mixer dissertation P-II

Linearity Optimization of RF Analog

Down Conversion Mixer Designed

for 2.4GHz Applications

Presented by:

Mr. Shiven Pandya

P.G Student, MEFGI, Rajkot.

Enroll. No. 130570742011

Guided by :

Mr. Amit Kumar

Asst. Professor, MEFGI, Rajkot.

1

Outlines

• Introduction

• Optimization Techniques

• Roadmap to Optimization

• Layout

• Parameters Comparison

• Conclusions

2

What is Mixer?

• Non-Linear Analog device

• Frequency Translator

4

Local

Osc.

LNA Down conversion Mixer

RF

LO

IF

Antenna

Fig.1: RF Front End / Role of mixer in generic receiver

If the inputs are sinusoids,

the ideal mixer output is the sum and difference frequencies given by

Vo = [A1cos(𝜔1𝑡)] [A2cos(𝜔2𝑡)] = 𝐴1𝐴22[cos 𝜔1− 𝜔2 𝑡 + [cos 𝜔1+ 𝜔2 𝑡]

Mixer Output Components

• Nonlinear relation between Input and Output is given by :

Vout = a0 + a1Vin + a2Vin2 + a3Vin

3 + Higher order terms

5

a0 --- DC Biased

a1Vin --- Linear Term (Amplifiers)

a2Vin2 --- Detectors / Mixers

a3Vin3 --- Undesired Spurious Signals (Noise)

Table showing harmonics of fout

Order of

Harmonics

Components of fout

1st order fRF , fLO

2nd order 2fRF , 2fLO

fLO - fRF , fLO + fRF

3rd order 3fRF , 3fLO

2fLO - fRF , 2fLO + fRF ,

fLO - 2fRF , fLO + 2fRF

Pow

er

Frequency

fIF = | fLO – fRF |

IF

RF

LO

Pow

er

Fig.2: Intermediate Frequency at output FFT

3rd Order Input Intercept Point

(Linearity)

• Input / Output relation of the transconductor is

Iout = a0 + a1Vin + a2Vin2 + a3Vin

3 + Higher order terms

6

Two Tones at Input : f1 and f2

Table showing harmonics

Order of Harmonics Components

1st order f1 , f2

2nd order 2f1 , 2f2

f1- f2 , f1 + f2

3rd order 3f1 , 3f2

2f1 – f2 , 2f1 + f2 ,

f1 - 2f2 , f1 + 2f2

Vin

Iout

Fig.3: RF Port

Pow

er

Fundamentals

Frequency

3rd order products

5th order products

7

• Damaging distortion products are odd-order

• 3rd order products will have the highest amplitude

• Defines the dynamic range

Fig. 4: Graph showing IIP3

Input Power(dBm)

Output Power(dBm)

IIP3IM3

Linearity

• Compound FET’s [6]

• Source Degeneration Resistors[10]

• Subharmonics Pumped Technique[11]

Conversion Gain

• Current Bleeding [4]

• Shunt Peaking Tuning[10]

• Current Injection [7]

9

Parameter

Optimization

Technique

Roadmap to Optimization

10

Gilbert Cell

Harmonic Reduction

Negative Feedback

Degenerative Resistor

Current Bleeding

Monte Carlo

Double Balanced Mixer(Gilbert cell)

11

Transconductance stage,

V to I Converter

Switch stage

I to V Converter

Fig.5: Double balanced Gilbert Cell Mixer [1]

12

Output will have: Even harmonics

Fourier series of LO signal

S(t) = 𝑎0

2+ 𝑛=1

∞ 𝑎𝑛 cos(𝑛𝜔𝐿𝑂𝑡 + 𝑏𝑛 sin(𝑛𝜔𝐿𝑂𝑡)

Harmonics as𝑎𝑛 ≠ 0Frequency

Gain

Fig.6: FFT of output signal

Harmonic Reduction

• LO is switching signal

• Operate in Cuttoff & Saturation Region

13

No perfect square wave ??

- Shifting signal upward and downward

- Change in duty cycle

+1

0

Sgn (sinωLO)

Fig.7: sin 𝜔𝐿𝑂𝑡 & sgn(sin 𝜔𝐿𝑂𝑡 )

14

Reduction of even harmonics after applying pulse

Design

Waveforms

Frequency

Frequency

Gain

Gain

With sin 𝜔𝐿𝑂𝑡 as LO

With Sgn(sin(𝑛𝜔𝐿𝑂𝑡))

15

Negative Feedback[1]

Causes of distortion due to intermodulation products

Vin

Iout

Linear Input

Nonlinear Output

Amplitude Distortions

Makes output drive in Saturation or Cuttoff

Frequency Distortions

Frequency dependent effects of reactive components

F(Hz)

GainPoor HF gain

F(Hz)

GainToo much HF gain

Fig.8: Amplitude Distortion

Fig.9: Frequency vs Gain

16

Negative Feedback

• Subtracts a fraction of its output from its input

• Higher Linearity (Reduces Distortions)

• Increases Bandwidth

R

INIF

Fig.10 Negative feedback[1]

Closed loop Gain Afb = AOL

1+ 𝛽AOL

Negative feedback Keeps gain at a constant level

Degenerative Resistors

R R

RF+ RF-

17

Principle of Linearization Reduce the dependence of the gain

of the circuit upon input level

𝐺𝑚 =𝑔𝑚

1 + 𝑔𝑚𝑅For large 𝑔𝑚𝑅 approaches

1

𝑅

Waveforms

Design

Fig.11 : R as Degenerative resistors[11]

18

Current Bleeding Technique

Fig.12 : Current bleeding circuit[5]

LO+ LO-

VDD

V

Driver Current

Pmos is used as a bleeding current source

Higher IP3 than the conventional mixer

Conversion Gain = 4

𝜋𝑔𝑚𝑅𝐿𝑜𝑎𝑑

IIP3 = 32 𝐼𝐷

3𝛽𝑛

For a Gilbert Cell Mixer :

19

Computational algorithms Repeated random sampling

to obtain numerical results

Monte Carlo Simulation

Component Values

WM1-M4, WM9-10 6µm

WM5,M6 0.59µm

WM7,M8 0.61µm

Values after Monte Carlo Simulation

Fig.13 : Gaussian Distribution

Switch

1

Switch

2

Switch

3

Switch

4

Current

Source

Current

Bleeding

Current

Bleeding

Negative

Feedback

Negative

Feedback

Driver/RF Stage

IF+ IF-

RF+ RF-

VDD

Load Load

LO+ LO+

LO-

21

Proposed Block Diagram

Proposed Design

22

Input Output waveforms

Frequencies at the outputWhere 100MHz is the required IF

If sine wave is given as LO input

Fig.15 Output FFT

Fig.14 Input output waveforms

Simulations

23

Fig. 16 : Output FFT will sine wave as LO signal

If sine wave is given as LO input

24

Fig.18 : IIP3 point

Fig.17 : Intermodulation products

Fig.19 : 1dB Compression point

2-Tone set up:Fund. Tone 1 : 2.40Fund. Tone 2 : 2.41

25

• Transistor in linear region behaves as a resistor

Fig. 20 : Proposed circuit with resistors

26

Component W VGATE

Current source

1.5µ 0.5V

50Ω 90µ 0.5V

113Ω 100µ 2.2V

255Ω 75µ 2.2V

1000Ω 20µ 2.2V

Fig. 21 : Transistors replacing resistors

27

Layout

Fig. 22 : Layout of proposed work

28

DRC Report

Prelayout in ADS

29Fig. 23 : Schematic in ADS

Fig. 24 : Symbol of the circuit

30

Converting the 11 pin IC to a 2 pin setup

Layout in ADS

31

Fig. 25 : Nmos and Pmos transistors

33

Fig. 26 : Autogenerated Layout in a ADS

35

Fig. 28 : Final Layout in ADS

Resistors are replaced by GPDK pwell resistors

36

Fig. 29 : Setup for HB 2-tone simulation of schematic

37

Fig. 30 : Setup for HB 2-tone simulation of layout

Comparing IIP3 Point

39

Prelayout

IIP3 = 10dBm

Postlayout

IIP3 = -5dBm

Table : Comparison of Performance Parameter

42

Parameters 1[2] 2[9] 3[10] 4[3] 5[4] 6[6] 7[7] Proposed work

RF(Hz) 2.45G 1.9G 3.1G 1.9G 2.4G 2.4G 1.5G 2.4G

LO(Hz) 3.15G 1.95G

3.3G 2 - 2.65G - 2.5G

IF(Hz) 700M 50M 200M 100M - 250M - 100M

Tech.(µm) 0.18 0.18 0.18 0.5 0.18 0.25 1.2 0.18

Supply(V) 1.8 - 1.8 2.5 - - 1.2 1.8

IIP3(dBm) - 20.45

12.4 2.17 20 - 45.2 13.13

P1dBm -16 12.8 - -8.2 - -8.3 - -9.94

Gain(dB) 8.9 1.6 4.7 3.35 2 6.1 18 9.98

PowerDissipation(W)

9m - - 10m - 5.6m 1.3n 9m

Conclusions• Design Optimized for a Down Conversion Mixer for ISM Band 2.4GHz

application with 0.18 µm CMOS Technology

• Incorporating the degenerative resistors, current bleeding technique and

negative feedback in the standard gilbert cell mixer

• Monte Carlo optimization algorithm gives 3rd order Input Intercept point

(IIP3) at 13.13dBm

• Tools used : Mentor Graphics Tool Suite, Hspice (Synopsys) and ADS.

46

References

1. B. Razavi, RF Microelectronics, New Jersey: Prentice-Hall, 1998.

2. Pokle, “VLSI design of ISM band RF down conversion mixer”, IEEE Thirdinternational conference on emerging trends in engineering and technology, 2010.

3. Kilicaslan, Ismail, “A 1.9 GHz CMOS RF Down Conversion Mixer”, IEEEProceedings of the 40th Midwest Symposium on Circuits and Systems, 1997.

4. Tsai, “Design of 40-80 GHz Low power and High Speed CMOS Down ConversionRing Mixer for Multistand MMW Radio Applications”, IEEE Transactions ofMicrowave Theory and Techniques, Vol.60,No. 3,March 2012.

5. Lee, “ Current-reuse bleeding Mixer”, IEEE Electronics Letters , Vol.36 No.8, April2000.

6. Siddiqi,“2.4 GHz RF Down-conversion Mixers in Standard Cmos Technology”, IEEEJournal of Solid State Circuits, vol. 36, NO. 12, 2004.

47

7. Wei, “A 1.5V High Linearity Down-Conversion Mixer for WiMAX Application”, IEEE

2nd International Conference on Mechanical and Electronics Engineering, vol. 2 , 2010.

8. MacEachern,Manku “A Charge Injection Method for Gilbert Cell Biasing”, IEEE

Canadian Conference on Electrical and Computer Engineering, vol. 1, 1998.

9. Islam,Huq “ High Performance CMOS Converter Design in TSMC 0.18µm Process”,

IEEE Southeast Conference, 2005.

10. Pandram “A Low Power down Conversion CMOS Gilbert Mixer for WirelessCommunications”, Int. Journal of Engineering Research and Applications, Vol. 4, Issue7(Version 1), July 2014.

11. Lu Hung, “A 0.18µm CMOS High Linearity Flat Conversion Gain Down-conversion

Mixer for UWB Receiver”, IEEE Asia Pacific Conference on Circuits and Systems , 2008.

48

Thank You . .

49

50

51

52

Waveforms

53

RF = 2.4 GHz at 0.5 VLO = 2.5 GHz at 1.8VConversion Gain: 4.13 dB

54

Waveforms

55

RF = 2.4 GHz at 0.5 VLO = 2.5 GHz at 1.8VConversion Gain: 9.93 dB

56

57

V4 RF2 0 0.5V5 N007 0 PULSE(0 2.0 0 0 0 0.2ns 0.4ns)M7 OUT2 N003 0 N004 NMOS l=180n w=xM8 OUT1 N002 0 N005 NMOS l=180n w=xR2 N003 OUT2 aR10 OUT1 N002 aR11 N001 OUT2 255R12 N001 OUT1 255

V3 N015 N016 dc=0 z0=50 power=1 $50 Ohm src

+ HB Pin:W 0 1 1 $ tone 1+ HB Pin:W 0 1 2 $ tone 2

.HB tones=2400MEG,2410MEG nharms=3 3 intmodmax=7 Sweep monte = 100 + SWEEP Pin:dBm -50.0 0 2.0

*.print HB P(R12) P(R12)[1,0] P(R12)[2,-1]*.probe HB P(R12) P(R12)[1,0] P(R12)[2,-1]

.OPTION LIMPROBE = 10000