and parasitics will be included. Results of the Sonnet simulations of the individual 50-mW power amplifier, 100-mW power
47
amplifier, 425-MHz low noise amplifier, and the 425-MHz BPSK modulator agreed well with the ADS and MWO simulations.
Test chip ARL25 contains a copy of the narrowband 425-MHz low noise amplifier included on the first three designs. The simulation results using ADS are similar to MWO with a slight increase in the noise figure (figures 50, 51, and 52). Sonnet was used to simulate the actual layout (figure 53) to check for any unsimulated parasitics. The three PHEMT connections are simulated as internal ports, and then the S-parameter file created for the Sonnet simulation is combined with the PHEMTs in ADS (or MWO) for comparison to the original simulations. Figures 54 and 55 show good agreement between the original simulations and the Sonnet simulations of the actual physical layout. Since the Sonnet EM simulation used lossless 2-D metal, the low noise amplifier gain tends to be slightly higher and the noise figure slightly lower, but overall agrees very well with the linear ADS/MWO simulations.
70.5 uV
70.5 uV 70.5 uV
70.5 uV
70.5 uV
70.5 uV
560 mV
560 mV560 mV
560 mV560 mV
560 mV
560 mV560 mV 560 mV
560 mV
560 mV
2.80 V
2.80 V
2.80 V2.80 V
2.80 V
2.80 V
2.78 V
2.78 V2.78 V
0 V
0 V0 V
0 V
560 mV
560 mV
2.78 V
426 uA
0 A
-2.38 fA
tqped_phssQ5
Ng=1W=wd2 um
VARVAR1
VGS=0.52RB=720wd2=2VDS=2.8
EqnVar
426 uA
-426 uA
0 A
tqped_ehssQ4
Ng=1 tW=12 um t
-2.72 mA
tqped_mrindL13
LVS_Ind="LVS_Value"n=40s=10 umw=10 um
0 Atqped_capC6c=4.65 pF t
0 Atqped_capC7c=38.5 pF t
0 Atqped_capC5c=4.25 pF t
0 A
tqped_mrindL12
LVS_Ind="LVS_Value"n=40s=10 umw=16 um0 A
tqped_resR12
w=3 umR=3 kOhm
0 Atqped_capC8c=39 pF
-6.37 nA
tqped_resR10
w=3 umR=2 kOhm
-57.3 uA
tqped_resR11
w=2 umR=350 Ohm
2.78 mA
-2.78 mA
6.37 nA
tqped_ehssQ2
Ng=4 tW=15 um t
-3.21 mA
V_DCSRC2Vdc=VDS V
PortP6Num=3
PortP5Num=2
PortP7Num=4Port
P4Num=1
0 A
TermTerm2
Z=50 OhmNum=1
0 A
tqped_padP3
0 A
tqped_padP2
0 A
tqped_padP1
3.21 mA
tqped_sviaV1
-3.21 mALL6
R=L=1000000 nH
2.78 mA
I_ProbeI_Probe1
0 A
TermTerm1
Z=50 OhmNum=2
Figure 50. Schematic of the 425-MHz low noise amplifier, ARL25 (ADS).
48
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.0 1.0
-20
-15
-10
-5
0
5
10
-25
15
freq, GHz
dB
(S(2
,1))
m8
dB
(S(1
,1)) m11
dB
(S(2
,2))
m12
m8freq=dB(S(2,1))=12.197
425.0MHz
m11freq=dB(S(1,1))=-4.533
425.0MHzm12freq=dB(S(2,2))=-20.446
425.0MHz
Figure 51. S-parameter simulation of the 425-MHz low noise amplifier, ARL25 (ADS).
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.0 1.0
1
2
0
3
freq, GHz
NF
min
nf(2
)
m1
m1freq=nf(2)=2.246
425.0MHz
Figure 52. Noise figure simulation of the 425-MHz low noise amplifier, ARL25 (ADS).
49
Figure 53. Sonnet EM layout simulation of the 425-MHz low noise amplifier, ARL25 (ADS).
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.0 1.0
-20
-15
-10
-5
0
5
10
-25
15
freq, GHz
dB
(S(2
,1))
m8
dB
(S(1
,1))
m11
dB
(S(2
,2))
m12
dB
(ln
a4
25
m..
S(2
,2))
dB
(ln
a4
25
m..
S(1
,1))
dB
(ln
a4
25
m..
S(2
,1))
m8freq=dB(S(2,1))=13.532
425.0MHz
m11freq=dB(S(1,1))=-5.294
425.0MHzm12freq=dB(S(2,2))=-13.030
425.0MHz
Figure 54. Sonnet (dotted) vs. ADS S-parameter simulation of the 425-MHz low noise amplifier, ARL25.
50
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.0 1.0
1
2
0
3
freq, GHz
NF
min
nf(
2)
m1
lna
42
5m
..nf(
2)
lna
42
5m
..NF
min
m1freq=nf(2)=2.024
425.0MHz
Figure 55. Sonnet (dotted) vs. ADS noise figure simulation of the 425-MHz low noise amplifier, ARL25.
A broadband low noise amplifier design that worked well from the 1st pass designs was slightly modified to reduce the current consumption. Modifying the feedback and stabilizing resistors to 1100 and 360 ohms as well as reducing the size of the active load PHEMT from a 4x15 to 4x10 µm, reduced the current consumption of the low noise amplifier, while still providing sufficient broadband gain and good performance. Figures 56, 57, and 58 show the ADS schematic and simulations of the modified broadband amplifier used in ARL24DB. This low noise amplifier subcircuit is included as a probe-able test circuit at the bottom center of the layout.
51
0 V
0 V
1.95 V
1.95 V
1.95 V
1.95 V 1.95 V
180 uV
180 uV
180 uV
2.80 V
1.66 mV
1.66 mV 1.66 mV
480 mV 480 mV
480 mV480 mV
480 mV1.95 V1.95 V 1.95 V 1.95 V
1.95 V
0 V
VARVAR3
VGS=0.52RB=360VDS=2.8
EqnVar
1.33 mA
tqped_resR10
w=4 umR=RB Ohm
-8.18 mA
8.18 mA
0 A
tqped_phssQ3
Ng=4W=10 um
6.84 mA
-6.84 mA
8.81 nA
tqped_ehssQ2
Ng=6 tW=75 um t
1.33 mAtqped_resR9
w=4 umR=1100 Ohm t
S2PSNP1File="E450V3I21.S2P"
21
Ref
-8.18 mA
V_DCSRC2Vdc=VDS V
0 Atqped_capC7c=30 pF t
0 Atqped_capC6c=25 pF t
0 Atqped_capC5c=25 pF t
6.84 mA
tqped_mrindL9
LVS_Ind="LVS_Value"n=5s=10 umw=10 um
0 A
tqped_padP3
0 A
tqped_padP2
0 A
tqped_padP1
8.18 mA
tqped_sviaV1
0 A
TermTerm2
Z=50 OhmNum=1
-8.18 mA
LL6
R=L=1000000 nH
8.18 mA
I_ProbeI_Probe1
0 A
TermTerm1
Z=50 OhmNum=2
Figure 56. Schematic of the modified (8 mA) broadband low noise amplifier, ARL24DB (ADS).
1 20 3
-10
-5
0
5
10
15
-15
20
freq, GHz
dB
(S(2
,1))
m4
dB
(S(1
,1))
m6dB
(S(2
,2))
m2
ou
r_m
axg
m4freq=dB(S(2,1))=15.732
400.0MHz
m6freq=dB(S(1,1))=-6.895
400.0MHzm2freq=dB(S(2,2))=-7.962
400.0MHz
Figure 57. Simulation of the modified (8 mA) broadband low noise amplifier, ARL24DB (ADS).
52
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.0 1.0
1
2
0
3
freq, GHz
NF
min
nf(2
)
m1
m1freq=nf(2)=1.939
400.0MHz
Figure 58. NF simulation of the modified (8 mA) broadband low noise amplifier, ARL24DB (ADS).
Test chip ARL25 contains a copy of the narrowband 425 MHz 100 mW PA included on the first design (ARL21M425). The simulation results using ADS are similar to MWO (see figures 59, 60, and 61). Sonnet was used to simulate the physical layout (figure 62) to check for any unsimulated parasitics. The three PHEMT connections are simulated as internal ports, and then the S-parameter file created for the sonnet simulation is combined with the PHEMTs in ADS (or MWO) for comparison to the original simulations. Figure 63 shows good agreement between the original simulation and the Sonnet simulation of the physical layout.
2.80 V
2.80 V
2.80 V
2.80 V2.80 V
1.73 mV
1.73 mV
1.73 mV
1.73 mV1.73 mV
639 mV
639 mV
639 mV
639 mVVGS
2.60 V2.60 V
2.60 V
639 mV
639 mV
639 mV
0 V
0 VVin
0 VVout0 V
2.60 VVDS426 uA
-426 uA
0 A
tqped_ehssQ6
Ng=1W=6 um
426 uA
0 A
-2.38 fA
tqped_phssQ7
Ng=1W=W2 um
0 Atqped_capC12c=62.5 pF
0 Atqped_capC14c=17.7 pF
482 nAtqped_resR5
w=4 umR=100 Ohm
0 A0 Atqped_capC15c=23.2 pF
0 A
TermTerm2
Z=25 OhmNum=2
0 A
TermTerm1
Z=50 OhmNum=1
78.3 mA
78.7 mA
-78.7 mA
PortP5Num=2
0 A-78.7 mA
PortP7Num=4
0 A I_ProbeIIn
0 A I_ProbeIOut
78.3 mA
-78.3 mA
482 nA
tqped_ehssQ5
Ng=10W=96 um
-78.3 mA
0 Atqped_capC13c=31.2 pF
482 nA tqped_resR3
w=5 umR=35 Ohm
-482 nA
tqped_resR4
w=4 umR=120 Ohm
Figure 59. Schematic of the 425-MHz, 100-mW power amplifier, ARL25 (ADS).
53
0.5 1.0 1.5 2.0 2.5 3.0 3.50.0 4.0
-20
-15
-10
-5
0
5
10
15
20
-25
25
freq, GHz
dB(S
(2,1
)) m4
m7
dB(S
(1,1
))
m6
dB(S
(2,2
))
m2
m4freq=dB(S(2,1))=2.626
2.400GHz
m7freq=dB(S(2,1))=24.183
425.0MHz
m6freq=dB(S(1,1))=-15.440
425.0MHzm2freq=dB(S(2,2))=-7.146
425.0MHz
Figure 60. Simulation of the 425-MHz, 100-mW power amplifier, ARL25 (ADS).
-14 -12 -10 -8 -6 -4 -2 0 2-16 4
6
8
10
12
14
16
18
20
4
22
10
20
30
40
50
0
60
RFpower
dB
m(V
ou
t[::
,1])
m1
PA
E1
m2
m1indep(m1)=plot_vs(dBm(Vout[::,1]), RFpower)=18.095
2.000
m2indep(m2)=plot_vs(PAE1, RFpower)=58.999
2.000
Figure 61. Power simulation of the 425-MHz, 100-mW power amplifier, ARL25 (ADS).
54
Figure 62. Sonnet EM layout simulation of the 425-MHz, 100-mW power amplifier, ARL25 (ADS).
0.5 1.0 1.5 2.0 2.50.0 3.0
-20
-15
-10
-5
0
5
10
15
20
-25
25
freq, GHz
dB
(S(2
,1)) m4
m7
dB
(S(1
,1))
m6
dB
(S(2
,2))
m2
dB
(pa
42
5m
_1
00
m_
v28
..S
(2,2
))d
B(p
a4
25
m_
10
0m
_v2
8..
S(1
,1))
dB
(pa
42
5m
_1
00
m_
v28
..S
(2,1
))
m4freq=dB(S(2,1))=1.387
2.400GHz
m7freq=dB(S(2,1))=23.973
425.0MHz
m6freq=dB(S(1,1))=-9.741
425.0MHzm2freq=dB(S(2,2))=-12.905
425.0MHz
Figure 63. Sonnet (dotted) vs. ADS S-parameter simulation of the narrowband 100-mW power amplifier, ARL25.
55
A copy of the narrowband 425-MHz, 50-mW power amplifier from the second design (ARL22M425) is also included on test chip ARL25. The simulation results using ADS are similar to MWO (figures 64, 65, and 66). Sonnet was used to simulate the physical layout (figure 67) to check for any unsimulated parasitics. The three PHEMT connections are simulated as internal ports, and then the S-parameter file created for the sonnet simulation is combined with the PHEMTs in ADS (or MWO) for comparison to the original simulations. Figure 68 shows good agreement between the original simulation and the Sonnet simulation of the physical layout.
VGS
0 V
0 V
2.75 VVDS
2.80 V
0 VVin
0 VVout
426 uA
-426 uA
0 A
tqped_ehssQ6
Ng=1W=6 um
-39.5 mA
0 A
tqped_capC17c=12 pF
0 Atqped_capC15c=30.9 pF
0 A
tqped_capC16c=1.5 pF
0 A
240 nA tqped_resR3
w=5 umR=30 Ohm
-240 nA
tqped_resR4
w=4 umR=135 Ohm
0 Atqped_capC12c=62.5 pF
39.5 mA
-39.5 mA
240 nA
tqped_ehssQ5
Ng=6W=80 um
0 A I_ProbeIOut
0 A I_ProbeIIn
PortP7Num=4
-39.9 mA0 APort
P5Num=2
-39.9 mA
39.9 mA
39.5 mA
0 A
TermTerm1
Z=50 OhmNum=1
0 A
TermTerm2
Z=25 OhmNum=2
426 uA
0 A
-2.38 fA
tqped_phssQ7
Ng=1W=W2 um240 nA
tqped_resR5
w=4 umR=100 Ohm
0 Atqped_capC14c=17.7 pF 0 A
tqped_capC13c=31.2 pF
Figure 64. Schematic of the 425-MHz, 50-mW power amplifier, ARL25 (ADS).
56
0.5 1.0 1.5 2.0 2.5 3.0 3.50.0 4.0
-20
-15
-10
-5
0
5
10
15
20
-25
25
freq, GHz
dB(S
(2,1
))
m4m7dB
(S(1
,1))
m6
dB(S
(2,2
))
m2
m4freq=dB(S(2,1))=21.347
425.0MHz
m7freq=dB(S(2,1))=21.347
425.0MHz
m6freq=dB(S(1,1))=-19.152
425.0MHzm2freq=dB(S(2,2))=-3.361
425.0MHz
Figure 65. Simulation of the 425-MHz, 50-mW power amplifier, ARL25 (ADS).
-14 -12 -10 -8 -6 -4 -2 0 2-16 4
6
8
10
12
14
16
18
20
4
22
10
20
30
40
50
0
60
RFpower
dB
m(V
ou
t[::
,1])
m1
PA
E1
m2
m1indep(m1)=plot_vs(dBm(Vout[::,1]), RFpower)=16.30
2.000
m2indep(m2)=plot_vs(PAE1, RFpower)=50.015
2.000
Figure 66. Power simulation of the 425-MHz, 50-mW power amplifier, ARL25 (ADS).
57
Figure 67. Sonnet EM layout simulation of the 425-MHz, 50-mW power
amplifier, ARL25 (ADS).
0.5 1.0 1.5 2.0 2.50.0 3.0
-20
-15
-10
-5
0
5
10
15
20
-25
25
freq, GHz
dB
(S(2
,1))
m4
m7
dB
(S(1
,1))
m6dB
(S(2
,2))
m2
dB
(pa
42
5m
_5
0m
_v2
8_
nb
..S
(2,2
)d
B(p
a4
25
m_
50
m_
v28
_n
b..
S(1
,1)
dB
(pa
42
5m
_5
0m
_v2
8_
nb
..S
(2,1
)
m4freq=dB(S(2,1))=-3.770
2.400GHz
m7freq=dB(S(2,1))=20.979
425.0MHz
m6freq=dB(S(1,1))=-12.129
425.0MHzm2freq=dB(S(2,2))=-3.767
425.0MHz
Figure 68. Sonnet (dotted) vs. ADS S-parameter simulation of the narrowband 50-mW power amplifier, ARL25
58
The TR switch design that worked well from the 1st pass design, but had higher than desired insertion loss, was modified for the 2nd pass to reduce the insertion loss to an expected 0.5 dB while retaining the positive control logic. ADS was used to re-simulate the TR switch (figure 69 shows the schematic) and the simulations are similar to that obtained with MWO (figure 70).
2 V
2 V
2 V2 V2 V
2.00 V
2.00 V 2.00 V2.00 V
2.00 V
2 V
2 V 2 V
2 V
2 V
0 V
0 V
0 V
2.00 V
0 V 0 V2.00 V
0 V 0 V
0 V2.00 V
2.00 V
2.00 V0 V0 V 0 V
About 8 Ohms Extra Resistance?Also, designed with 25pf caps, but built with 14-15 pF caps!
PortP13Num=7
-36.2 nA
LL3
R=L=1000 nH
-36.2 nAV_DCSRC3Vdc=Voff V
0 A
tqped_padP6
0 Atqped_capC2c=100 pF
0 Atqped_capC3c=100 pF
0 Atqped_capC5c=100 pF
0 Atqped_capC1c=100 pF
0 A SRLSRL1
L=0 nHR=0 Ohm
12.1 nA
12.1 nA
-24.1 nA
tqped_phssQ9
Ng=8W=150 um
-6.04 nA
6.04 nA
567 fA
tqped_phssQ1
Ng=8W=150 um
6.04 nA
6.04 nA
-12.1 nA
tqped_phssQ10
Ng=6W=100 um
-12.1 nA
12.1 nA
283 fA
tqped_phssQ8
Ng=6W=100 um
284 fAtqped_resR16
w=3 umR=2000 Ohm
-12.1 nAtqped_resR17
w=3 umR=2000 Ohm
567 fAtqped_resR19
w=3 umR=2000 Ohm
-24.1 nAtqped_resR18
w=3 umR=2000 Ohm
18.1 nA
tqped_resR25
w=3 umR=4000 Ohm
18.1 nA
tqped_resR26
w=3 umR=4000 Ohm
0 A
TermTerm3
Z=50 OhmNum=3
0 A
TermTerm2
Z=50 OhmNum=2
0 A
tqped_padP4
0 A
tqped_padP5
0 A
tqped_sviaV1
PortP10Num=4
PortP11Num=5
PortP12Num=6
PortP7Num=1
0 A
TermTerm1
Z=50 OhmNum=1
0 A
tqped_padP2
VARVAR3
VbAN=2 - VbAVoff=2VbA=0
EqnVar
36.2 nA
V_DCSRC1Vdc=VbA V
0 A
tqped_padP1
0 A
tqped_padP3
36.2 nALL1
R=L=1000 nH
-850 fALL2
R=L=1000 nH
-850 fA
V_DCSRC2Vdc=VbAN V
PortP9Num=3
PortP8Num=2
Figure 69. Schematic of the TR switch, ARL25 (ADS)
59
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.80.0 2.0
-20
-15
-10
-5
-25
0
freq, GHz
dB(S
(1,1
))m2
dB(S
(2,2
))dB
(S(3
,3))
m4
m2freq=dB(S(1,1))=-19.928
425.0MHzm4freq=dB(S(3,3))=-0.930
425.0MHz
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.80.0 2.0
-40
-35
-30
-25
-20
-15
-10
-5
-45
0
freq, GHz
dB(S
(2,3
))dB
(S(1
,2))
m3
dB(S
(1,3
))
m1
m3freq=dB(S(1,2))=-0.267
425.0MHzm1freq=dB(S(1,3))=-44.725
425.0MHz
Figure 70. Simulation of the TR switch, ARL25 (ADS)
The BPSK modulators, both the narrowband 425-MHz and the dual 425/900-MHz designs were not included as standalone test circuits. A narrowband version was very slightly retuned from the original 1st pass design to shift the center frequency from 450 to 425 MHz. The dual 425/ 900-MHz modulator adds two more elements to the high pass and low pass filters to provide a broadband 180° phase shift. The topology of low pass filter was reduced to two series inductors rather than three to save space. Only the narrowband modulator was simulated with Sonnet for comparison to ADS. The schematic and simulation results for the 425-MHz BPSK modulator using ADS are shown in figures 71, 72, and 73. Sonnet was used to simulate the physical layout (figure 74) to check for any unsimulated parasitics. The four switching PHEMTs were not part of the Sonnet EM simulation. For comparison, the S-parameter file created for the layout using Sonnet is combined with the PHEMT switches in ADS (or MWO) with good agreement to the original simulations. Figures 75 and 76 show good agreement between the original simulations
60
and using Sonnet to simulate the physical layout. Since the Sonnet EM simulation used lossless 2-D metal, the modulator insertion loss tends to be underestimated, but otherwise agrees very well with the linear ADS/MWO simulations.
tqped_mrindL24
LVS_Ind="LVS_Value"l2=445 uml1=540 umn=28
V_DCSRC1Vdc=VbiasN V
V_DCSRC2Vdc=Vbias V
LL4
R=L=1000 nH
LL5
R=L=1000 nH
tqped_resR15
w=3 umR=8000 Ohm
tqped_padP2
TermTerm2
Z=50 OhmNum=2
tqped_mrindL25
LVS_Ind="LVS_Value"l2=445 uml1=540 umn=28
tqped_mrindL14
LVS_Ind="LVS_Value"l2=500 uml1=480 umn=28s=10 um
tqped_capC16c=7.52 pF
tqped_capC17c=7.52 pF
tqped_capC15c=7.52 pF
tqped_resR5
w=3 umR=4000 Ohm
tqped_resR2
w=3 umR=4000 Ohm
tqped_demlD3w=10 um
tqped_demlD2w=10 um
tqped_resR3
w=3 umR=4000 Ohm
tqped_resR4
w=3 umR=4000 Ohm
tqped_resR8
w=3 umR=8000 Ohmtqped_res
R6
w=3 umR=4000 Ohm
tqped_ehssQ6
Ng=6W=133 um
tqped_ehssQ3
Ng=6W=133 um
tqped_ehssQ5
Ng=6W=133 um
tqped_ehssQ4
Ng=6W=133 um
tqped_resR1
w=2 umR=2000 Ohm
PortP9Num=5
PortP8Num=4
PortP7Num=3
tqped_sviaV3
tqped_sviaV2
tqped_padP10
tqped_padP4
PortP6Num=2
tqped_padP1
TermTerm1
Z=50 OhmNum=1
Figure 71. Schematic of the 425-MHz BPSK modulator (ADS).
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.0 1.0
-25
-20
-15
-10
-5
-30
0
freq, GHz
dB(S
(2,1
))
m1
dB(S
(1,1
)) m6
dB(S
(2,2
))
m1freq=dB(S(2,1))=-1.174Vbias=0.000000
425.0MHzm6freq=dB(S(1,1))=-14.880Vbias=0.000000
425.0MHz
Figure 72. Insertion loss simulation of the 425-MHz BPSK modulator (ADS).
61
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.1 1.0
-180
-135
-90
-45
0
45
90
135
-225
180
freq, GHz
phas
e(S
(2,1
))
m2
m3
phas
e(S
(2,1
))[0
,::]
-pha
se(S
(2,1
))[1
,::]
m7
m2freq=phase(S(2,1))=98.015Vbias=3.000000
425.0MHzm3freq=phase(S(2,1))=-84.102Vbias=0.000000
425.0MHz
m7freq=phase(S(2,1))[0,::]-phase(S(2,1))[1,::]=-182.117
425.0MHz
Figure 73. Phase simulation of the 425-MHz BPSK modulator (ADS).
Figure 74. Sonnet EM layout simulation of the 425-MHz BPSK modulator.
62
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.0 1.0
-25
-20
-15
-10
-5
-30
0
freq, GHz
dB(S
(2,1
))
m1
dB(S
(1,1
))
m6
dB(S
(2,2
))dB
(bps
k_42
5m..S
(2,2
))dB
(bps
k_42
5m..S
(1,1
))dB
(bps
k_42
5m..S
(2,1
))
m1freq=dB(S(2,1))=-0.677Vbias=0.000000
425.0MHzm6freq=dB(S(1,1))=-16.859Vbias=0.000000
425.0MHz
Figure 75. Sonnet (dotted) vs. ADS S-parameter simulation of the narrowband BPSK modulator.
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.1 1.0
-180
-135
-90
-45
0
45
90
135
-225
180
freq, GHz
phas
e(S
(2,1
))
m2
m3
phas
e(S
(2,1
))[0
,::]
-pha
se(S
(2,1
))[1
,::]
m7
phas
e(bp
sk_4
25m
..S
(2,1
))[0
,::]
-pha
se(b
psk_
425m
..S
(2,1
))[1
,::]
m2freq=phase(S(2,1))=97.056Vbias=3.000000
425.0MHzm3freq=phase(S(2,1))=-85.107Vbias=0.000000
425.0MHz
m7freq=phase(S(2,1))[0,::]-phase(S(2,1))[1,::]=-182.163
425.0MHz
Figure 76. Sonnet (dotted) vs. ADS phase simulation of narrowband BPSK modulator.
63
5. Integration of RFIC Matching Circuits
Most RFIC transceivers require quite a few external matching elements. Not only do these elements take up board space in the system, but they also tend to narrow the operation to a specific frequency band. For instance the Texas Instruments (TI) CC1000 sub-1 GHz transceiver and the more recent TI CC1100 sub-1 GHz transceiver were of interest for targeting a combined RFIC with the booster IC. While the transceiver may theoretically operate over several bands, the RFIC manufacturers expect the system to include all of the matching elements external to their RFIC. Multiple matching circuits are recommended depending on which band is desired. It would be difficult to broadband the matching circuits required by the RFIC. Also, it would save considerable board space if the matching circuits and balun circuits for the various RFIC transceivers could be integrated into a single small IC or package. A broadband or dual band antenna will be needed for the system, but we do not discuss that issue here. A small compact efficient system would include the RFIC, a small packaged IC incorporating all of the matching/balun elements, and the active booster IC to improve the RF performance for enhanced operating range or bandwidth. Figure 77 illustrates the concept of integrating the typical matching elements required between the RFIC and the antenna. The antenna connection in the figure can easily be extended to a booster IC plus the antenna for enhanced performance.
Figure 77. Concept of integrating RFIC matching elements into an IC for size, weight, and power (SWAP).
64
To demonstrate the integration of matching elements for improved SWAP, the data sheets for the CC1000 and CC1100 were used for initial designs of an integrated matching circuit. These two designs illustrate some of the difficulties and pitfalls of designing and integrating these matching circuits required by the RFIC. The older CC1000 design has a separate single ended RF input and RF output but the recommended matching circuit combines these two connections to a single antenna connection. It is not clear from the data sheet what the affect of this internal TR switch has on using the RF input and RF output as separate connections. Also, in attempts to simulate the recommended matching circuits and match to the specified impedances, there seem to be discrepancies for every frequency band. Unfortunately, there were no systems available with these RFICs to take direct measurements of the impedances and discern the best approach for matching. To complicate matters further, the newer design of the CC1100 has a very different set of impedances and recommended matching circuits that are not compatible with the earlier CC1000 RFIC. Likewise the simulations of the recommended CC1100 matching circuits do not seem to match the specified impedances. The CC1100 has the further complication that the RF connections are differential with a single pair of pads for a single RF input/output. Recall that the CC1000 RF connections were separate single ended connections combined into a single matching circuit with the possibility of some unspecified affect from an internal TR switch. At least the CC1100 looks promising in that the impedances recommended seem much closer to the center of a 50-ohm Smith chart, making their use more likely to be tolerant of matching errors and discrepancies. It would appear that the CC1100 is more tolerant of matching circuit variation than the older CC1000. However, there is no RFIC system available to make direct measurements for verifying the data sheets information.
Given the time constraints, the lumped element matching circuits were designed based on the best understanding of the data sheets for the CC1000 and CC1100. The main target was for the 425 MHz operation of each RFIC as well as a dual band matching circuit for the newer CC1100 design which uses switches to operate at either the 425 MHz band or the 900 MHz band. This was intended to be combined with the ARL24DB Booster IC to provide a small wireless system using the CC1100 RFIC, the integrated matching circuit, and the Booster IC for a switchable dual band operation at 425/900 MHz. At the least, the circuits will demonstrate the feasibility of the size advantages of integrating the matching elements for the RFIC into a single packaged IC. TriQuint does offer a high performance GaAs passive fabrication process (TQRLC) which is less expensive and requires fewer processing steps by omitting the active PHEMT devices, extra masks, and processing steps.
The following are descriptions of the passive designs:
• ARL26DB―This dual band design contains a matching circuit plus RF switches to connect the integrated matching circuit for a TI CC1100 RFIC for operation at either 425 or 900 MHz. One side of the design connects to the differential RF connections of the CC1100 RFIC and the other side switches to one of two single ended RF connections,
65
intended to be either an input to a transmitter/power amplifier and the other an output from a receiver/low noise amplifier. It is a 95x95 mil die.
• ARL27M425―This design contains discrete lumped elements circuits for an integrated match at 425-MHz circuit for a TI CC1000 RFIC. It separates the recommended matching circuit into an RF input and RF output match for separate connections as either an input to a transmitter/power amplifier or an output from a receiver/low noise amplifier rather than a single-ended connection to an antenna. It is a 95x50 mil die.
• ARL28M900―This design contains discrete lumped elements circuits for an integrated match at 900-MHz circuit for a TI CC1000 RFIC. It separates the recommended matching circuit into an RF input and RF output match for separate connections as either an input to a transmitter/power amplifier or an output from a receiver/low noise amplifier rather than a single-ended connection to an antenna. It is similar to ARL27M425 except for the frequency of operation. It is a 95x50 mil die.
• ARL29M425―This single band design contains a matching circuit to connect the integrated matching circuit for a TI CC1100 RFIC for operation at 425 MHz. One side of the design connects to the differential RF connections of the CC1100 RFIC and the other side connects to a single-ended RF connections, intended to be the antenna connection. It is a 95x50 mil die.
• ARL29M425S―This single band design contains a matching circuit to connect the integrated matching circuit for a TI CC1100 RFIC for operation at 425 MHz. One side of the design connects to the differential RF connections of the CC1100 RFIC and the other side connects to RF switches intended for connection to the input of a transmitter/power amplifier or an output from a receiver/low noise amplifier rather than the single-ended connection to an antenna of ARL29M425 which is a similar design. It is a 95x50 mil die.
The circuit designated as ARL26DB is designed for matching the CC1100 at 425 MHz and also 900 MHz, using PHEMT switches to choose between the two frequency bands (figure 78). The values of the lumped elements and the topologies of the matching circuits used are the ones recommended in the CC1100 data sheet Connections to the CC1100 on the upper-left and lower-left sides of the layout use a differential RF connection while the right-hand side of the layout has switches with a separate control to select between one of two RF single-ended connections, presumably the transmit and receive paths of the booster IC. A preliminary wire bond diagram of the die in a 4x4 mm SEMPAC, 20-pin, 4x4 mm QFN package is shown in figure 79. Note that this circuit is mostly passive, i.e., uses lumped matching elements, but also uses PHEMTs as switches to choose the frequency band, and to choose one of two single-ended RF connections.
66
Figure 78. Layout of ARL26DB (2.41x2.41 mm die).
67
FA
RFOUT2
RFIN2
RFOUT1
TA
TB
RFIN1
FB
Figure 79. Wire-bond diagram of ARL26DB in a 4x4 mm QFN package.
The circuit designated as ARL27M425 is designed for matching the CC1000 at 425 MHz (figure 80). The values of the lumped elements in the matching circuits are those recommended in the CC1000 data sheet. However, the topology is split into an RF input and an RF output. The CC1000 has two single ended RF connections, one intended for input and one intended for output with an internal TR switch whose effect on the impedance is not specified in the data sheets. Originally, the topology connects a shunt inductor/series capacitor matching element to each of the CC1000 RF connections, which then connect to a simple low pass filter leading to a single RF antenna connection. Since the booster IC has a separate transmit path and receive path, the topology of this design uses two separate paths with two separate low pass filters, each connected to a series inductor/shunt capacitor combination for matching. In the layout, the upper-left connects to the booster low noise amplifier/receive path and the lower-left connects to the CC1000 RF input. Likewise, the upper right connects to the Booster power amplifier/transmit path and the lower right connects to the CC1000 RF output. A preliminary wire-bond diagram of the die in a 4x4 mm, SEMPAC, 20-pin, 4x4 mm QFN package is shown in figure 81. Note that this circuit is entirely passive, i.e., uses only lumped elements.
68
Figure 80. Layout of ARL27M425 (2.41x1.27 mm die).
RFIN2
RFOUT2 RFOUT1
RFIN1
Figure 81. Wire-bond diagram of ARL27M425 in a 4x4 mm QFN package.
69
The circuit designated as ARL28M900 is designed for matching the CC1000 at 900 MHz (figure 82). The values of the lumped elements in the matching circuits are those recommended in the CC1000 data sheet. This circuit’s topology is exactly the same as that described for ARL27M425, the difference is the change of lumped element values to suit the higher frequency band. In the layout, the upper left connects to the booster low noise amplifier/receive path and the lower left connects to the CC1000 RF input. Likewise, the upper right connects to the booster power amplifier/transmit path and the lower right connects to the CC1000 RF output. A preliminary wire bond diagram of the die in a 4x4 mm SEMPAC, 20-pin, 4x4 mm QFN package is shown in figure 83. Note that this circuit too is entirely passive, i.e., uses only lumped elements.
Figure 82. Layout of ARL28M900 (2.41x1.27 mm die).
70
RFIN2
RFOUT2 RFOUT1
RFIN1
Figure 83. Wire-bond diagram of ARL28M900 in a 4x4 mm QFN package.
The circuit designated as ARL29M425 is designed for matching the CC1100 at 425 MHz (figure 84). The values of the lumped elements and the topologies of the matching circuits are those recommended in the CC1100 data sheet. In the layout, connections to the CC1100 are on the upper-left and lower-left sides of the layout creating a differential RF connection while the right-hand side of the layout is a single-ended RF connection, intended to be the antenna. This circuit will demonstrate the improved SWAP by integrating the matching elements for the CC1100. However, this design has a single RF connection, while the booster IC has a separate single ended RF connection for the transmit and receive paths—see ARL29M425S. A preliminary wire-bond diagram of the die in a 4x4 mm SEMPAC, 20-pin, 4x4 mm QFN package is shown in figure 85. Note that this circuit is entirely passive, i.e., uses only lumped elements.
71
Figure 84. Layout of ARL29M425 (2.41x1.27 mm die).
RFOUT1
RFOUT2
ANT
Figure 85. Wire-bond diagram of ARL29M425 in a 4x4 mm QFN package.
The circuit designated as ARL29M425S is designed for matching the CC1100 at 425 MHz (figure 86) and is based on the ARL29M425 design. The difference between the designs is the addition of a single pole double throw TR switch to connect the single-ended “antenna”
72
connection to either the transmit or receive RF connections of the booster IC. In the layout, the differential RF connection to the CC1100 is on the upper-left and lower-left sides while the right-hand side has a TR switch to connect to either a singled-ended RF connection to the right or to the bottom. A preliminary wire bond diagram of the die in a 4x4 mm SEMPAC, 20-pin, 4x4 mm QFN package is shown in figure 87. Note that this circuit is mostly passive, i.e., uses lumped matching elements, but also uses PHEMTs as switches to choose between one of two single ended RF connections.
Figure 86. Layout of ARL29M425S (2.41x1.27 mm die).
73
RFOUT1
RFOUT2 RFIN1
RFIN2
VR VA VB
Figure 87. Wire-bond diagram of ARL29M425S in a 4x4 mm QFN package.
The CC1110 data sheet gives the differential port impedance towards the antenna as 116 + j41 Ω at 433 MHz, and 86.5 + j43 Ω at 868 MHz. For the CC1000, the singled-ended receive mode impedance is at 70 + j26 Ω 433 MHz and 52 + j4 Ω at 915 MHz, while the singled-ended transmit mode impedance is at 140 Ω 433 MHz and 80 Ω at 868 MHz. Simulations of recommended matching circuits for both the CC1100 and CC1000 resulted in slightly different answers for both sets of circuits. Also, the simulations for the CC1000 seemed to be very narrowly resonant circuits that were not near the 50-ohm match point of the Smith chart. At least the CC1100 matching circuit simulations seemed to have a wider resonance and were closer to the 50-ohm match point of the Smith chart. At these relatively low frequencies, the lumped element parasitics should be fairly small. A design kit of the suggested 0402 Murata chip capacitor and chip inductor elements was obtained for Agilent’s ADS program, and the matching circuits were re-simulated with results similar to using ideal elements. There were no boards for either the CC1000 or CC1100 RFICs to measure and verify the discrepancies in the data sheets. So, it is likely that these matching circuit designs will not be ideal for this 1st pass of designs. From the simulations, it would appear that the CC1100 circuits are more likely to be “close” to the desired impedance match. At least, the matching circuits should demonstrate the system SWAP improvement and feasibility of integrating the many lumped element matching elements required by a typical RFIC transceiver into a single compact IC. Also, it should be noted that the data sheet for the CC1000 has a component designated as C41, which TI recommends omitting
74
for the CC1000 UltraCSP package. But that omission may be the cause of some problems seen in a tag design by Sandia. Their experience seems to indicate that the system works better if you include C41, an indication that impedance matching for the CC1000 may be critical and very sensitive, in addition to verifying an error in the data sheet.
The schematic for the CC1100 matching circuit at 433 MHz is shown in figure 88 and for 900 MHz in figure 89. Note that they are similar but slightly different topologies. It would be good to explore the possibility of creating a single broadband topology of lumped elements that could create the desired match for both bands without having to use the switched topology approach used in the ARL26DB design. Simulations showing similar agreement between the TriQuint element matching circuit and the original ideal element match are shown in figure 90 for the 433-MHz design and figure 91 for the 900-MHz design. A layout was created for the 900-MHz matching circuit but it was not included in this tile for fabrication, only the 433-MHz design was created as ARL29M425 with a single RF connection to an antenna, and as ARL29M425S, which includes a TR switch to connect to the transmit and receive RF connections of the booster IC.
Figure 88. Schematic of the CC1100 matching circuit for 425/433 MHz using TriQuint elements.
75
Figure 89. Schematic of the CC1100 matching circuit for 900 MHz using TriQuint elements.
Figure 90. Simulation of the CC1100 matching circuit for 425/433 MHz (TQS and ideal).
76
Figure 91. Simulation of the CC1100 matching circuit for 900 MHz (TQS and ideal).
The schematic for the CC1000 matching circuit at 433 MHz is shown in figure 92 and for 900 MHz in figure 93. Note that they are identical except for the capacitor (C41), which is recommended for the lower frequency bands if the UltraCSP package is not used. It would be good to explore a single broadband topology of lumped elements that could create the desired match for both bands, but it looks like it would be very difficult for this part based on these simulations. Also, the topology of the matching circuit assumes a single RF connection to an antenna. For use with the booster IC, the matching circuits were split into a separate transmit and receive path as shown in figure 94. Figure 95 shows a Smith chart simulation of the RF input and RF output impedances for the recommended matching circuits of the CC1000 at 315, 433, and 868 MHz. It is a cluttered plot, but what is evident is that each band seems to have a narrow resonance that is not as close to the 50-ohm Smith chart center as one would want. Also the simulated impedances do not match the recommended values, and there is the possibility that the internal TR switch of the CC1000 has some effect on the impedances that is not documented. For simulation purposes, the matching circuits were split into a low pass filter and two series capacitor/shunt inductor matching elements for the RF input and output. Then these sub circuits using TriQuint elements were tuned to match the corresponding ideal element versions, shown in the simulations at 433 MHz (figure 96). The only dual band matching design was for the newer CC1100 RFIC parts. Figure 97 shows the schematic of ARL26DB where a set of three switches
77
select either the matching circuits at 433 or 900 MHz followed by a switch to select either the transmit or receive RF connection of the booster IC.
Figure 92. Schematic of the CC1000 matching circuit for 433 MHz using ideal elements.
Figure 93. Schematic of the CC1000 matching circuit for 900 MHz using ideal elements.
78
Figure 94. Split of the CC1000 matching circuit into separate transmit and receive circuits.
Figure 95. Simulation of the CC1000 matching circuits for 315, 433, and 900 MHz (ideal).
79
Figure 96. Simulation of the CC1000 split matching circuits for 433 MHz (TQS and ideal).
Figure 97. Schematic of the CC1100 dual band matching circuit (ARL26DB).
80
6. Design Rule Checking (DRC)
All designs within the tile must be checked according to the process design rules supplied by TriQuint. TriQuint has design rules for their fabrication processes that work with the open-source ICED program. Design verification typically consists of design rule checking (DRC) and layout versus schematic checking (LVS). For fabrication feasibility, TriQuint only cares that the tile passes DRC checks and can be fabricated. While some design rules may be oriented towards maximizing yield and are not fatal errors, any violations of the design rules must be removed or granted a waiver to fabricate by the foundry. Additionally, it is up to the designer to verify that the designs are connected as desired and that parameter values are correct by doing additional LVS checking. After completing DRCs with ICED, a final DRC check was performed on the tile layout using TriQuint’s “mailDRC” service. TriQuint provides a free e-mail based service to provide a final DRC check using Cadence’s Assura software. Next, there is a checklist for the foundry to verify higher level fabrication issues such as labeling. Each individual die must have a unique label using the correct metal layer and be of sufficient size. Every die name starts with the letters “ARL” to help identify the customer so that the individual dice can be sorted and returned appropriately. The second ARL tile passed the mailDRC checking by TriQuint and was accepted for integration into the fabrication process.
7. Layout versus Schematic (LVS) Checking
Each design is verified against a schematic (i.e., a netlist), which can be generated from MWO or ADS. Both tools require manual editing of the generated netlist, which typically requires an iterative LVS check until all device connections are verified. Foundries do not require LVS checking, only DRC checking, but it is imperative for the designer to verify the connections in the layout. A DRC correct layout does not guarantee that there are no shorts or opens that could cause the circuit to fail. Verifying the connections requires LVS checking followed by a final DRC check if any layout changes occur. If the design passes DRC and LVS checking, then the probability of functional success is extremely high. In addition to checking the connections, the LVS check also notes parametric differences between the schematic and layout. If the capacitors, resistors, or PHEMTs differ in size, type, or value, the LVS check will list the mismatches. Inductors are not checked parametrically, only their connectivity is verified. All of the designs on the second ARL tile passed LVS checking. The LVS checks did find at least one major error in connectivity that was not found by DRC checks alone.
81
Another check beyond DRC and LVS checks are electrical rules. It is a good idea to check the high current paths in every design—i.e., the active designs for this tile. The path for drain current to the PHEMTs in the power amplifiers is the most likely area for concern. A minor increase in a line width for the drain connection to the 50-mW, 425-MHz power amplifier was performed after examining the designs.
8. Tile Layout
The prototype fabrication offered by TriQuint provides a 5x10 mm quarter tile, which the designer can partition into multiple designs. There are a number of limitations, for example, all the rows and columns must align. Typically the row and column dimensions are limited to a few choices and may even use multiples of some fixed value. Since these designs were targeted for a 4x4 mm QFN package, the 95-mil dimension was chosen to provide the maximum circuit design area. Die size versus the number of different designs is a tradeoff in the tile layout. For the smaller matching circuits, the width was made 95 mils to fit in the 4x4 mm QFN package, but the height was reduced to 50 mils to yield an extra row of designs (two). Half of the 10 designs in the tile are active booster IC designs and half are dedicated to integrating the RFIC matching circuits into a single IC to improve SWAP. Figure 98 shows a plot of the final 4.82x9.77 mm ARLTILE2. Table 19 lists the design locations within the tile. Each design will get diced into individual gel packs with typically 50 copies of each of the 10 designs for test.
82
Figure 98. Layout plot of ARLTILE2 (10 designs–4.8x9.8 mm)
83
Table 19. Map of design layout within ARLTILE2.
Die Size ( Y µm)
Row 2410 X µm
C1 C2
1270 R1 ARL29M425S ARL28M900
1270 R2 ARL29M425 ARL27M425
2410 R3 ARL26DB ARL25
2410 R4 ARL23M425 ARL24DB
2410 R5 ARL21M425 ARL22M425
The designs were submitted August 24, 2010, to TriQuint for fabrication. It is expected that they will be returned in late October 2010 when probe testing of the individual die will be performed. Designs will also be packaged for testing the die plus package. The goal is to integrate the most promising designs into a wireless system to demonstrate the enhanced performance and improved SWAP of the booster IC and integrated matching circuit designs.
9. Conclusion
While I performed the designs, the end product was a team effort. A design review held with many participants, mostly in the RDRL-SER-E branch, was very helpful in making sure that these were the best designs achievable under the existing constraints. A number of people helped with the testing, reviewing, and documenting of the 1st pass booster IC designs, which were then used as a basis for these optimized 2nd pass designs. For further documentation of the 1st pass designs and TriQuint design package for the 2nd pass, see the references 1–5.
84
10. References
1. Mitchell, G.; Penn, J. Preliminary Gallium Arsenide (GaAs) Integrated Circuit Design for Radio Frequency Booster Chips at 450, 900, and 2400 MHz; ARL-TR-4970; U.S. Army Research Laboratory: Adelphi, MD, September 2009.
2. Penn, J. GaAs Microwave Integrated Circuit Designs Submitted to TriQuint Semiconductor for Fabrication; ARL-TN-0381; U.S. Army Research Laboratory: Adelphi, MD, December 2009.
3. Mitchell, G.; Penn, J. Results of Bare Die Probing for RF Booster Chip at 400, 900, 2400 MHz; ARL-TR-5170; U.S. Army Research Laboratory: Adelphi, MD, April 2010.
4. Penn, J. Testing of GaAs Microwave Integrated Circuit Designs in QFN Packages; ARL-TR-5131; U.S. Army Research Laboratory: Adelphi, MD, March 2010
5. Penn, J. GaAs Microwave Integrated Circuit Designs Submitted to TriQuint Semiconductor for Fabrication (ARL Tile #2); ARL-TN-0404; U.S. Army Research Laboratory: Adelphi, MD, September 2010.
85
List of Symbols, Abbreviations, and Acronyms
ADS Advanced Design System
AWR Applied Wave Research
BPSK binary phase shift keying
CAD computer-aided design
DMODE depletion mode
DRC design rule checking
EM electromagnetic
EMODE enhancement mode
GaAs gallium arsenide
IC integrated circuit
ICED Integrated Circuit Editor
LVS layout versus schematic checking
MMIC monolithic microwave integrated circuit
MWO Microwave Office
PAE power added efficiency
PHEMT pseudomorphic high electron mobility transistor
QFN quad flat no lead
RF radio frequency
RFIC RF integrated circuit
TI Texas Instruments
TR transmit/receive
UHF ultra-high frequency
86
NO. OF COPIES ORGANIZATION 1 DEFENSE TECHNICAL (PDF INFORMATION CTR only) DTIC OCA 8725 JOHN J KINGMAN RD STE 0944 FORT BELVOIR VA 22060-6218 1 DIRECTOR US ARMY RESEARCH LAB IMNE ALC HRR 2800 POWDER MILL RD ADELPHI MD 20783-1197 1 DIRECTOR US ARMY RESEARCH LAB RDRL CIM L 2800 POWDER MILL RD ADELPHI MD 20783-1197 1 DIRECTOR US ARMY RESEARCH LAB RDRL CIM P 2800 POWDER MILL RD ADELPHI MD 20783-1197 9 DIRECTOR US ARMY RESEARCH LAB RDRL SER PAUL AMIRTHARAJ RDRL SER E ROMEO DEL ROSARIO GREG MITCHELL JAMES WILSON GLEN BIRDWELL ROB REAMS JOHN PENN ED VIVEIROS RDRL SER M ERIC ADLER 2800 POWDER MILL RD ADELPHI MD 20783-1197
NO. OF COPIES ORGANIZATION 1 CERDEC I2WD RDER IWR CI BOB GROSS SUITE D 6240 GUARDIAN GATEWAY APG MD 21005 1 I2WD STEVE HAUGHT CDR, USACERDEC RDER-IWR-CI FT MONMOUTH, NJ 07703 ABERDEEN PROVING GROUND 1 DIR USARL RDRL CIM G (BLDG 4600)