1. A 24 GHz Low Noise Amplifier: Design, Construction, and Testing
An Undergraduate Senior Design Project
Jeff Keyzer
Tony Long

2. Motivation Building an Amateur Radio 24 GHz Transceiver
Parts above 10 GHz expensive, not usually available through local surplus market
Amateur designs available as references
Senior Design Project Requirement
To go where no undergrad has gone before!

3. Design Requirements > 12dB Gain 2.0 - 2.5dB Noise Figure
Overcome noisy/lossy 1st mixer
For passive mixer, Tsys drops from 7402K to 464K
For active mixer, Tsys drops from 5612K to 393K
Maximum of two stages
Ease of construction - packaged parts a must!!

4. Our Design Tools
Guillermo Gonzalez, Microwave Transistor Amplifiers (Simon & Schuster 1997)
LNA is based on a very simple “by the book” design.
Amateur Radio Microwave Enthusiasts Worldwide
Agilent Technologies Advanced Design Suite 1.3
(Thanks to Dr. Asbeck’s High Speed Devices Group!)
Smith Chart
Many calculations first done on paper, optimized in ADS!

5. FET Selection Choices: Fujitsu FMM5701LG Stanford SPF-1576
NEC NE32584C
NEC NE32984D
Selection Criteria
Packaging
Stability
Cost/Availability
S-Parameters
24 GHz

6. Single Stage LNA Layout

7. Matching Networks Procedure:
1. Plot noise and gain circles for device at 24 GHz.
2. Pick desired input match based on gain/noise tradeoff.
3. Use Smith Chart to match input to 50 using series line and shunt open circuit stub.
4. Optimize line lengths using ADS, correct for capacitance, etc.
5. Find Zout of device with input matching network in place.
(Device is not unilateral!)
6. Use Smith Chart to make conjugate match on output.
7. Optimize in ADS.

8. DC Blocking Coupled Lines
Isolate bias voltages from input/output and between stages
1/4 wave stubs
6 mil width
4 mil separation
Taper reduces reflections

9. Bias Networks
Supply bias to FETs while isolating bias circuits from RF path
Extensive use of /4 impedance transformers
Radial stubs exhibit more ideal performance
C = 1000pF, R = 100 or 1k

10. The Printed Circuit Board
Substrate - 10 mil Rogers Duroid (PTFE) #5880
Conductor mil rolled copper
Needed to be precise
$350 manufacturing cost

11. Test fixtures Required little machining Allow for quick board change
Provide mechanical support for connectors
Allow for flexible connector placement

12. Grounding Plated vias not an option Used “Z” wires
Test fixture provides flat ground plane

13. Connectors Traditional SMA connectors good to 18 GHz
K connectors are expensive and difficult to use
3.5 mm connectors - good through 26.5 GHz

14. Testing 1 or Fighting With Oscillations
First test was to trace oscillations
Not a question of were there oscillations, but where they were.
Addition of Z-wires eliminated oscillations immediately

15. Testing 2 or Fighting With Calibration Standards
3.5mm cal kit is VERY sensitive
Pin offsets of even fractions of a mm are detrimental

16. Noise Figure
TBA.0 dB

17. The Result
Amplifier works! Gives 14 dB gain

18. What’s next? Noise figure measurement
Waveguide transitions rather than coaxial connectors
Integration into receive downconverter
Re-tune for use as a low output power amplifier.