Analysis and Design of a CMOS Mixer

1
Analysis and Designof a CMOS Mixer

• Brian N. Limketkai
• Robert W. Brodersen
• Department of EECS
• University of California
• Berkeley

2
Test Chip
3
Gilbert Cell

• Conversion gain for an ideal square-wave mixing
  function
• Conversion gain for a non-ideal square-wave
  mixing function

4
Low-Power Mixer Issues

• Power is limited by noise
• flicker noise from LO switches
• thermal noise from LO switches
• thermal noise from input differential pair
• thermal noise from load
• Reducing switch size reduces LO power, but
  increases mixer noise
• Flicker noise from LO switches is important in
  direct-conversion systems

5
Output Noise Due to the Switches

• When LO voltage is greater than Vdsat of switches
• LO switch behaves like a cascode transistor
• Minimal output noise contribution

• When LO voltage is less than Vdsat of switches
• LO switches behave like a differential pair
• Maximum output noise contribution

6
Thermal Noise from Switches
For a white noise process N(t) multiplied by a
unit amplitude square wave
where is the duty cycle. For , the
noise spectral density of N(t), we find
This can be simplified to
7
Flicker Noise from Switches

• Flicker noise is negligible at high frequencies,
  thus the output noise contribution is
• Flicker noise is comparable to thermal noise at
  the output for small LO switches because of the
  increase of flicker noise.

8
Distortion of the Differential Pair

• Increase Vdsat
• improves linearity
• improves noise
• reduces headroom
• decrease channel length
• improves linearity
• increases mismatch

9
Interfacing to the Outside World

• Coupling with the PLL
• tradeoff between size of LO switches and overall
  power consumption
• DC input level of LO switches should match DC
  output level of VCO to reduce circuitry
• Coupling with LNA
• capacitance of input differential pair is part of
  the LNAs tank
• Because of the LNAs high output bias, additional
  biasing circuitry required