1. Single Balanced Mixer Design ECE 6361
July 30, 2002
Prof. J. S. Kenney
L01: Group1
Dennis Mahoney
Adam Toner
Minsik Ahn

2. Objectives Single-balanced mixer 4-layer PPE printed circuit boards
GHz RF, GHz LO, 140 MHz IF (5 MHz bandwidth)
4-layer PPE printed circuit boards
2.6” by 1.7”
Upconversion and downconversion

4. Description of Design
Balun was realized as a 380 mil long broadside coupled line on layers 1 and 2.
RF was fed into diode through a Wilkinson power divider for balance.
Microstrip stubs were used for reflective terminations.

5. Alternate Design Similar design, but balun outputs on single layer
Simulated conversion loss: ~8 dB
Measured conversion loss: ~10.5 dB
Tapered LO input and moved vias in CAD
LO-RF transmission null centered at 2.8 GHz instead of 2.4 GHz indicating poor balun performance
Poor RF port return loss

6. Simulation
Design, simulation, optimization, and layout in Agilent ADS 2001
Harmonic balance simulation and optimization with seven harmonics
Curve fit diode for n, Rs; Cj from datasheet

7. ADS Simulation Model

8. Wilkinson Power Divider in ADS

9. ADS Simulation Results: Downconverter

10. ADS Simulation Results: Upconverter

11. Mixer Layout

12. Specifications and Results: Downconverter
Actual Results
(Simulated Results)
RF:2400 MHz
RF:2442 MHz
RF:2485 MHz
Specification
Conversion Loss
8.9 dB
(9.325 dB)
9.5 dB
(9.253 dB)
11.2 dB
(9.456 dB)
< 9.5 dB
LO-RF Rejection
29.4 dB
( dB)
33.3 dB
( dB)
38.0 dB
( dB)
< -20 dB
LO-IF Rejection
36.7 dB
( dB )
40.1 dB
( dB)
43.3 dB
( dB)
< -30 dB

13. Specifications and Results: Upconverter Spurious Response
Actual Results
(Simulated Results)
RF:2400 MHz
RF:2442 MHz
RF:2485 MHz
Specification
No Spurs
(No Spurs)
-50 dBc
MHz
-24.3 dBc
at GHz
( dBc at GHz)
-25.4 dBc
at GHz
( dBc at GHz)
-28.4 dBc
at GHz
( dBc at GHz)
-40 dBc
MHz
No spurs
(No spurs)
Only LO
(Only LO)
MHz
-21.9 dBc at GHz
( dBc at GHz)
-22.0 dBc at GHz
( dBc at GHz)
22.2 dBc at GHz
( dBc at GHz)
-30 dBc

14. Measurement Notes
Downconverter conversion loss measured with VNA, accuracy ~0.2 dB
Increased signal generator amplitude to compensate for measured cable loss
Frequency offset mode, calibrated IF output with filter and cables, connected RF directly to test port
Upconverter conversion loss measured with spectrum analyzer, results were better than downconverter but accuracy was ~1 dB
Spurious response measurements with spectrum analyzer set to appropriate RBW and span
Verified IF bandwidth (>> 5 MHz) with swept input, measured output on spectrum analyzer

15. Problems Simulated without vias One via to ground not drilled
Vias greatly increased simulation time
Performance margin built into design
One via to ground not drilled
Repaired by drilling, inserting wire, soldering
Top balun trace not centered
Decreases coupling efficiency, increases conversion loss and spurious products
Trace moved in CAD

16. Problems (continued)
Simulated with straight reactive terminations, fabricated curved stubs
Mixer operating range shifted down in frequency
Conversion loss improved from simulation in low end but increased in high end
Low end spurs increased to out-of-spec from simulation (stub tuning offset down)
Simulated with nominal dielectric constant

17. Improvements
Increase spacing around stubs to reduce parasitic coupling
Reduce loss in RF input bends
Optimize further in ADS
Limited time for first prototype
Agilent diode improved conversion loss in simulation by ~3 dB
Investigate alternate design problems

18. Conclusions Two designs fabricated
One was consistent with simulation
Conversion loss for alternate design was ~3 dB worse than simulation
Primary design met most specifications
Some conversion loss and spurious specs missed
Room for improvement in both designs
Can reduce size and improve performance