Dung Trinh, PhD HCMUT / 2019

Chapter 2
Noise, Nonlinear Distortion and
System Parameters

Trinh Xuan Dung, PhD

[email protected]

Department of Telecommunications
Faculty of Electrical and Electronics Engineering
Ho Chi Minh city University of Technology
Dept. of Telecoms Engineering 1
Dung Trinh, PhD HCMUT / 2019

Contents

1. General Considerations
2. Nonlinear Distortion in Microwave System
3. Noise in RF/Microwave System

Dept. of Telecoms Engineering 2

Dung Trinh, PhD HCMUT / 2019

1. Units in RF Design
 Units in RF Design:
• The voltage gain and power gain are expressed in decibels (dB):
𝑉𝑜𝑢𝑡
𝐴𝑣 𝑑𝐵 = 20𝑙𝑜𝑔
𝑉𝑖𝑛
𝑃𝑜𝑢𝑡
𝐴𝑃 𝑑𝐵 = 10𝑙𝑜𝑔
𝑃𝑖𝑛
• These two quantities are equal (in dB) only if the input and output voltages appear
across equal impedances. For example, an amplifier having an input resistance of
R0 (50Ω) and driving a load resistance of R0 satisfies the following equations:
2
𝑉𝑜𝑢𝑡
𝑃𝑜𝑢𝑡 𝑅0 𝑉𝑜𝑢𝑡
𝐴𝑃 𝑑𝐵 = 10𝑙𝑜𝑔 = 10𝑙𝑜𝑔 2 = 20𝑙𝑜𝑔 = 𝐴𝑣 𝑑𝐵
𝑃𝑖𝑛 𝑉𝑖𝑛 𝑉𝑖𝑛
𝑅0
• The absolute signal levels are often expressed in dBm rather than watts or volts.
The unit dBm refers to “dBs” above “1mW”. To express the signal power, Psig, in
dBm, we write:
𝑃𝑠𝑖𝑔
𝑃𝑠𝑖𝑔 𝑑𝐵𝑚 = 10𝑙𝑜𝑔
1 𝑚𝑊
Dept. of Telecoms Engineering 3
Dung Trinh, PhD HCMUT / 2019

1. Units in RF Design
 Example 1: An amplifier senses a sinusoidal signal and delivers a power of 0 dBm to a
load resistance of 50Ω. Determine the peak-to-peak voltage swing across the load.
 Solution: 0dBm is equivalent to 1mW, we have:
2
𝑉𝑃𝑃
= 1𝑚𝑊
8𝑅𝐿
𝑉𝑃𝑃 = 632𝑚𝑉.

 Output voltage of the amplifier is of interest in the above example. This may occur if
the circuit following the amplifier does not present a 50Ω input impedance, and hence
the power gain and voltage gain are not equal in dB.
 Only for a sinusoid can we assume that the rms value is equal to the peak-to-peak value
divided by 2 2. Fortunately, for a narrowband 0-dBm signal, it is still possible to
approximate the (average) peak-to-peak swing as 632mV.
Dept. of Telecoms Engineering 4
Dung Trinh, PhD HCMUT / 2019

1. General Considerations
 Example 2: A GSM receiver senses a narrowband (modulated) signal having a level of
-100dBm. If the front-end amplifier provides a voltage gain of 15 dB, calculate the
peak-to-peak voltage swing at the output of the amplifier.
 Solution:
• -100dBm is 100 dB below 632 mVpp.
• 100 dB for voltage quantities is equivalent to 105.
• Thus, -100 dBm is equivalent to 6.32𝜇𝑉𝑃𝑃 .
• This input level is amplified by 15𝑑𝐵 ≈ 5.62, resulting in an output swing of
35.5 𝜇𝑉𝑃𝑃 .

Dept. of Telecoms Engineering 5

Dung Trinh, PhD HCMUT / 2019

1. Nonlinearity
 A system is called “memoryless” or “static” if its output does not depend on the past
values of its input.
 For a memoryless linear system, the input/output characteristic is given by:
𝑦 𝑡 = 𝛼𝑥(𝑡)
 For a memoryless nonlinear system, the input/output characteristic can be approximated
with a polynomial:
𝑦 𝑡 = 𝛼0 + 𝛼1 𝑥 𝑡 + 𝛼2 𝑥 2 𝑡 + +𝛼3 𝑥 3 𝑡 + ⋯
 The following figure shows a common-source stage as an example of a memoryless
nonlinear circuit (at low frequencies). If M1 operates in the saturation region and can be
approximated as a square-law device, then
1
𝑉𝑜𝑢𝑡 = 𝑉𝐷𝐷 − 𝐼𝐷 𝑅𝐷 = 𝑉𝐷𝐷 − 𝑘𝑛 𝑉𝑖𝑛 − 𝑉𝑇𝐻 2 𝑅𝐷
2

 The system has “odd symmetry” if y(t) is an odd function of

x(t). This occurs if 𝛼𝑗 = 0 for even j. Such system is called
“balanced” as exemplified by the differential pair.

Dept. of Telecoms Engineering 6

Dung Trinh, PhD HCMUT / 2019

1. Nonlinearity
 Example 3: The MOS transistors operating in saturation, characteristic can be
expressed as
1 4𝐼𝑆𝑆 2
𝑉𝑜𝑢𝑡 = − 𝑘𝑛 𝑉𝑖𝑛 − 𝑉𝑖𝑛 𝑅𝐷
2 𝑘𝑛
If the differential input is small, approximate the characteristic by a polynomial.
 Solution:
2 4𝐼𝐷𝑆𝑆
• Assuming 𝑉𝑖𝑛 ≪ and applying the approximation 1 − 𝜀 ≈ 1 − 𝜀/2, we have
𝑘
𝑘𝑛 2
𝑉𝑜𝑢𝑡 ≈ − 𝑘𝑛 𝐼𝐷𝑆𝑆 𝑉𝑖𝑛 1 − 𝑉 𝑅
8𝐼𝐷𝑆𝑆 𝑖𝑛 𝐷
3
𝑘𝑛 2 3
𝑉𝑜𝑢𝑡 ≈ − 𝑘𝑛 𝐼𝐷𝑆𝑆 𝑅𝐷 𝑉𝑖𝑛 + 𝑅𝐷 𝑉𝑖𝑛
8 𝐼𝐷𝑆
• The first term on the right-hand side represents linear
operation, revealing the small signal voltage gain of the
circuit (−𝑔𝑚 𝑅𝐷 ).
• Due to symmetry, even-order nonlinear terms are absent.
Dept. of Telecoms Engineering 7
Dung Trinh, PhD HCMUT / 2019

2. Nonlinear Distortion in Microwave Systems

 A system is linear if its output can be expressed as a linear combination
(superposition) of responses to individual inputs:
𝑦1 𝑡 = 𝑓 𝑥1 𝑡
𝑎𝑦1 𝑡 + 𝑏𝑦2 𝑡 = 𝑓 𝑎𝑥1 𝑡 + 𝑏𝑥2 𝑡
𝑦2 𝑡 = 𝑓 𝑥2 𝑡

 For a linear system: 𝑣𝑜 = 𝑎𝑣𝑖

 For a nonlinear system: The

input/output characteristic of a
memory-less nonlinear system can be
approximated with a polynomial:

𝑣𝑜 = 𝑎0 + 𝑎1 𝑣𝑖 + 𝑎2 𝑣𝑖2 + 𝑎3 𝑣𝑖3 + ⋯

Dept. of Telecoms Engineering 8

Dung Trinh, PhD HCMUT / 2019

2. Nonlinear Distortion in Microwave Systems

 The nonlinearity of the system causes the following issues:
 Harmonic generation (multiples of a fundamental signal)
 Gain Compression (gain reduction in an amplifier)
 Inter-modulation Distortion (products of a two-tone input signal)
 Cross-modulation (modulation transfer from one signal to another)
 AM-PM conversion (amplitude variation causes phase shift)
 Spectral regrowth (intermodulation with many closely spaced signals)

Dept. of Telecoms Engineering 9

Dung Trinh, PhD HCMUT / 2019

2A. Nonlinear – Harmonic Distortion

 Input signal of the system:

𝑣𝑖 𝑡 = 𝑉0 𝑐𝑜𝑠𝜔0 𝑡

 Output signal of the system:

2 3
𝑣0 𝑡 = 𝑎0 + 𝑎1 𝑉0 𝑐𝑜𝑠𝜔0 𝑡 + 𝑎2 𝑉0 𝑐𝑜𝑠𝜔0 𝑡 + 𝑎3 𝑉0 𝑐𝑜𝑠𝜔0 𝑡 +⋯
1 3 1 1
= 𝑎0 + 𝑎2 𝑉02 + 𝑎1 𝑉0 + 𝑎3 𝑉03 𝑐𝑜𝑠𝜔0 𝑡 + 𝑎2 𝑉02 𝑐𝑜𝑠2𝜔0 𝑡 + 𝑎3 𝑉03 𝑐𝑜𝑠3𝜔0 𝑡 + ⋯
2 4 2 4

DC Fundamental Second Third

Harmonic Harmonic

 Even-order harmonics result from αj with even j

 nth harmonic grows in proportion to An.

Dept. of Telecoms Engineering 10

Dung Trinh, PhD HCMUT / 2019

2A. Nonlinear – Harmonic Distortion

 Problems:
 More signal loss and distortion.
 Interference to other systems.
 Sometimes can be used to create
frequency multipliers.

GSM900 Band

Dept. of Telecoms Engineering 11

Dung Trinh, PhD HCMUT / 2019

2A. Nonlinear – Harmonic Distortion

 Example 4: An analog multiplier “mixes” its two inputs, ideally producing 𝑦 𝑡 =
𝑘𝑥1 (𝑡)𝑥2 (𝑡) where k is a constant. Assume 𝑥1 𝑡 = 𝐴1 𝑐𝑜𝑠 𝜔1 𝑡 and 𝑥2 𝑡 =
𝐴2 𝑐𝑜𝑠 𝜔2 𝑡 .
a. If the mixer is ideal, determine the output frequency components.
b. If the input port sensing 𝑥2 (𝑡) suffers from third-order nonlinearity, determine the
output frequency components
 Solution:
a. We have: 𝑦 𝑡 = 𝑘𝐴1 𝑐𝑜𝑠 𝜔1 𝑡 𝐴2 𝑐𝑜𝑠 𝜔2 𝑡
𝑘𝐴1 𝐴2 𝑘𝐴1 𝐴2
= 𝑐𝑜𝑠 𝜔1 + 𝜔2 𝑡 + 𝑐𝑜𝑠 𝜔1 − 𝜔2 𝑡
2 2
b. The third harmonic of 𝑥2 𝑡 is 𝛼3 𝐴32 /4, we have:
𝛼3 𝐴32
𝑦 𝑡 = 𝑘𝐴1 𝑐𝑜𝑠 𝜔1 𝑡 𝐴2 𝑐𝑜𝑠 𝜔2 𝑡 + 𝑐𝑜𝑠 3𝜔2 𝑡
4
𝑘𝐴1 𝐴2 𝑘𝐴1 𝐴2
= 𝑐𝑜𝑠 𝜔1 + 𝜔2 𝑡 + 𝑐𝑜𝑠 𝜔1 − 𝜔2 𝑡
2 2
𝑘𝛼3 𝐴1 𝐴32 𝑘𝛼3 𝐴1 𝐴32
+ 𝑐𝑜𝑠 𝜔1 + 3𝜔2 𝑡 + 𝑐𝑜𝑠 𝜔1 − 3𝜔2 𝑡
8 8
Dept. of Telecoms Engineering 12
Dung Trinh, PhD HCMUT / 2019

2A. Nonlinear – Harmonic Distortion

The mixer now produces two “spurious” component at 𝜔1 + 3𝜔2 and 𝜔1 − 3𝜔2 ,
one or both of which often prove problematic. For example, if 𝜔1 = 2𝜋 × 850𝑀𝐻𝑧 and
𝜔2 = 2𝜋 × 900𝑀𝐻𝑧, then 𝜔1 − 3𝜔2 = 2𝜋 × 1850𝑀𝐻𝑧, an undesired component that is
difficult to filter out because it lies close to the desired component at 𝜔1 + 3𝜔2 =
2𝜋 ×1750MHz.

Dept. of Telecoms Engineering 13

Dung Trinh, PhD HCMUT / 2019

2B. Nonlinear – Gain Compression

1 3 1 1
𝑣0 = 𝑎0 + 𝑎2 𝑉02 + 𝑎1 𝑉0 + 𝑎3 𝑉03 𝑐𝑜𝑠𝜔0 𝑡 + 𝑎2 𝑉02 𝑐𝑜𝑠2𝜔0 𝑡 + 𝑎3 𝑉03 𝑐𝑜𝑠3𝜔0 𝑡 + ⋯
2 4 2 4

𝟑
𝒂𝟏 𝑽𝟎 + 𝟒 𝒂𝟑 𝑽𝟑𝟎 𝟑
Fundamental GAIN: 𝑮 = = 𝒂𝟏 + 𝒂𝟑 𝑽𝟐𝟎
𝑽𝟎 𝟒

14
Most RF/microwave
component/circuits
Expansive are compressive ! Compressive
Dept. of Telecoms Engineering
Dung Trinh, PhD HCMUT / 2019

2B. Nonlinear – Gain Compression

 P1dB: power at 1-dB compression.
 Psat: power at saturation.
 If we need very good linearity, we should operate a few dBs below the P1dB point;
this is called power back-off.

𝟑
𝟐𝟎𝒍𝒐𝒈 𝒂𝟏 + 𝒂𝟑 𝑽𝟐𝒊𝒏,𝟏𝒅𝑩 = 𝟐𝟎𝒍𝒐𝒈 𝒂𝟏 − 𝟏
𝟒

𝒂𝟏
𝑽𝒊𝒏,𝟏𝒅𝑩 = 𝟎. 𝟏𝟒𝟓
𝒂𝟑

Dept. of Telecoms Engineering 15

Dung Trinh, PhD HCMUT / 2019

2B. Nonlinear – Gain Compression

 Effect on modulated signal:
 Frequency Modulation:

 Amplitude Modulation:

 FM signal carries no information in its amplitude and hence tolerates compression.

 AM contains information in its amplitude, hence distorted by compression.
 Desensitization: the receiver gain is reduced by the large excursions produced by the
interferer even though the desired signal itself is small.

vi  V1 cos 1t  V2 cos 2t

 3 3 
vo  ...   a1V1  a3V13  a3V23  cos 0t  ...
 4 2 

Dept. of Telecoms Engineering 16

Dung Trinh, PhD HCMUT / 2019

2B. Nonlinear – Gain Compression

 Example 5: A 900-MHz GSM transmitter delivers a power of 1W to the antenna. By
how much must the second harmonic of the signal be suppressed (filtered) so that it
does not desensitize a 1.8-GHz receiver having 𝑃1𝑑𝐵 = −25dBm? Assume the receiver
is 1m away and the 1.8-GHz signal is attenuated by 10 dB as it propagates across this
distance.
 Solution:
The output power at 900MHz is equal to 30dBm. With an attenuation of 10dB, the
second harmonic must not exceed -15dBm at the transmitter antenna so that it is below
𝑃1𝑑𝐵 of the receiver. Thus, the second harmonic must remain at least 45dB below the
fundamental at the TX output. In practice, this interference must be another several dB
lower to ensure the RX does not compress.

Dept. of Telecoms Engineering

Dung Trinh, PhD HCMUT / 2019

2C. Nonlinear – Inter-modulation Distortion

 Single Signal: Harmonic distortion

 Single and one

Desensitization
large interferer:

 Single and two

Intermodulation
large interferers:

Dept. of Telecoms Engineering 18

Dung Trinh, PhD HCMUT / 2019

2C. Nonlinear – Inter-modulation Distortion

 A received small desired signal along with two large interferers
 Intermodulation product falls onto the desired channel, corrupts signal.
interferers

desired

𝟐𝝎𝟐 − 𝝎𝟏 = 𝝎𝟎

Dept. of Telecoms Engineering 19

Dung Trinh, PhD HCMUT / 2019

2C. Nonlinear – Inter-modulation Distortion

 The two interferers at frequency 𝜔1 and 𝜔2 go into the system:

𝑣𝑖 𝑡 = 𝑉0 𝑐𝑜𝑠𝜔1 𝑡 + 𝑐𝑜𝑠𝜔2 𝑡

 The signal at the output of the system:

𝑣0 𝑡 = 𝑎0 + 𝑎1 𝑉0 𝑐𝑜𝑠𝜔1 𝑡 + 𝑐𝑜𝑠𝜔2 𝑡 + 𝑎2 𝑉02 𝑐𝑜𝑠𝜔1 𝑡 + 𝑐𝑜𝑠𝜔2 𝑡 2

+ 𝑎3 𝑉03 𝑐𝑜𝑠𝜔1 𝑡 + 𝑐𝑜𝑠𝜔2 𝑡 3
+⋯

1 1
= 𝑎0 + 𝑎1 𝑉0 𝑐𝑜𝑠𝜔1 𝑡 + 𝑎1 𝑉0 𝑐𝑜𝑠𝜔2 𝑡 + 𝑎2 𝑉02 1 + 𝑐𝑜𝑠2𝜔1 𝑡 + 𝑎2 𝑉02 1 + 𝑐𝑜𝑠2𝜔2 𝑡
2 2
+ 𝑎2 𝑉02 𝑐𝑜𝑠 𝜔1 − 𝜔2 𝑡 + 𝑎2 𝑉02 𝑐𝑜𝑠 𝜔1 + 𝜔2 𝑡

3 1 3 1
+ 𝑎3 𝑉03 𝑐𝑜𝑠𝜔1 𝑡 + 𝑐𝑜𝑠3𝜔1 𝑡 + 𝑎3 𝑉03 𝑐𝑜𝑠𝜔2 𝑡 + 𝑐𝑜𝑠3𝜔2 𝑡
4 4 4 4
3 3 3
+ 𝑎3 𝑉03 𝑐𝑜𝑠𝜔1 𝑡 + 𝑐𝑜𝑠 𝟐𝝎𝟏 − 𝝎𝟐 𝑡 + 𝑐𝑜𝑠 2𝜔1 + 𝜔2 𝑡
2 4 4
3 3 3
+ 𝑎3 𝑉03 𝑐𝑜𝑠𝜔2 𝑡 + 𝑐𝑜𝑠 𝟐𝝎𝟐 − 𝝎𝟏 𝑡 + 𝑐𝑜𝑠 2𝜔2 + 𝜔1 𝑡 + ⋯
2 4 4

Dept. of Telecoms Engineering 20

Dung Trinh, PhD HCMUT / 2019

2C. Nonlinear – Inter-modulation Distortion

Harmonics Formulation

DC 𝑎0 + 𝑎2 𝑉02

3 3
Fundamental 𝑎1 𝑉0 + 𝑎3 𝑉0 𝑐𝑜𝑠𝜔1 𝑡 + 𝑎1 𝑉0 + 𝑎3 𝑉03 𝑐𝑜𝑠𝜔2 𝑡
3
4 4

2nd order IMD 𝑎2 𝑉02 𝑐𝑜𝑠 𝜔1 − 𝜔2 𝑡 + 𝑎2 𝑉02 𝑐𝑜𝑠 𝜔1 + 𝜔2 𝑡

1 1
2nd harmonic 𝑎2 𝑉0 𝑐𝑜𝑠2𝜔1 𝑡 + 𝑎2 𝑉02 𝑐𝑜𝑠2𝜔2 𝑡
2
2 2
3 3
3rd order IMD (IM3) 𝑎3 𝑉0 𝑐𝑜𝑠 2𝜔1 − 𝜔2 𝑡 + 𝑎3 𝑉03 𝑐𝑜𝑠 2𝜔2 − 𝜔1 𝑡
3
4 4
3 3
3rd order IMD (IM3) 𝑎3 𝑉03 𝑐𝑜𝑠 2𝜔1 + 𝜔2 𝑡 + 𝑎3 𝑉03 𝑐𝑜𝑠 2𝜔2 + 𝜔1 𝑡
4 4

Dept. of Telecoms Engineering 21

Dung Trinh, PhD HCMUT / 2019

2C. Nonlinear – Inter-modulation Distortion

Example 6: Suppose four Bluetooth users operate in a room as shown in figure below.
User 4 is in the receive mode and attempts to sense a weak signal transmitted by User 1
at 2.410 GHz. At the same time, Users 2 and 3 transmit at 2.420 GHz and 2.430 GHz,
respectively. Explain what happens.

 Since the frequencies transmitted by Users 1, 2, and 3 happen to be equally spaced,

the intermodulation in the LNA of RX4 corrupts the desired signal at 2.410 GHz.

Dept. of Telecoms Engineering 22

Dung Trinh, PhD HCMUT / 2019

2C. Nonlinear – Inter-modulation Distortion

 IM3 products degrades system performance because they are very close to the
passband of the system
 The power of IM3 products increases more rapidly than the fundamental output
when the input power is increased

3
𝑎3 𝑽𝟑𝟎 𝑐𝑜𝑠 2𝜔1 − 𝜔2 𝑡
4

3
𝑎3 𝑽𝟑𝟎 𝑐𝑜𝑠 2𝜔2 − 𝜔1 𝑡
4

Dept. of Telecoms Engineering 23

Dung Trinh, PhD HCMUT / 2019

2C. Nonlinear – Inter-modulation Distortion

 Third-order intercept point designates the point where the third-order
intermodulation product has the same power as the fundamental output power.
 The input power at the IP3 point is call input-referred IP3: IIP3.
 The output power at the IP3 point is call output-referred IP3: OIP3.

 In practice, the amplifier never

works close to the IP3 point
because it would have
saturated way before that.
 IP3 is extrapolated from
measurements of the
fundamental output and IM3
products in the linear region of
the amplifier.
 IP3 is really a measure of
linearity, rather than
power handling.

Dept. of Telecoms Engineering 24

Dung Trinh, PhD HCMUT / 2019

2C. Nonlinear – Inter-modulation Distortion

 Third Intercept Point: is the point where the output power at 𝜔1 equals to output
power at 2𝜔1 − 𝜔2 .

3 4 𝑎1
𝑎1 𝑉𝐼𝐼𝑃3 = 𝑎3 𝑽𝟑𝑰𝑰𝑷𝟑 → 𝑉𝐼𝐼𝑃3 =
4 3 𝑎3

𝑎1 𝑉 4
𝑉1𝑑𝐵 = 0.145 → 𝐼𝐼𝑃3 = ≃ 9.6𝑑𝐵
𝑎3 𝑉1𝑑𝐵 0.435

Dept. of Telecoms Engineering 25

Dung Trinh, PhD HCMUT / 2019

2C. Nonlinear – Inter-modulation Distortion

 Example 7: A Bluetooth receiver employs a low-noise amplifier having a gain of 10
and an input impedance of 50Ω. The LNA senses a desired signal level of -80dBm at
2.410 GHz and two interferers of equal levels at 2.420 GHz and 2.430 GHz. For
simplicity, assume the LNA drives a 50Ω load.
a. Determine the value of 𝛼3 that yields a 𝑃1𝑑𝐵 of -30dBm.
b. If each interferer is 10 dB below 𝑃1𝑑𝐵 , determine the corruption experienced by the
desired signal at the LNA output.
 Solution:
a. Note that −30𝑑𝐵𝑚 = 20𝑚𝑉𝑃𝑃 = 10𝑚𝑉𝑃 . We have:
𝑎1
𝑉1𝑑𝐵 = 0.145 = 10𝑚𝑉𝑃
𝑎3
• Since 𝛼1 = 10 then 𝛼3 = 14.500 𝑉 −2
b. Each interferer has a level of -40dBm (6.32 m𝑉𝑃𝑃 ). Setting 𝐴1 = 𝐴2 = 6.32 m𝑉𝑃𝑃 ,
we determine the amplitude of the IM product
3𝛼3 𝐴12 𝐴2
= 0.343𝑚𝑉𝑃 = −59.3𝑑𝐵𝑚
4
• The desired signal is amplified by 0 (20dB), produce an output at level -60dBm.
Dept. of Telecoms Engineering 26
Dung Trinh, PhD HCMUT / 2019

2C. Nonlinear – Inter-modulation Distortion

 Input IP is the point where the output power

at 1 equals to output power at (21 - 2 )

Dept. of Telecoms Engineering 27

Dung Trinh, PhD HCMUT / 2019

2C. Nonlinear – Inter-modulation Distortion

Example 9: A low-noise amplifier senses a -80dBm signal at 2.410 GHz and two -
20dBm interferers at 2.420 GHz and 2.430 GHz. What IIP3 is required if the IM
products must remain 20 dB below the signal? For simplicity, assume 50Ω interfaces at
the input and output.
Solution:
 At the LNA output:
3
20𝑙𝑜𝑔10 𝑎1 𝑉𝑠𝑖𝑔 − 20 = 20𝑙𝑜𝑔10 𝑎3 𝑽𝟑𝒊𝒏𝒕
4
 Then:
30
𝑎1 𝑉𝑠𝑖𝑔 = 𝑎3 𝑽𝟑𝒊𝒏𝒕
4

 Thus:

4 𝑎1
𝐼𝐼𝑃3 = = 3.65𝑉𝑃 = 15.2(𝑑𝐵𝑚)
30 𝑎3

Dept. of Telecoms Engineering 28

Dung Trinh, PhD HCMUT / 2019

2D. Nonlinear – Cross Modulation

 Suppose that the interferer is an amplitude-modulated signal:

𝑣𝑖𝑛𝑡𝑒𝑟𝑓𝑒𝑟𝑒𝑟 𝑡 = 𝑉0 1 + 𝑚𝑐𝑜𝑠𝜔𝑚 𝑡 𝑐𝑜𝑠𝜔2 𝑡

 Then:
3
𝑣𝑜 𝑡 = 𝑉0 1 + 𝑚𝑐𝑜𝑠𝜔𝑚 𝑡 𝑐𝑜𝑠𝜔2 𝑡 𝑎1 𝑉0 + 𝑎3 𝑉03 𝑐𝑜𝑠𝜔0 𝑡
4

Dept. of Telecoms Engineering 29

Dung Trinh, PhD HCMUT / 2019

3. Noise in RF/Microwave System

 The effect of noise is critical to the performance of most RF and microwave
communications, radar, and remote sensing systems.
 Noise ultimately determines the threshold for the minimum signal that can be reliably
detected by a receiver.

 Noise power in a receiver will be introduced from:

 The external environment through the receiving antenna – External Noise.
 Internal circuits of receivers – Internal Noise.

Dept. of Telecoms Engineering 30

Dung Trinh, PhD HCMUT / 2019

3. Noise in RF/Microwave System

 External noise may be introduced into a system either by a receiving antenna or by
electromagnetic coupling. Some sources of external RF noise include the following:
 Thermal noise from the ground
 Cosmic background noise from the sky
 Noise from stars (including the sun)
 Lightning
 Gas discharge lamps
 Radio, TV, and cellular stations
 Wireless devices
 Microwave ovens
 Deliberate jamming devices

Noise vs. Interferer

Dept. of Telecoms Engineering 31
Dung Trinh, PhD HCMUT / 2019

3. Noise in RF/Microwave System

 Internal Noise generated in a device or component is usually caused by random
motions of charges or charge carriers in devices and materials. Such motions may be
due to any of several mechanisms, leading to various types of noise:
 Thermal noise is the most basic type of noise, being caused by thermal vibration
of bound charges. It is also known as Johnson or Nyquist noise.
 Shot noise is due to random fluctuations of charge carriers in an electron tube or
solid-state device.
 Flicker noise occurs in solid-state components and vacuum tubes. Flicker noise
power varies inversely with frequency, and so is often called 1/ f -noise.

Dept. of Telecoms Engineering 32

Dung Trinh, PhD HCMUT / 2019

3. Thermal Noise
 A resistor not connected to anything can have a voltage across it.
 The average voltage is of course 0 and its amplitude is random and very small
 Random motion of charge carriers produce a random variation of voltage (or current)
with respect to time  noise voltage and/or noise current.
 A conductor (equivalently a resistor) has lots of charge carriers.
 The higher the temperature, the more random motion, and therefore the higher the
noise.

Higher temperature

Dept. of Telecoms Engineering 33

Dung Trinh, PhD HCMUT / 2019

3. Thermal Noise
 The noisy resistor can be replaced
with a Thevenin equivalent circuit
of a noiseless resistor and a random
noise voltage generator Vn.
4ℎ𝑓𝐵𝑅
 It can be shown that *: 𝑉𝑛2 = ℎ𝑓
h = 6.626 x 10-34 J∙s, Planck’s constant
k = 1.38 x 10-23 J/K, Boltzmann’s constant
𝑒 𝑘𝑇 −1
B, bandwidth in Hz
ℎ𝑓
ℎ𝑓 f, center frequency of the bandwidth
 At low frequencies (f < 5 THz), 𝑒 𝑘𝑇 −1≃ :
𝑘𝑇 R, resistance in Ω
T, temperature in Kelvin
 Noise voltage: 𝑽𝒏 = 𝟒𝒌𝑻𝑩𝑹
𝟐
𝑽𝒏
 Maximum Noise Power: 𝑷𝒏 = 𝑹 = 𝒌𝑻𝑩
𝟐𝑹
𝑷𝒏 𝒅𝑩𝒎 = 𝟏𝟎𝒍𝒐𝒈𝟏𝟎 𝟏. 𝟑𝟖 × 𝟏𝟎−𝟐𝟑 𝑻𝑩 = 𝟏𝟎𝒍𝒐𝒈𝟏𝟎 𝑻𝑩 − 𝟏𝟑𝟖. 𝟓
𝑇 = 290𝐾, 𝐵 = 1𝐻𝑧 → 𝑷𝒏 = −𝟏𝟕𝟒𝒅𝑩𝒎
When: 𝑇 = 290𝐾, 𝐵 = 1𝐾𝐻𝑧 → 𝑷𝒏 = −𝟏𝟒𝟒𝒅𝑩𝒎
𝑇 = 290𝐾, 𝐵 = 1𝑀𝐻𝑧 → 𝑷𝒏 = −𝟏𝟏𝟒𝒅𝑩𝒎
* For derivation, refer to: http://www.pas.rochester.edu/~dmw/ast203/Lectures/Lect_20.pdf
Dept. of Telecoms Engineering 34
Dung Trinh, PhD HCMUT / 2019

3. Thermal Noise
 At most RF and microwave frequencies, the thermal noise from a resistor appear
uniform with respect to frequency, as is evident from the absence of a frequency
variable in the Vn expression.
 We call this type of frequency independent noise “white noise”.

 We also note that the noise power is independent of the resistance!

Dept. of Telecoms Engineering 35

Dung Trinh, PhD HCMUT / 2019

3. Noise Figure
 Noise Figure: measures the degradation in the signal to noise ratio between the input
and output of the component.
𝒅𝒆𝒔𝒊𝒓𝒆𝒅 𝒔𝒊𝒈𝒏𝒂𝒍 𝒍𝒆𝒗𝒆𝒍
Signal to Noise Ratio (SNR): 𝑆𝑁𝑅 =
𝒖𝒏𝒅𝒆𝒔𝒊𝒓𝒆𝒅 𝒏𝒐𝒊𝒔𝒆 𝒍𝒆𝒗𝒆𝒍

 It is important to understand that while amplifiers are great at making your signal
stronger, they also make noise stronger!
 Noise Factor (linear scale):
𝑺𝑵𝑹𝒊𝒏
𝐹=
𝑺𝑵𝑹𝒐𝒖𝒕
 Noise Figure (dB scale): 𝑁𝐹 = 𝟏𝟎𝒍𝒐𝒈𝟏𝟎 𝑭

Dept. of Telecoms Engineering 36

Dung Trinh, PhD HCMUT / 2019

3. Noise Figure
Example 10: Consider a device of bandwidth 𝐵 = 20𝑀𝐻𝑧 and having a gain 𝐺 =
60𝑑𝐵 with noise figure of 𝑁𝐹 = 5𝑑𝐵. What is the noise power seen at the output of the
device?
Solution:

Nout = -35.82dBm.

Dept. of Telecoms Engineering 37

Dung Trinh, PhD HCMUT / 2019

3. Noise Figure
 The noise contributions from subsequent stages are suppressed by the preceding
stages. It follows that the first stage in a cascade system needs to have low noise and
high gain.
 Amplifiers specifically designed to do this are called low noise amplifiers (LNA).
 LNAs are often designed for the lowest noise, sacrificing efficiency, linearity,
and even gain.
 To a large extent, LNAs determine the system sensitivity.
𝑁1 = 𝐺1 𝑘𝑇0 𝐵 + 𝐺1 𝑘𝑇𝑒1 𝐵 𝑁0 = 𝐺2 𝑁1 + 𝐺2 𝑘𝑇𝑒2 𝐵
1
= 𝐺1 𝐺2 𝑘𝐵 𝑇0 + 𝑇𝑒1 + 𝑇
𝐺1 𝑒2
𝑇𝑒2 𝑇𝑒3
𝑇𝐶𝐴𝑆 = 𝑇𝑒1 + + +⋯
𝐺1 𝐺1 𝐺2
𝑇𝑒 = (𝐹 − 1)𝑇0
𝐹2 − 1 𝐹3 − 1
𝐹𝐶𝐴𝑆 = 𝐹1 + + +⋯
𝐺1 𝐺1 𝐺2

Dept. of Telecoms Engineering 38

Dung Trinh, PhD HCMUT / 2019

3. Noise Figure
Example 11: The block diagram of a wireless receiver front-end is shown in following
figure. Compute the overall noise figure of this subsystem. If the input noise power
from a feeding antenna is 𝑁𝑖 = 𝑘𝑇𝐴 𝐵, where 𝑇𝐴 = 150𝐾, find the output noise power in
dBm. If we require a minimum signal-to-noise ratio (SNR) of 20𝑑𝐵 at the output of the
receiver, what is the minimum signal voltage that should be applied at the receiver
input? Assume the system is at temperature 𝑇𝑜 , with a characteristic impedance of 50Ω
and an IF bandwidth of 10MHz.
Solution:

Dept. of Telecoms Engineering 39

Dung Trinh, PhD HCMUT / 2019

3. Noise Figure
 We have:
𝐺𝑎 = 10𝑑𝐵 = 10 𝐺𝑓 = −1𝑑𝐵 = 0.79 𝐺𝑚 = −3𝑑𝐵 = 0.5
𝐹𝑎 = 2𝑑𝐵 = 1.58 𝐹𝑓 = 1𝑑𝐵 = 1.26 𝐹𝑚 = 4𝑑𝐵 = 2.51
 The overall noise figure of the system:
𝐹𝑓 −1 𝐹𝑚 −1 1.26−1 2.51−1
𝐹 = 𝐹𝑎 + + = 1.58 + + = 1.8 = 2.55𝑑𝐵
𝐺𝑎 𝐺𝑎 𝐺𝑓 10 10×0.79

 The equivalent noise temperature of the overall system:

𝑇𝑒 = 𝐹 − 1 𝑇0 = 1.8 − 1 × 290 = 232𝐾
 The output noise power:
𝑁𝑜 = 𝑘 𝑇𝐴 + 𝑇𝑒 𝐵𝐺 = 1.38 × 10−23 150 + 232 × 107 × 3.95 = −96.8𝑑𝐵𝑚
 For an output SNR of 20dB, the input signal power must be:
𝑆𝑜 𝑆𝑜 𝑁𝑜 2.08×10−13
𝑆𝑖 = = = 100 = −82.8𝑑𝐵𝑚
𝐺 𝑁𝑜 𝐺 3.95
 For a 50Ω system impedance, this corresponds to an input signal voltage of
𝑉𝑖 = 𝑍0 𝑆𝑖 = 1.62 × 10−5 𝑉 = 16.2𝜇𝑉

Dept. of Telecoms Engineering 40

Dung Trinh, PhD HCMUT / 2019

3. Noise Figure of a Lossy Element

 Let’s assume that a resistor at the same physical temperature is connected to the input of
the network. Since they in thermal equilibrium, the resistor network can be treated as
one noise source at T, the output noise power should be

𝑁0 = 𝑘𝑇𝐵

 We can also treat the input resistor as the signal source, then the output noise power
should be:
1 1
𝑁0 = 𝑘𝑇𝐵 + 𝑘𝑇𝑒 𝐵
𝐴 𝐴
 Equating the two equation gives: 𝑇𝑒 = 𝐴 − 1 𝑇
 Also we have: 𝑇𝑒 = 𝐹 − 1 𝑇. Finally, 𝐹 = 𝐴.

A = attenuation, defined as Pin/Pout

B, bandwidth in Hz
T, physical temperature (in Kelvin) of the
network

Dept. of Telecoms Engineering 41

Dung Trinh, PhD HCMUT / 2019

3. Noise Figure Measurement

 Cold source method: a termination impedance at room temperature, also called “cold
source”, is placed at the DUT input.
 If the gain of the amplifier is well known over the frequency of interest it can be
subtracted from the output power noise resulting in the amplifier’s excess noise factor F.

𝑁𝑜𝑢𝑡 = 𝑁𝑖𝑛 × 𝐺 × 𝐹

 The gain G of the amplifier can be accurately measured over a bandwidth B by a

calibrated two-port measurement with a VNA.

https://www.youtube.com/watch?v=IgXKS0yJ1-E
Dept. of Telecoms Engineering 42
Dung Trinh, PhD HCMUT / 2019

3. Noise Figure Measurement

 Y factor method: the idea is
simple: measure the noise from
two known noise sources and
compare the noise power
difference at the output.

𝑁1 = 𝐺𝑘𝑇1 𝐵 + 𝐺𝑘𝑇𝑒 𝐵 𝑁1 𝑇1 + 𝑇𝑒 𝑇1 − 𝑌𝑇2

𝑌= = 𝑇𝑒 =
𝑁2 = 𝐺𝑘𝑇2 𝐵 + 𝐺𝑘𝑇𝑒 𝐵 𝑁2 𝑇2 + 𝑇𝑒 𝑌−1

 It seems that taking one measurement is enough to

calculate Te but:
 That would be less accurate; you would have to
know GB precisely.
 Taking two measurements allows you to get rid of
measurement errors that are common to both
measurements.
 You need two or more measurement points to
https://www.youtube.com/watch?v=bbdTRX4_2DE
accurately extrapolate the zero crossing value.
Dept. of Telecoms Engineering 43
Dung Trinh, PhD HCMUT / 2019

3. Noise Figure
Example 12: An X-band amplifier has a gain of 20 dB and a 1 GHz bandwidth. Its
equivalent noise temperature is to be measured via the Y-factor method. The following
data are obtained:
For 𝑇1 = 290𝐾, 𝑁1 = −62𝑑𝐵𝑚
For 𝑇2 = 77𝐾, 𝑁2 = −64.7𝑑𝐵𝑚
Determine the equivalent noise temperature of the amplifier. If the amplifier is used
with a source having an equivalent noise temperature of 𝑇𝑆 = 450𝐾, what is the output
noise power from the amplifier, in dBm?
Solution:

 The Y factor in dB is:

𝑌 = 𝑁1 − 𝑁2 = −62 + 64.7 = 2.7𝑑𝐵 = 1.86
 This gives the equivalent noise temperature as
𝑇1 −𝑌𝑇2 270−1.86×77
𝑇𝑒 = = = 170𝐾
𝑌−1 1.86−1
 If 𝑇𝑆 = 450𝐾, the output noise power from the amplifier is
𝑁𝑜 = 𝐺𝑘𝑇𝑠 𝐵 + 𝐺𝑘𝑇𝑒 𝐵 = 100 × 1.38 × 10−23 × 109 × 450 + 170 = −60.7𝑑𝐵𝑚
Dept. of Telecoms Engineering 44
Dung Trinh, PhD HCMUT / 2019

3. Noise Figure Measurement

 In the OFF or cold state, the noise
source is switched off,
corresponding to a temperature TC
and the noise power at the output is
then:
𝑃𝑜𝑓𝑓 = 𝑘𝑇𝐶 𝐵𝐺𝐹

 With the noise source ON or in hot

state, the excess noise PN of the
noise source is added to the output
noise power:
𝑃𝑜𝑛 = 𝑘𝑇𝐶 𝐵𝐺𝐹 + 𝑃𝑁 𝐺𝐵 = 𝑘𝑇ℎ 𝐵𝐺𝐹
 The Excess Noise Ratio of the used
noise figure is calculated as:
𝑇ℎ − 𝑇𝑜 𝐸𝑁𝑅
𝐸𝑁𝑅 = 𝐹=
𝑇𝑜 𝑌−1
Dept. of Telecoms Engineering 45
Dung Trinh, PhD HCMUT / 2019

3. Noise Figure Measurement

 Would it help if we take more than two
measurements?
 That would require more noise sources:
 Good accurate noise sources can be
expensive
 Getting two noise sources with
drastically different noise temperature
can also be difficult
 Remember that temperature is specified
in Kelvin; 0°C and 100°C are only about
30% different.

 Example: Calculate the Noise Figure NF of your Spectrum Analyzer:

1. Connect a 50Ω termination at the analyzer’s input
2. Select a resolution bandwidth RBW, e.g. RBW=10KHz
3. Place a marker at the center frequency and read out the power level 𝑃𝑁 in
dBm, e.g. 𝑃𝑁 = −110𝑑𝐵𝑚.
Dept. of Telecoms Engineering 46
Dung Trinh, PhD HCMUT / 2019

3. Sensitivity
Sensitivity: Minimum detectable signal by the receiver according to a fixed SNR
determined by the BER.

 To find Receiver NF: Transmit Power

 Transmit Power – FCC regulated.
 Path loss.
Path Loss
 Receiver sensitivity – govern by
standards and applications.
 Required SNR – depends on BER Receiver Sensitivity
requirement and modulation Required SNRo
scheme. Input Noise Floor (No/G)
 Noise floor – thermal noise or
circuit noise limited depending on Noise Figure
the modulation schemes. Input Noise (Ni)

 Example: Mobile network: -100dBm; GPS: -127.5dBm.

Dept. of Telecoms Engineering 47

Dung Trinh, PhD HCMUT / 2019

Q&A

Dept. of Telecoms Engineering 48

Dung Trinh, PhD HCMUT / 2019

Chapter 3
Passive Components, Resonators
and Impedance Matching

Trinh Xuan Dung, PhD

[email protected]

Department of Telecommunications
Faculty of Electrical and Electronics Engineering
Ho Chi Minh city University of Technology
Dept. of Telecoms Engineering 1
Dung Trinh, PhD HCMUT / 2019

Contents

1. High frequency characteristics of passive components

2. RLC Networks and Resonators
3. Impedance Matching
4. Exercises

Dept. of Telecoms Engineering 2

Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Passive Components

 Resistance of a wire above a ground plane:
• Current tends to flow near the external surface of a conductor at high frequencies.
• The current density directed along the axis of a conducting wire is largest at the
surface of the conductor, and falls to small values inside the conductor.
• Most of the current flows within the cylindrical shell within one skin depth from
the surface. The skin depth is denoted by 𝛿[m]:
1
𝛿=
𝜋𝑓𝜇𝜎
where 𝜇, 𝜎 are the permeability and conductivity of the
conductor. 𝜇0 = 4𝜋 × 10−7 (𝐻/𝑚)
• The skin depth in copper at 100 kHz, 10 MHz, and 1 GHz
is approximately 0.2mm, 0.02mm, and0.002 mm.

Dept. of Telecoms Engineering 3

Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Passive Components

 Resistance of a wire above a ground plane:
• At low frequency:
𝑙 𝑙
𝑅𝐷𝐶 = = 2
𝜎𝑆 𝜋𝑟 𝜎
• At higher frequencies, where the skin depth is small compared to the wire radius.
The AC resistance can then be written as:
𝜋𝑟 2 𝑟
𝑅𝐴𝐶 = 𝑅𝐷𝐶 = 𝑅𝐷𝐶
𝜋𝑑𝛿 2𝛿
• Obviously, the wire resistance at high frequency is much greater than its DC
resistance, especially at high frequencies.

Dept. of Telecoms Engineering 4

Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Passive Components

 Example 1: Consider AWG 22 wire. Compute resistance per unit length of the wire
𝑅𝐷𝐶 and 𝑅𝐴𝐶 at 10MHz and 1GHz.
 Solution:
𝑙 𝑙 1
𝑅𝐷𝐶 = = = = 0.053 Ω/𝑚
𝜎𝑆 𝜎𝜋𝑟 2 5.8 × 107 × 𝜋 × (0.643 2)2 × 10−6
• At 10MHz:
1 1
𝛿= = ≃ 21(𝜇𝑚)
𝜋𝑓𝜎𝜇 7 7
𝜋 × 10 × 5.8 × 10 × 4𝜋 × 10 −7

𝑟 0.643 2 × 10−3
𝑅𝐴𝐶 = 𝑅𝐷𝐶 = 0.053 × ≃ 0.4 Ω/𝑚
2𝛿 2 × 23 × 10−6
• At 1GHz: 𝑅𝐴𝐶 ≃ 4 Ω/𝑚

 AWG (American Wire Gauge) 22 gauge wire is a

typical wire used for connections and component leads
which has diameter 𝑑 = 25.3𝑚𝑖𝑙𝑠 or 0.643𝑚𝑚.

Dept. of Telecoms Engineering 5

Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Passive Components

 Inductor of a wire above a ground plane:
• The inductance, per unit length, of a wire with length l, diameter d, and distance h
from a ground plane is (when 𝑑 ≪ 𝑙 and ℎ ≪ 𝑙):
𝜇0 −1
2ℎ
𝐿= cosh
2𝜋 𝑑
𝜇0 4ℎ
• When ℎ 𝑑 > 1 then the following approximation is useful: 𝐿 ≃ ln
2𝜋 𝑑
Note that cosh−1 𝑥 = ln 𝑥 + 𝑥 2 − 1
• For values of ℎ/𝑑 in the range 1 to 100, above equation predicts that L ranges from
2.8nH/cm to 12 nH/cm.
 For short wires, the resistance is often small enough to be ignored, however the inductance of the
wire is often significant. In a circuit where impedances are relatively low, the series impedance of
even a short connecting lead may have a significant impact on circuit performance.
 This leads to a fundamental rule of RF circuit design:
• At high frequencies it is important to keep the length of interconnecting wires and circuit-board
traces short in order to minimize lead inductance.
• When components are separated by significant distances, interconnections must be treated as
distributed circuit elements, and transmission line models are used to model the conductors that
interconnect components.
Dept. of Telecoms Engineering 6
Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Passive Components

 Example 2: A AWG 22 copper wire placed directly on top surface of a PCB with
dielectric thickness of 1.57mm (standard thickness) and a ground plane on the bottom
of the board. Compute the inductance of the wire per unit length, inductive reactance at
100MHz and 1GHz.
 Solution:
• We are given ℎ = 1.57𝑚𝑚 and 𝑑 = 0.643𝑚𝑚. Because ℎ/𝑑 > 1, then:
𝜇0 4ℎ 4𝜋 × 10−7 4 × 1.57
𝐿≃ ln ≃ ln = 0.46(𝜇𝐻/𝑚)
2𝜋 𝑑 2𝜋 0.643
• At 100MHz:
𝑍𝐿 = 𝑗𝜔𝐿 = 𝑗2𝜋 × 108 × 0.46 × 10−6 = 𝑗300(Ω/𝑚)
• At 1GHz:
𝑍𝐿 = 𝑗𝜔𝐿 = 𝑗2𝜋 × 109 × 0.46 × 10−6 = 𝑗3 (𝑘Ω/𝑚)

Dept. of Telecoms Engineering 7

Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Resistors

 Several types of resistors are used in RF circuits, including wire-wound, carbon
composition, thick film, and thin film units:
• Wire-wound resistors consist of a length of lossy wire that is coiled up to fit into a
small package. This type of resistor is seldom used at RF because they have
relatively large inductance.
• Carbon composition resistors consist of a lossy dielectric material sandwiched
between two conducting electrodes.
• Thick or thin-film resistors consist of a film of conducting material deposited on
an insulating substrate. Film resistors are available in cylindrical packages with
attached connecting leads and also as surface mount devices (SMDs). Thick or
thin film resistors in a surface mount package ( aka, “chip” resistors) are the most
common type for RF applications.

A wire wound resistor A carbon composition resistor A Thick film resistor

Dept. of Telecoms Engineering 8
Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Resistors

 Over a fairly wide frequency range, resistors can be modeled using the equivalent
circuit shown in Fig.1.
• The inductor represents the inductance associated with the current path through
the resistor.
• The capacitor represents the capacitance between the two electrodes used to
connect the resistor to external circuitry.
 The inductance is primarily determined by the length of the current path within the
element. The capacitance is determined by the size and separation of the contact
electrodes, as well as the dielectric permittivity of the material between the electrodes.
 Generally, the inductance and capacitance associated with a miniature surface mount
resistor package are on the order of 1 nH and 1 pF, respectively.

Standard equivalent circuit of a resistor

Dept. of Telecoms Engineering 9
Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Resistors

 Significantly higher inductance would be associated with a part in an axial package
with wire leads.
 Even with very short leads, such a package would have a typical inductance value on
the order of 10 nH.
 For many purposes, a simpler (standard) model can be used even for resistors with
external connecting leads if the inductance is taken to be the sum of the resistor
inductance and lead inductance and the shunt capacitance is taken to be the sum of
package capacitance and capacitance between the leads.
 When the resistance is small (≪ 100Ω), and the frequency is not too high, the series
LR brand of the model has a much lower impedance than the capacitance shunt to it.
The capacitive reactance can be neglected.

Equivalent circuit of a resistor with connecting leads Equivalent circuit of a small resistor
Dept. of Telecoms Engineering 10
Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Resistors

 Example 3: Consider a 50Ω resistor with a lead inductance of 𝐿 = 10𝑛𝐻, shunt
capacitor 𝐶 = 1𝑝𝐹. Draw the magnitude of the impedance when the frequency varies
from 1Hz to 1GHz.
 Solution:
• The series inductance acts to increase the impedance at high frequencies.
• The equivalent circuit of a small resistor can be applied to compute the impedance.

Dept. of Telecoms Engineering 11

Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Resistors

 Example 4: Consider a 10𝑘Ω resistor with a lead inductance of 𝐿 = 10𝑛𝐻, shunt
capacitor 𝐶 = 1𝑝𝐹. Draw the magnitude of the impedance when the frequency varies
from 1Hz to 1GHz.
 Solution:
• When the resistance is large (≫ 100Ω), and the frequency is not too high the
inductive reactance will be small compared to R and can be neglected.
• The shunt capacitance is dominant and tends to “short out” the resistance, resulting
in a dramatic reduction in the impedance at high frequencies.

Equivalent circuit of a resistor with

high resistance

Dept. of Telecoms Engineering 12

Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Resistors

 Example 5: Consider a 200Ω resistor with a lead inductance of 𝐿 = 10𝑛𝐻, shunt
capacitor 𝐶 = 1𝑝𝐹. Draw the magnitude of the impedance when the frequency varies
from 1Hz to 1GHz.
 Solution:
• In this case neither of the parasitic reactances is significant compared to 200Ω and
the impedance variation is relatively small up through 1GHz.
• Notice that the impedance of a resistor can be much larger or smaller than the DC
resistance, depending on the resistance value and frequency.

Dept. of Telecoms Engineering 13

Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Capacitors

 Capacitors are constructed by separating two conducting electrodes by an insulating
medium such as air, or a low-loss dielectric material.
• Loss in the dielectric is modeled as a resistance (𝑅𝑝 ) in parallel with the intrinsic
capacitance.
• The inductance associated with the current path through the electrodes and any
connected leads appears in series.
• A resistance in series with the inductor (𝑅𝑠 ) models losses in the electrodes and
leads.
 If 𝑅𝑝 ≫ 1 𝜔𝐶 , the model can be transformed into a series RLC model.
 The sum of the ohmic resistance and the transformed dielectric loss resistance is
termed the equivalent series resistance (ESR). The ESR of a capacitor is dominated by
dielectric loss at sufficiently low frequencies:
1
𝑅𝑒𝑠𝑟 = 𝑅𝑠 +
𝑅𝑝 𝜔 2 𝐶 2

Standard equivalent circuit of a capacitor Equivalent RLC model of a capacitor, valid if 𝑅𝑝 ≫ 1 𝜔𝐶

Dept. of Telecoms Engineering 14
Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Capacitors

 The Q of a capacitor at any frequency is the ratio of the reactance and the ESR
𝑋
𝑄=
𝑅𝑒𝑠𝑟
where the reactance 𝑋 = 𝜔𝐿 − 1 𝜔𝐶 . The dissipation factor, d, is the inverse of Q:
1 𝑅𝑒𝑠𝑟
𝑡𝑎𝑛𝛿 = 𝑑 = 𝑄 = 𝑋
 The dissipation factor is also called the loss tangent.
 All capacitors will have a series resonant frequency, 𝑓𝑠 = 1 2𝜋 𝐿𝐶 . Above this
frequency, the inductive dominates. Hence capacitors are inductive at frequency above
𝑓𝑠 .
 Circuit designers make explicit use of the fact that
capacitor impedance is smallest at and near the series
resonant frequency. When a capacitor is used as a DC
bias circuit decoupling element or as an inter-stage DC-
blocking coupling element, it is sometimes possible to
choose the capacitor so that the intended frequency of
operation falls near the series resonant frequency of the
capacitor, where the impedance is smallest.
Dept. of Telecoms Engineering 15
Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Capacitors

 Example 6: Consider a 0.01𝜇𝐹 capacitor with total lead length of 1𝑐𝑚. Given typical
value of lead inductance is 10nH/cm. Compute resonant frequency of the capacitor.
Draw the capacitor’s impedance and reactance for the frequency range from 1Hz to
100MHz.
 Solution:
• Series resonant frequency of the capacitor is 15.9MHz.

Log amplitude of capacitor’s impedance. Reactance for a 0.01𝜇𝐹 capacitor.

Dept. of Telecoms Engineering 16
Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Inductors

 A reasonably accurate equivalent circuit for an inductor consisting of a coil of wire
wound as a solenoid, around a torus, or in a plane includes
• A series resistance to model the ohmic loss in the wire.
• A shunt capacitance to model distributed capacitance between the turns of the coil
 The inductance of the wire results in voltage differences between the different parts of
the coil. This voltage difference sets up an electric field in the air and in any dielectric
material near the coil.
 The effect of this stored electric energy can be modeled with a capacitance shunted
across the terminals of the coil. This effective capacitance is called the distributed
capacity of the coil.
 The coil resistance can be estimated from the AC resistance of an isolated wire having
the length equal to the total length of wire in the coil.
𝑛𝜋𝑑𝐷
𝑅𝐴𝐶 = 𝑟𝐷𝐶
4𝛿
• d is the wire diameter, D is the coil
diameter, n is the number of turns in the coil
and 𝛿 is the skin depth in the wire. Equivalent circuit of an inductor
Dept. of Telecoms Engineering 17
Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Inductors

 Example 7: Consider a 10𝜇𝐻 inductor with series resistance of 15Ω and distributed
capacitor of 20pF. Compute the resonant frequency of the inductor. Draw the
inductor’s impedance and reactance for the frequency range from 1Hz to 100MHz.
 Solution:

Impedance of the inductor Reactance for the inductor.

Dept. of Telecoms Engineering 18
Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Inductors

 In many cases, an inductor is used to provide a DC bias signal to a circuit, but it is
necessary to isolate the bias supply from the circuit at the operating frequency.
 In this case, an inductor with parallel resonant frequency near the operating frequency
may be employed to provide a low impedance to DC and large impedance to RF
signals. In such an application the inductor element is referred to as an RF choke.

Example use of RFC

Dept. of Telecoms Engineering 19

Dung Trinh, PhD HCMUT / 2019

1. High frequencies Characteristics of Inductors

 When considering a particular inductor for use in a circuit, the designer needs to be
aware of the parallel resonant frequency as well as the “Quality Factor,” or Q, of the
inductor. The Q of an inductor is defined to be the ratio of inductive reactance and
resistance associated with the component
𝑋
𝑄=
𝑅𝑠
 The higher the Q, the better the inductor approximates an “ideal” component. The Q is
an important parameter if the inductor is to be used in a resonant circuit, filter, or
matching network.

The quality factor for the inductor in example 7

Dept. of Telecoms Engineering 20

Dung Trinh, PhD HCMUT / 2019

2. RLC Networks and Resonators

 In RF communication systems, resonant circuits are extensively used to select the
wanted signal and reject the unwanted signal.
 Consider the series RLC circuit in a filter configuration where the output voltage is
taken across the resistor. The voltage transfer function is
𝑉𝑜𝑢𝑡 (𝑠) 𝑅
𝐻 𝑠 = =
𝑉𝑖𝑛 (𝑠) 1
𝑅 + 𝑠𝐿 + 𝑠𝐶
 Consider sinusoidal excitation under steady-state condition, the frequency response
𝑅 𝑅
𝐻 𝑗𝜔 = =
1 1
𝑅 + 𝑗𝜔𝐿 + 𝑅 + 𝑗𝜔 𝐿 − 2
𝑖𝜔𝐶 𝜔 𝐿𝐶
1
 When 𝜔 = , the phase shift of the transfer function is zero, this is called the
𝐿𝐶
“resonant frequency”, 𝜔0 .

Series RLC circuit as a filter

Dept. of Telecoms Engineering 21

Dung Trinh, PhD HCMUT / 2019

2. Series RLC Networks

 The transfer function depends on R,L and C but only two parameters. Define another
quantity 𝑄𝑠 , where
𝜔0 𝐿 1 𝐿
𝑄𝑠 = =
𝑅 𝑅 𝐶
 The frequency transfer function can be written in terms of only 𝜔0 and 𝑄𝑠
𝑅 1
𝐻 𝑗𝜔 == =
1 𝜔 𝜔
𝑅 + 𝑗𝜔 𝐿 − 2 1 + 𝑗𝑄𝑠 𝜔 − 𝜔0
𝜔 𝐿𝐶 0

Magnitude and
phasor of frequency
response in linear
scale

Dept. of Telecoms Engineering 22

Dung Trinh, PhD HCMUT / 2019

2. Series RLC Networks

 The RLC filter has a “bandpass” characteristic. The 3dB bandwidth of the filter is
𝜔0
𝐵𝑊 = Δ𝜔 = 𝜔2 − 𝜔1 =
𝑄𝑠

3dB bandwidth of a filter

Magnitude and
phasor of frequency
response in log scale

Dept. of Telecoms Engineering 23

Dung Trinh, PhD HCMUT / 2019

2. Series RLC Networks

 Example 8: Use a series RLC circuit to couple a voltage source with negligible source
resistance to a 50Ω load as shown in following figure. The circuit should have a center
frequency of 5 MHz and a 3dB bandwidth of 100 kHz. Determine value of L and C.
 Solution:
𝑓0 5𝑀𝐻𝑧
𝑄𝑠 = = = 50
Δ𝑓 100𝑘𝐻𝑧
𝜔𝑜 𝐿 𝑄𝑠 𝑅 50 × 50
𝑄𝑠 = →𝐿= = = 79.6(𝜇𝐻)
𝑅 𝜔𝑜 2𝜋 × 5 × 106
1 1
𝐶= 2 = = 12.7(𝑝𝐹)
𝜔0 𝐿 2𝜋 × 5 × 106 2 × 79.6 × 10−6

Dept. of Telecoms Engineering 24

Dung Trinh, PhD HCMUT / 2019

2. Parallel RLC Networks

 A parallel RLC network being driven by an ideal source as shown in following figure.
The input current and output voltage are related by an impedance function:
−1
𝑉𝑜𝑢𝑡 1 1
𝑍 𝑠 = = + + 𝑠𝐶
𝐼𝑖𝑛 𝑅𝑝 𝑠𝐿
 For sinusoidal steady-state excitation:
−1
1 1 𝑅𝑝
𝑍 𝑗𝜔 = + + 𝑗𝜔𝐶 =
𝑅𝑝 𝑗𝜔𝐿 𝜔 𝜔
1 + 𝑗𝑄𝑝 𝜔 − 𝜔0
0
where:
1
𝜔0 = 𝐿𝐶
𝑅𝑝 𝐶
𝑄𝑝 = 𝜔 = 𝑅𝑝
0 𝐿 𝐿

 The transfer function has exactly the same form as that of series RLC circuit except
the scaling factor 𝑅𝑝 . The 3dB bandwidth:
𝜔0
𝐵𝑊 = Δ𝜔 = 𝜔2 − 𝜔1 =
𝑄𝑝

Dept. of Telecoms Engineering 25

Dung Trinh, PhD HCMUT / 2019

2. Unloaded and Loaded Q of RLC networks

 Compared to the case with infinite source impedance, the finite source impedance
causes the Q to be reduced and, hence, the bandwidth to be increased.
 It is common practice to call the Q of the resonant circuit alone (either series or
parallel RLC) the unloaded Q, and the Q of the composite circuit, which includes the
source resistance and any other resistances that are external to the LC resonator, the
loaded Q.
 The loaded Q is always smaller than the unloaded Q.

Loaded parallel RLC network Loaded series RLC network

𝑅𝑝 ∥ 𝑅𝑠 𝜔0 𝐿
𝑄𝑝 = 𝑄𝑠 =
𝜔0 𝐿 𝑅 + 𝑅𝑠

Dept. of Telecoms Engineering 26

Dung Trinh, PhD HCMUT / 2019

2. Quality Factor Q
 Q describes the frequency selectivity of the Resonators and Circuit characteristics.
 The general definition of Q for a system is:
Maximum instantaneous stored energy
𝑄 = 2𝜋
Energy dissipated per cycle
Maximum instantaneous stored energy
= 2ߨf
Time − average power dissipated
 This definition can be applied to resonant and non-resonant circuits. If this energy
based definition is applied to resonant second-order RLC circuits the result is
compatible with the (𝑄𝑠 , 𝑄𝑝 ) defined in the previous section.
 The energy definition is also commonly applied to characterize lossy inductors or
capacitors which, by themselves, are non-resonant.
 At resonant frequency:
𝑉𝑜𝑢𝑡 = 𝑉𝑖𝑛 = 𝐼𝑖𝑛 𝑅
 The maximum stored energy is:
1 2 1 𝑉𝑖𝑛 2
𝐸𝑚𝑎𝑥 = 𝐿 𝐼𝑖𝑛 = 𝐿 𝑅2
2 2
1 2 1 𝐼𝑖𝑛 2
𝐸𝑚𝑎𝑥 = 2
𝐶 𝑉𝐶 = 2
𝐶 𝜔𝐶2
0

Dept. of Telecoms Engineering 27

Dung Trinh, PhD HCMUT / 2019

2. Quality Factor Q
 The fact that the voltage across the capacitor is 90 degrees out of phase with the
current through the inductor means that the current maximizes at the time when the
capacitor voltage is zero.
 At that instant in time, all of the stored energy resides in the inductor and the
magnitude of the current phasor (which is the peak current magnitude) can be used to
calculate the total stored energy.
 Alternatively, at the time instant when the capacitor voltage is maximum then the
current in the system is zero and all of the stored energy resides in the capacitor. The
magnitude of the capacitor voltage phasor can then be used to calculate the stored
energy at that time.
 The stored energy comes out the same either way. Actually, it can be shown that the
total stored energy in this driven resonant RLC circuit is a constant, so that the
maximum instantaneous stored energy is equal to the energy stored at any instant of
time.

Dept. of Telecoms Engineering 28

Dung Trinh, PhD HCMUT / 2019

2. Quality Factor Q
 The time-averaged power delivered to (and dissipated in) the network is:
1 ∗ 1 𝑉𝑖𝑛 2
𝑃𝑎𝑣𝑔 = ℝ𝑒 𝑉𝑖𝑛 𝐼𝑖𝑛 =
2 2 𝑅
 Using the energy definition of Q:
𝐸𝑚𝑎𝑥 𝜔0 𝐿
𝑄 = 2𝜋𝑓0 =
𝑃𝑎𝑣𝑔 𝑅
 For non-resonant circuit, consider RL circuit:
𝑉𝑖𝑛
𝐼𝑖𝑛 =
𝑅 + 𝑗𝜔𝐿
1 2
1 2
1
𝐸𝑚𝑎𝑥 = 𝐿 𝐼𝑖𝑛 = 𝐿 𝑉𝑖𝑛
2 2 𝑅2 + 𝜔 2 𝐿2
 The time-averaged power delivered is:
1 ∗ 1 1 1 𝑅
𝑃𝑎𝑣𝑔 = 2 ℝ𝑒 𝑉𝑖𝑛 𝐼𝑖𝑛 = 2 𝑉𝑖𝑛 2 ℝ𝑒 𝑅−𝑗𝜔𝐿
= 2 𝑉𝑖𝑛 2
𝑅2 +𝜔2 𝐿2
 Then
𝐸𝑚𝑎𝑥 𝜔0 𝐿
𝑄 = 2𝜋𝑓0 =
𝑃𝑎𝑣𝑔 𝑅
 The concept of component Q is often used to describe the properties of arbitrary circuit
elements at a particular frequency.
Dept. of Telecoms Engineering 29
Dung Trinh, PhD HCMUT / 2019

2. Series to Parallel Transformations

 Any circuit element has both a series and a parallel representation. Since the energy
storage and dissipation properties of the element do not depend on how we represent
it, the Q is the same for either representation.
𝑋𝑠
 The circuit Q is given by: 𝑄 = 𝑅𝑠
 The equivalent parallel resistance and reactance are:
𝑅𝑝 = 𝑅𝑠 1 + 𝑄2
1
𝑋𝑝 = 𝑋𝑠 1 + 2
𝑄
 If 𝑄 ≫ 1, then: 𝑅𝑝 ≃ 𝑅𝑠 𝑄2 and 𝑋𝑝 ≃ 𝑋𝑠 .
 A complex parallel impedance can be transformed to series by inversing the above
equations.

Dept. of Telecoms Engineering 30

Dung Trinh, PhD HCMUT / 2019

2. Series RLC Networks

 Example 9: A 1𝜇𝐻 inductor has a component Q of 100 at 10MHz. Find a parallel
representation for the inductor.
 Solution:
• At 10MHz:
𝑍𝐿 = 𝜔𝐿 = 2𝜋 × 107 × 1 × 10−6 = 62.8 (Ω)
𝜔𝐿
𝑄𝐿 = 100 = → 𝑅 = 0.628 (Ω)
𝑅
• 𝑋𝑝 = 𝑋𝑠 = 62.8(Ω)
• 𝑅𝑝 ≃ 𝑅𝑠 𝑄2 = 6.28(𝑘Ω)

Dept. of Telecoms Engineering 31

Dung Trinh, PhD HCMUT / 2019

3. Impedance Matching Network

 A source with impedance 𝑍𝑆 = 𝑅𝑆 + 𝑗𝑋𝑆 is connected to a load 𝑍𝐿 = 𝑅𝐿 + 𝑗𝑋𝐿 . The
peak voltage of the source is 𝑉𝑆 . The time-averaged real power delivered to the load:
1 1 1 1 𝑅𝐿
𝑃𝐿 = ℝ𝑒 𝑉𝐿 𝐼𝐿∗ = 𝑉𝐿 2 ℝ𝑒 ∗ = 𝑉𝑆 2
2 2 𝑍𝐿 2 𝑍𝐿 + 𝑍𝑆 2
 The above equation can be used to determine what conditions are required to
maximize the power delivered to the load. The problem is then to choose 𝑅𝐿 and 𝑋𝐿 to
maximize 𝑃𝐿 . 𝑃𝐿 is maximized if
𝑍𝐿 = 𝑍𝑆∗
 The maximum power that the source can deliver to an external passive load is referred
to as the power available from the source, 𝑃𝑎𝑣𝑠
𝑉𝑆 2
𝑃𝑎𝑣𝑠 = 𝑃𝐿 𝑍𝐿=𝑍𝑆∗ =
8𝑅𝑆
 The degree to which the actual power delivered to an arbitrary load is smaller than the
available power can be quantified in terms of a mismatch factor (MF)
𝑃𝐿 4𝑅𝑆 𝑅𝐿
𝑀𝐹 = =
𝑃𝑎𝑣𝑠 𝑍𝐿 + 𝑍𝑆 2
 Mismatch Loss: 𝑀𝐿 = −20 log(𝑀𝐹)
Dept. of Telecoms Engineering 32
Dung Trinh, PhD HCMUT / 2019

3. Impedance Matching Network

 Example 10: Suppose a 50Ω signal generator has available power of 1mW. The
generator is to drive a load impedance of 250 + 𝑗100Ω. What is the power delivered
to the load?
 Solution:
𝑃𝐿 4𝑅𝑆 𝑅𝐿 4 × 50 × 250
𝑀𝐹 = = = = 0.5 = 3 (𝑑𝐵)
𝑃𝑎𝑣𝑠 𝑍𝐿 + 𝑍𝑆 2 250 + 𝑗100 + 50 2
𝑃𝐿 = 𝑀𝐹 × 𝑃𝑎𝑣𝑠 = 0.5 × 1𝑚𝑊 = 0.5 𝑚𝑊 = −3 (𝑑𝐵𝑚)

Dept. of Telecoms Engineering 33

Dung Trinh, PhD HCMUT / 2019

3. Impedance Matching Network

 In many applications it is desirable to maximize the transfer of power from the source
to the load. This can be achieved by using a lossless 2-port network inserted between
the source and the load.

 Another way to look at this is to note that the original source plus the matching
network can be viewed as a new source with available power 𝑃𝑎𝑣𝑠 and source
impedance 𝑍𝑆 = 𝑍𝐿∗ . Most lumped-element matching networks are versions of ladder
networks.

Dept. of Telecoms Engineering 34

Dung Trinh, PhD HCMUT / 2019

3. Impedance Matching with lossless L-networks

 Two resistances is to be matched with a lossless L-network. The goal is to transform
𝑅1 to 𝑅2 at one frequency.

 The unknown reactances 𝑋𝑆 and 𝑋𝑃 are easily found with the use of a parallel-to-series
transformation.

 Solving for 𝑋𝑆 and 𝑋𝑃 gives

𝑅2
𝑋𝑃 = ±𝑅1 𝑎𝑛𝑑 𝑋𝑆 = ∓ 𝑅1 𝑅2 − 𝑅22
𝑅1 − 𝑅2
 The solutions yield real values for 𝑋𝑆 and 𝑋𝑃 only if 𝑅1 > 𝑅2 . This leads to a rule for
using a lossless L-network to match two resistances.
The shunt arm of the L-network is connected across the larger of the two resistances.
Dept. of Telecoms Engineering 35
Dung Trinh, PhD HCMUT / 2019

3. Q of an L-network
 The transformed L-network looks like a series resonant circuit. Treating this circuit in
the same manner as a resonant RLC network, the Q of the network is:
𝑋𝑃 𝑅12
𝑅12 + 𝑋𝑃2 𝑅1 𝑅1
𝑄= = = −1
𝑅1 𝑋𝑃2 𝑋𝑃 𝑅2
𝑅12 + 𝑋𝑃2
 This suggests that the BW of the matching network depends only on the ratio 𝑅1 𝑅2 .
 We have defined the Q of the L-network by analogy with the series RLC network. This
analogy works well only when 𝑅1 ≫ 𝑋𝑃 or equivalently, Q ≫ 1.
𝑅 𝑅1
 For moderate or small values of Q the expression 𝑄 = 𝑋1 = − 1 is still valid. But
𝑃 𝑅2
a simple relationship between Q and bandwidth does not exist, since the series
reactance that results from the parallel-to-series transformation has a different
frequency dependence from that of a simple inductor or capacitor.
 Thus an L-network does not behave exactly like a series RLC circuit. Only when the Q
is very large will the equivalent series reactance behave approximately like a capacitor
or inductor.

Dept. of Telecoms Engineering 36

Dung Trinh, PhD HCMUT / 2019

3. Q of an L-network
 Example 11: Match a 100Ω source to a 25Ω load with a lossless L-network. Compare
the power delivered to the load with and without the matching network in place.
Suppose the peak voltage of the source is 1 Volt.

 Solution:
𝑅𝑏𝑖𝑔 100
𝑄= −1= −1= 3
𝑅𝑠𝑚𝑎𝑙𝑙 25
𝑅𝑏𝑖𝑔 100
𝑋𝑃 = 𝑋𝐶 = − =− = −57.7 (Ω)
𝑄 3
𝑋𝑆 = 𝑋𝐿 = 𝑄𝑅𝑠𝑚𝑎𝑙𝑙 = 25 3 = 43.3 (Ω)
• The power delivered to the load without the matching network:
2
1 𝑉𝑜𝑢𝑡 2 1 1
𝑃𝐿 = = × = 0.8 (𝑚𝑊)
2 25 50 5
• With the matching network, the power delivered to the matching network is:
1 1/2 2
𝑃𝑖𝑛 = = 1.25 (𝑚𝑊)
2 100
Dept. of Telecoms Engineering 37
Dung Trinh, PhD HCMUT / 2019

3. Matching complex loads with a lossless L-network

 When complex source and loads are involved, there are two basic conceptual
approaches that can be used:
 Absorption - “absorb” the source or load reactance into the matching network.
 Resonance - series or parallel resonate the source or load reactance at the
frequency of interest.

 Example 12: Match the source and load of the following circuit at 100MHz with a
lossless L-network.

Dept. of Telecoms Engineering 38

Dung Trinh, PhD HCMUT / 2019

3. Matching complex loads with a lossless L-network

 Solution:
• 𝑋𝑆′ and 𝑋𝑃′ can be found by using the design equations:
750
𝑄= − 1 = 2.55
100
𝑋𝑆′ = ±100 × 2.55 = ±255 (Ω)
𝑋𝑃′ = ∓750/2.55 = ∓294.1(Ω)
• The lump reactances can be obtained:
129 (Ω)
𝑋𝑆 = 𝑋𝑆′ − 126 =
−381 (Ω)
250𝑋𝑃′ 1667 (Ω)
𝑋𝑃 = =
250 + 𝑋𝑃′ 135 (Ω)

Dept. of Telecoms Engineering 39

Dung Trinh, PhD HCMUT / 2019

3. Matching complex loads with a lossless L-network

 To apply the resonance concept, the source and load are augmented with reactances
that resonate with the source and load reactances.

 Example 13: Redesign the matching network in Example 12 with resonance method.

Dept. of Telecoms Engineering 40

Dung Trinh, PhD HCMUT / 2019

3. Matching complex loads with a lossless L-network

 Solution:
• An L-network is designed to match these two rea:
258.76
𝑄= − 1 = 1.57
75
𝑋𝑆′ = ±100 × 1.57 = ±117.8 (Ω)
𝑋𝑃′ = ∓258.76/1.57 = ∓164.8(Ω)
• The solution of the design is as follows

Dept. of Telecoms Engineering 41

Dung Trinh, PhD HCMUT / 2019

3. Three-element Matching Networks

 The L-network does not give the designer freedom to choose the Q (bandwidth) or
phase shift of the matching network.
 The addition of a third matching element makes it possible to design for a match and a
specified phase shift or Q. The Pi- and T-networks can be used to design matching
circuits with specific bandwidth (Q).
 We consider a Pi-network:

 The Pi-network can be thought of as two back-to-back L-networks that act to match
both 𝑅1 and 𝑅2 to a “virtual resistance” 𝑅𝑣 .

Dept. of Telecoms Engineering 42

Dung Trinh, PhD HCMUT / 2019

3. Pi Matching Networks
 Because the series arms of both L-networks are connected to 𝑅𝑣 , it is clear that 𝑅𝑣 is
smaller than 𝑅1 and 𝑅2 . Define the Q’s of the two L-networks to be 𝑄1 and 𝑄2 :
𝑅1 𝑅2
𝑄1 = −1 and 𝑄2 = −1
𝑅𝑣 𝑅𝑣

 Assume 𝑅2 > 𝑅1 , therefore 𝑄2 > 𝑄1 . For most practical purposes the Q of the Pi-
network can be approximated by 𝑄2 . This is especially true if 𝑅2 ≫ 𝑅1 .
 The design process is as follows:
 Determine the required Q of the matching network. This Q is taken to be equal to
𝑄2 , and thus the virtual resistance, 𝑅𝑣 , is determined.
 Once 𝑅𝑣 is found, the values of 𝑋𝐶1 , 𝑋𝐿1 , 𝑋𝐿2 and 𝑋𝐶2 can be calculated:
𝑅 𝑅1 𝑅2
𝑋𝐶2 = − 𝑄2 𝑋𝐶1 = − 𝑅
2 𝑄22 +1 −𝑅2
1

𝑅1 2
𝑅2 𝑄2 + 𝑅2 𝑅2 𝑄2 + 1 − 1
𝑋𝐿 =
𝑄22 + 1

Dept. of Telecoms Engineering 43

Dung Trinh, PhD HCMUT / 2019

3. T-Matching Networks
 The Pi-network is most useful for matching when the values of 𝑅1 and 𝑅2 are not too
small. If 𝑅1 and 𝑅2 are small, the virtual resistance will be even smaller, and the
capacitor values will turn out to be impractically large.
 If either terminating resistance is significantly less than 50Ω, the T-network will
usually be a more practical choice.

 The Q’s of the two networks are:

𝑅1 𝑅2
𝑄1 = −1 and 𝑄2 = −1
𝑅𝑣 𝑅𝑣

 The design formulas are: 𝑋𝐶1 = −𝑅1 𝑄1

𝑅1 2
𝑅1 (𝑄12 + 1)
𝑋𝐿2 = 𝑅2 𝑋𝐿3 =
𝑅2 𝑄1 + 1 − 1 𝑅1 2
𝑄1 − 𝑅2 𝑄1 + 1 − 1
Dept. of Telecoms Engineering 44
Dung Trinh, PhD HCMUT / 2019

4. Exercises
Exercise 1: You are given a black box with two terminals. Suppose that you know that the
box contains a passive circuit that is constructed from 3 elements: a resistor (R), lossless
capacitor (C), and lossless inductor (L). Your task is to figure out how the elements are
connected, and what their values are. You make some measurements of the impedance of
the box 𝑍(𝑓) = 𝑅(𝑓) + 𝑗𝑋(𝑓). The results of the measurements are shown in following
figures.
a. Sketch the circuit that is inside of the box. [R in parallel with C, all in series with L]
b. Estimate the element values. [𝑅 = 1𝑘Ω , 𝐶 = 24.36𝑝𝐹, 𝐿 = 7.96𝜇𝐻].

Dept. of Telecoms Engineering 45

Dung Trinh, PhD HCMUT / 2019

4. Exercises
Exercise 2: You are given a black box with two terminals. Suppose that you know that the
box contains a passive circuit that is constructed from 3 elements: a resistor (R), lossless
capacitor (C), and lossless inductor (L). Your task is to figure out how the elements are
connected, and what their values are. You make some measurements of the impedance of
the box 𝑍(𝑓) = 𝑅(𝑓) + 𝑗𝑋(𝑓). The results of the measurements are shown in following
figures.
a. Sketch the circuit that is inside of the box..
b. Estimate the element values.

Dept. of Telecoms Engineering 46

Dung Trinh, PhD HCMUT / 2019

4. Exercises
Exercise 3: The circuit shown in following figure is used as a model for a realistic resistor
or inductor. It can also be used to model a realistic parallel resonant circuit.
a. Find an expression for the impedance of this circuit.
b. The frequency at which the impedance is purely resistive is called the “resonant
frequency.” Under what conditions will there be a frequency (> 0) at which the impedance
is purely resistive? Show that if this condition is satisfied, and if 𝐶𝑅2 /𝐿 ≪ 1, then the
resonant frequency is given approximately by:
1 𝐶𝑅2
𝜔0 = 1−
𝐿𝐶 2𝐿
c. Assume that the term 𝐶𝑅2 /𝐿 can be ignored and find the magnitude of the impedance
at the parallel resonant frequency. How does it depend on R?
1 𝐿
d. Find an approximate expression for the impedance valid when 𝜔 ≪ and 𝑅2 𝐶 ≫ 1.
𝐿𝐶
Draw a simplified equivalent circuit that is valid under these conditions.
1 𝐿
e. Find an approximate expression for the impedance valid when 𝜔 ≪ and ≪ 1.
𝐿𝐶 𝑅2 𝐶
Draw a simplified equivalent circuit that is valid under these conditions.

Dept. of Telecoms Engineering 47

Dung Trinh, PhD HCMUT / 2019

4. Exercises
Exercise 4: The circuit shown in following figure is usually a good model for a parallel
LC circuit implemented with real, lossy components. The element values are 𝐶 = 800𝑝𝐹,
𝐿 = 15𝜇𝐻, 𝑟 = 1Ω, 𝑅 = 10𝑘Ω. Use series to parallel transformations to transform this
circuit into an equivalent parallel RLC circuit and find the approximate resonant
frequency and 𝑄𝑃 of the equivalent circuit.

Exercise 5: The circuit in following figure is a model for the operation of a “ferrite
loopstick” antenna that is commonly employed in AM broadcast band radios. The antenna
is tuned to resonance by a capacitor C which is usually adjustable to allow the circuit to
cover the entire broadcast band. This circuit also performs the function of the pre-selector.
The voltage source 𝑉𝑠 represents the emf induced in the coil as a result of an incident
electromagnetic wave with frequency 𝜔. The resistance 𝑟 represents the losses in the coil
and 𝑅 represents the input impedance of the following stage.

Dept. of Telecoms Engineering 48

Dung Trinh, PhD HCMUT / 2019

4. Exercises
Exercise 5 (cont):
a. Find an exact expression in terms of r, R, L, and C for the frequency where the output
voltage is a maximum (the resonance frequency).
b. Suppose that 𝑉𝑠 = 1(𝑚𝑉) cos 2𝜋𝑓𝑐 𝑡 , 𝐿 = 100𝜇𝐻, 𝑅 = 100𝑘Ω, and 𝑟 = 6Ω. Find
the range of values that C must cover in order for the circuit to tune the AM broadcast
band (540-1700 kHz). Note: For this purpose you need an expression for the value of C
that maximizes the voltage response at a given frequency. An approximate analysis is
acceptable, but be sure to carefully state and justify your assumptions.
c. Now suppose that the source frequency, 𝑓𝑠 , is swept from 540 to 1700 kHz. The zero-
to-peak value of the source voltage is held constant at 1mV as the frequency is swept. Also
suppose that the circuit is tuned to follow the frequency of the source so that the output
voltage is always maximized. Plot the output voltage as a function of frequency.

Dept. of Telecoms Engineering 49

Dung Trinh, PhD HCMUT / 2019

4. Exercises
Exercise 6: Consider the circuit shown in following figure. The current source has
constant amplitude and frequency 𝑓𝐶 , and it drives a bandpass filter consisting of a lossy
inductor in parallel with a variable capacitor and a resistor. You may assume that any
capacitance associated with the inductor has been incorporated into the variable capacitor
indicated in the schematic. The variable capacitor C can be set to any value in the range 36
to 365pF, and 𝑟 = 10Ω, 𝑅 = 100𝑘Ω. Suppose that the frequency of the current source can
be adjusted to any frequency in the range 540-1700 kHz. For a given value of the source
frequency, 𝑓𝐶 , the variable capacitor will be tuned to maximize the output voltage.
Approximate the filter as a parallel RLC circuit to answer the following questions:
a. Specify a single value of L that would allow the variable capacitor to tune the filter
(i.e. maximize the output voltage) to any frequency in the range 540-1700kHz.
b. With the value of L that was determined in part (a), determine the approximate 3dB
bandwidth of the filter when 𝑓𝐶 = 540𝑘𝐻𝑧 and C is adjusted to maximize the output
voltage.
c. Same as part (b) but for 𝑓𝐶 = 1700𝑘𝐻𝑧.

Dept. of Telecoms Engineering 50

Dung Trinh, PhD HCMUT / 2019

4. Exercises
Exercise 7: Consider the following circuit and its Thevenin equivalent circuit.
a. If 𝑉𝑠 = 2𝑉, 𝑅 = 1𝑘Ω, 𝐶 = 1𝑛𝐹, 𝐿 = 10𝜇𝐻, 𝑟 = 1Ω, find the (non-zero) frequency,
𝜔0 , where the indicated Thevenin equivalent circuit would be valid. In other words, find
the frequency where the Thevenin impedance is purely resistive. Clearly state any
assumptions or approximations that you make.
b. Find 𝑅𝑡ℎ and 𝑉𝑡ℎ at the frequency found in part (a). For 𝑉𝑡ℎ , specify magnitude and
phase (in degrees).

Dept. of Telecoms Engineering 51

Dung Trinh, PhD HCMUT / 2019

4. Exercises
Exercise 8: The circuit shown in following figure is a capacitive transformer with
resonating inductance L. Suppose that 𝑅 = 50Ω, 𝐶1 = 3183𝑝𝐹, 𝐶2 = 3183𝑝. Use
parallel-series and series-parallel transformations, with appropriate approximations, and
find:
a. The inductance, L, required to resonate the circuit at 10 MHz. At resonance, the
impedance Z will be purely real.
b. The input impedance, Z at 10 MHz.
c. The Q of the circuit can be approximated by the 𝑄𝑃 of the equivalent parallel RLC
circuit (at 10 MHz). Find the approximate Q.
To save time, note that the reactances of 𝐶1 and 𝐶2 have magnitude 5Ω at 10MHz.

Dept. of Telecoms Engineering 52

Dung Trinh, PhD HCMUT / 2019

4. Exercises
Exercise 9: You are given a “black box” with two output terminals (a “1-port"). You play
around with the box for awhile and make the following observations:
a. The output voltage from the box is sinusoidal.
b. The peak magnitude of the open circuit voltage at the output is found to be 5V.
c. You connect a 50Ω resistor across the terminals and find that the peak magnitude of
the voltage across the resistor is 2.795V.
d. You short the output of the box and find the peak magnitude of the short circuit
current is 100 mA.
Find the power available from the source. Express your result in dBm.

Dept. of Telecoms Engineering 53

Dung Trinh, PhD HCMUT / 2019

4. Exercises
Exercise 10: Consider the design of a lossless L-network to match the source and load
shown in following figure. All resistances and reactances are in ohms, and the current
source magnitude is the peak value.
a. Find the power available from the source. Express your result in dBm.
b. How much power would be delivered to the load if a lossless matching network was
not used, i.e., if the load is connected directly to the source? Express your result in dBm.
c. There are four possible solutions for the matching network if an L-network is used.
Find all four. Sketch all solutions and indicate whether the elements are inductors or
capacitors.
d. Verify two of your designs by plotting the path from the load to the source on a Smith
Chart.

Dept. of Telecoms Engineering 54

Dung Trinh, PhD HCMUT / 2019

4. Exercises
Exercise 11: Consider the source and load shown in the following figure.
a. Find the power available from the source. Express your answer in dBm.
b. Match this source and load using a lossless L-network.
c. Now suppose that you can only use a single lossless inductor or capacitor, either in
series or in shunt, to couple the source to the load. If the goal is to maximize the power
delivered to the load under this constraint, find the best possible solution.

Exercise 12: A 50Ω source has 𝑃𝑎𝑣𝑠 = 4𝑚𝑊. It is necessary to couple the source to a load
with impedance 𝑍𝐿 = 2 = 𝑗20Ω.
a. Find the power delivered to the load when the load is connected directly to the source.
Express your answer in dBm.
b. There are 4 lossless LC L-networks that will match this source and load. Find the
solution that has a capacitive series arm and the shunt arm connected across the load.
Dept. of Telecoms Engineering 55
Dung Trinh, PhD HCMUT / 2019

Q&A

Dept. of Telecoms Engineering 56

Dung Trinh, PhD HCMUT / 2019

Chapter 4
RF Oscillators and Frequency
Synthesizers

Trinh Xuan Dung, PhD

[email protected]

Department of Telecommunications
Faculty of Electrical and Electronics Engineering
Ho Chi Minh city University of Technology
Dept. of Telecoms Engineering 1
Dung Trinh, PhD HCMUT / 2019

Contents

1. Introduction
2. LC Oscillators
3. RC Oscillators
4. Crystal Oscillator
5. Voltage Controlled Oscillator and Phase Locked Loop
6. Exercises

Dept. of Telecoms Engineering 2

Dung Trinh, PhD HCMUT / 2019

1. Introduction
 Oscillators convert a DC input (the supply voltage) into an AC RF
output (the waveform), which can have a wide range of different
wave shapes and frequencies.
 Oscillators are used in a number of applications in which a reference tone is required.
For instance, they can be used as the clock for digital circuits or as the source of the
LO signal in transmitters.
 In receivers, oscillator waveforms are used as the reference frequency to mix down
the received RF to an IF or to baseband. In most RF applications, sinusoidal
references with a high degree of spectral purity (low phase noise) are required.
 Communications systems, digital systems (including computers), and test equipment
make use of oscillators.

Waveform generated by oscillators

Dept. of Telecoms Engineering 3
Dung Trinh, PhD HCMUT / 2019

1. Introduction
 Classification of sinusoidal Oscillators:
 RC Oscillators: Wien Oscillator, Phase Shift Oscillator.
• Use a resistance-capacitance network to determine the oscillator frequency.
• Suitable for low and moderate frequency applications (5Hz to 1MHz).
 LC Oscillators: Colpitts, Hartley Oscillator.
• Use a inductance-capacitance network to determine the oscillator frequency.
• Suitable for radio frequency (1 to 500MHz) applications.
 Crystal Oscillators:
• Use piezoelectric crystal (or quartz) which has very high degree of stability
and accuracy. Suitable for radio frequency applications.
 Microwave Oscillators: Gunn diode, cavity tuned, dielectric.

Example of RC oscillators Example of LC oscillators Example of crystal oscillators

Dept. of Telecoms Engineering 4
Dung Trinh, PhD HCMUT / 2019

1. Conditions for Oscillation

 An oscillator can be seen as unstable feedback systems. An unstable system is one in
which an initially small excitation or disturbance produces an output that grows in
time due to constructive, or positive, feedback.
 We have:
𝑉𝜀 (𝑠) = 𝑉𝑖 (𝑠) + 𝛽(𝑠)𝑉𝑜 (𝑠)
𝑉0 (𝑠) = 𝐴(𝑠)𝑉𝜀 (𝑠)
 Then:
𝑉𝑜 (𝑠) 𝐴(𝑠)
𝐴𝑓 𝑠 = =
𝑉𝑖 (𝑠) 1 − 𝛽 𝑠 𝐴(𝑠)
where T(s) = 𝛽 𝑠 𝐴(𝑠) is known as open loop gain or sometimes shortened to loop gain.
 Replace 𝑠 with 𝑗𝜔:
𝐴(𝑗𝜔)
𝐴𝑓 𝑗𝜔 =
1 − 𝑇(𝑗𝜔)

Block diagram of a feedback system

Dept. of Telecoms Engineering 5
Dung Trinh, PhD HCMUT / 2019

1. Conditions for Oscillation

 Note that the loop gain is the gain obtained by opening the feedback loop and taking
𝑉
the output at the point where the loop was opened ( 𝑓 𝑉𝑖).
 Suppose that the loop gain is equal to 1 at some frequency 𝜔0 , 𝑇 𝜔0 = 1. Then the
voltage transfer function is singular (infinite) which can be interpreted as finite output
for zero input.
 In other words, the circuit is a potential source of radio frequency energy at the
frequency where the loop gain is 1, even in the absence of any input excitation 𝑉𝑖 .
 Condition for the feedback loop to provide sinusoidal oscillations of frequency 𝜔0
(Barkhausen Criterion) is
• At 𝜔0 the phase of the loop gain should be zero arg 𝑇(𝜔0 ) = 0.
• At 𝜔0 the magnitude of the loop gain should be unity 𝑇(𝜔0 ) = 1.

Visualization of open loop gain

Dept. of Telecoms Engineering 6
Dung Trinh, PhD HCMUT / 2019

1. Conditions for Oscillation

 A negative-feedback oscillatory system can be built using “LC-tuned” amplifier
stages.

 At low frequencies, L1  At the resonant  At high frequencies, C1

dominates the load frequency dominates the load
𝑉𝑜𝑢𝑡 𝑉𝑜𝑢𝑡 𝑉𝑜𝑢𝑡
= −𝑔𝑚 𝑗𝜔𝐿1 = −𝑔𝑚 𝑅𝑃 = −𝑔𝑚 /𝑗𝜔𝐶1
𝑉𝑖𝑛 𝑉𝑖𝑛 𝑉𝑖𝑛
𝑉𝑜𝑢𝑡
•
𝑉𝑜𝑢𝑡
is very small and • The phase shift from the • dimished and
𝑉𝑖𝑛 𝑉𝑖𝑛
input to the output is 𝑉
𝑉
∠ 𝑉𝑜𝑢𝑡 ≃ −900 𝑉 ∠ 𝑉𝑜𝑢𝑡 ≃ 900
𝑖𝑛 ∠ 𝑉𝑜𝑢𝑡 = 1800 𝑖𝑛
𝑖𝑛

Dept. of Telecoms Engineering 7

Dung Trinh, PhD HCMUT / 2019

1. Conditions for Oscillation

 The circuit provides a phase shift of 1800 with possibly adequate gain 𝑔𝑚 𝑅𝑃 at 𝜔0 .
We simply need to increase the phase shift to 3600.

 Wave form is shown above (right). A unique attribute of inductive loads is that they
can provide peak voltages above the supply. The growth of VX and VY ceases when
M1 and M2 enter the triode region for part of the period, reducing the loop gain.

Dept. of Telecoms Engineering 8

Dung Trinh, PhD HCMUT / 2019

1. Conditions for Oscillation

 In practice the condition arg 𝑇(𝜔0 ) = 0 is usually applied to solve for the potential
frequency of oscillation, 𝜔0 . Then the condition 𝑇(𝜔0 ) = 1 is applied to determine
the gain of amplifier.
 For practical designs, the amplifier gain is set to a value somewhat higher than the
value determined by above equation to ensure that oscillations will reliably start and
grow when the circuit is powered up.
 Circuits with the topology shown in following figure are commonly employed as
oscillators. The active device could be a BJT or an FET.

Topology for one class of oscillator circuits

Dept. of Telecoms Engineering 9

Dung Trinh, PhD HCMUT / 2019

1. Conditions for Oscillation

 The passive elements of the model (e.g., 𝑟𝜋 , 𝐶𝜋 , 𝐶𝜇 etc.) can be lumped into the
external impedances 𝑍1 , 𝑍2 , 𝑍3 .
 To find the open loop gain of this circuit, we break the loop at a convenient point and
terminate that point in the impedance that it sees when the loop is closed.
 The termination is required so that the open-loop circuit operates under the same
conditions as when the loop is closed. The loop gain is computed by exciting the
circuit at the input to the opened loop and computing the output across 𝑍1:
𝑉 𝑍 𝑍 +𝑍
𝐴 = 𝑉 𝑜 = −𝑔𝑚 𝑍2 𝑍1 + 𝑍3 = −𝑔𝑚 𝑍2 +𝑍1 +𝑍3
𝑖𝑛 1 2 3
𝑉𝑓 𝑍1
𝛽= =𝑍
𝑉𝑜 1 +𝑍3
𝑉𝑓 𝑉𝑜𝑢𝑡 𝑍1 𝑍2
𝑇=𝑉 = = 𝐴𝛽 = −𝑔𝑚 𝑍
𝑖𝑛 𝑉𝑖𝑛 1 +𝑍2 +𝑍3

Feedback loop terminated with Z1.

Dept. of Telecoms Engineering 10

Dung Trinh, PhD HCMUT / 2019

2. LC Oscillator Analysis using Loop Gain

 We assume for the moment that 𝑍1 and 𝑍2 are purely reactive, i.e., 𝑍1 = 𝑗𝑋1 , 𝑍2 =
𝑗𝑋2 . We allow 𝑍3 to have a non-zero (positive) real part. Then:
𝑔𝑚 𝑋1 𝑋2
𝑇=
𝑗(𝑋1 + 𝑋2 ) + 𝑍3
 For oscillation to occur, the phase angle of 𝑇 must be zero at some frequency. Now
there are two possibilities for the signs of 𝑋1 and 𝑋2 :
• 𝑋1 > 0 (inductor), 𝑋2 > 0 (inductor): Hartley Oscillator
• 𝑋1 < 0 (capacitor), 𝑋2 < 0 (capacitor): Colpitts Oscillator
 The frequency of oscillation is easily found by finding the frequency where the phase
angle of 𝑍𝑠 = 𝑗 𝑋1 + 𝑋2 + 𝑍3 = 0. This leads to
1 1
• Colpitts: 𝜔0 = Hartley: 𝜔0 =
𝐶 𝐶
𝐿𝐶 1+𝐶2 𝐶(𝐿1 +𝐿2 )
1 2

Colpitts Oscillator Hartley Oscillator

Dept. of Telecoms Engineering 11
Dung Trinh, PhD HCMUT / 2019

2. LC Oscillator Analysis using Loop Gain

 In fact, if 𝑍3 was lossless, then 𝑍1, 𝑍2 and 𝑍3 would constitute a lossless resonant
circuit. A lossless circuit will oscillate indefinitely at the resonant frequency once it is
initially excited.
 When 𝑍3 is lossy, then the oscillations cannot be maintained unless the transistor is
added to make up for the losses.
 So far we have considered only half of the Barkhausen criterion (arg[𝑇] = 0). The
results yielded the potential frequency of oscillation. For oscillations to occur, the
gain of the transistor must be large enough to make 𝑇 = 1 𝜔=𝜔0 . This leads to:
𝑔𝑚 𝑅 𝐶1 +𝐶2
• Colpitts: = 1 → 𝑔𝑚 = 𝑅𝜔02 𝐶1 𝐶2 = .
𝑅𝜔02 𝐶1 𝐶2 𝐿
𝑔𝑚 𝜔02 𝐿1 𝐿2 𝑅 𝑅𝐶 𝐿1 +𝐿2
• Hartley: = 1 → 𝑔𝑚 = 𝜔2 𝐿 = .
𝑅 0 1 𝐿2 𝐿1 𝐿2

 The values of 𝑔𝑚 obtained in above equations are the values necessary for the circuit
to support steady-state oscillations, denoted by 𝑔𝑚,𝑠𝑠 . In practical applications the
transistor is biased to set the transconductance to a value somewhat larger, e.g. a
factor of 2 to 5 larger, than 𝑔𝑚,𝑠𝑠 . One reason for doing this is to ensure that
oscillations start reliably even if component values change slightly.

Dept. of Telecoms Engineering 12

Dung Trinh, PhD HCMUT / 2019

2. LC Oscillator Analysis using Loop Gain

 Setting 𝑇 > 1 means that the oscillations will not be maintained at a steady state;
rather, they will grow in amplitude. Growth will proceed until the active device is no
longer operating in the “small-signal” mode.
 As the oscillation grows, eventually the amplitude of the oscillation will be limited by
nonlinear effects. To first order, the onset of nonlinear operation coincides with a
reduction in the gain of the active device.
 This is called gain saturation, and the effect can be modeled as a decrease in the
transconductance, 𝑔𝑚 , and hence a decrease in the loop gain.
 As the amplitude of the oscillation grows, the transconductance is decreased to the
point where the magnitude of the loop gain is 1. At this point steady-state oscillation
will be maintained.

Dept. of Telecoms Engineering 13

Dung Trinh, PhD HCMUT / 2019

2. LC Oscillator Analysis using KCL

 Applying KCL at the output, we have:
𝑉𝑜 𝑉𝑖𝑛
𝑔𝑚 𝑉𝑖𝑛 + + =0
𝑍2 𝑍1
𝑍1 +𝑍3
where 𝑉𝑜 = 𝑉𝑖𝑛 . Replace 𝑉𝑜 in the above equation, then:
𝑍1
𝑉𝑖𝑛 𝑍1 + 𝑍3 𝑉𝑖𝑛
𝑔𝑚 𝑉𝑖𝑛 + + =0
𝑍2 𝑍1 𝑍1
1 1 𝑍3 Oscillator circuit
→ 𝑉𝑖𝑛 𝑔𝑚 + + + =0
𝑍1 𝑍2 𝑍1 𝑍2
 For oscillation, 𝑉𝑖𝑛 must not be zero, therefore it enforces:
1 1 𝑍3
𝑔𝑚 + + + =0
𝑍1 𝑍2 𝑍1 𝑍2
1 1
 For Colpitts oscillator, 𝑍1 = 𝑗𝜔𝐶 , 𝑍2 = 𝑗𝜔𝐶 , and 𝑍3 = 𝑅 + 𝑗𝜔𝐿3 , then:
1 2
1
𝜔0 = 𝐶 𝐶
𝐿𝐶 1+𝐶2
1 2
𝑅 𝐶1 + 𝐶2
𝑔𝑚 = 𝑅𝜔02 𝐶1 𝐶2 =
𝐿
Colpitts Oscillator
Dept. of Telecoms Engineering 14
Dung Trinh, PhD HCMUT / 2019

2. LC Oscillator Analysis using KCL

1
 For Hartley oscillator, 𝑍1 = 𝑗𝜔𝐿1 , 𝑍2 = 𝑗𝜔𝐿2 , and 𝑍3 = 𝑅 + , then:
𝑗𝜔𝐶3
1
𝜔0 =
𝐶(𝐿1 +𝐿2 )
𝑅 𝑅𝐶 𝐿1 + 𝐿2
𝑔𝑚 = 2 =
𝜔0 𝐿1 𝐿2 𝐿1 𝐿2

Hartley Oscillator Colpitts Oscillator

 Example 1: For the Colpitts oscillator in following figure, assume parameters of 𝐿 =

1𝜇𝐻 , 𝐶1 = 𝐶2 = 1𝑛𝐹 and 𝑅 = 4𝑘Ω. Determine the oscillator frequency and the
required value of 𝑔𝑚 . Is this value of 𝑔𝑚 reasonable for a MOSFET? Why?
 Solution:
1 1
𝑓0 = = = 7.12𝑀𝐻𝑧.
𝐶 𝐶 10−9 ×10−9
2𝜋 𝐿𝐶 1+𝐶2 2𝜋 10−6 × −9
1 2 10 +10−9
𝐶2 10−9
𝑔𝑚 = = = 0.25𝑚𝐴/𝑉.
𝑅𝐶1 4×103 ×10−9

Equivalent circuit

Dept. of Telecoms Engineering 15

Dung Trinh, PhD HCMUT / 2019

2. LC Oscillator Analysis using Negative Resistance

 The requirements for oscillation in a circuit can be derived using the so-called
negative resistance concept.
 Consider a circuit consisting of arbitrary impedances 𝑍𝐴 and 𝑍𝐵 , as shown in the
following figure. The Kirchoff’s loop equation is I 𝑍𝐴 + 𝑍𝐵 = 0. Hence, finite
current I is allowed if 𝑍𝐴 + 𝑍𝐵 = 0.
 It is instructive to analyze the oscillator circuit using this approach. Suppose that the
circuit is divided into two parts, by defining 𝑍𝐴 = 𝑍1 .
 The impedance 𝑍𝐵 is the impedance seen looking into the port defined by the base
and emitter terminals of the transistor, with 𝑍1 removed from the circuit.
 Applying KCL at the input and output, we have:
𝑣𝑥 − 𝑣𝑜 𝑣𝑜
𝑖𝑥 = = 𝑔𝑚 𝑣𝑥 +
𝑍3 𝑍3
𝑣𝑥 𝑍2 + 𝑍3
→ 𝑍𝐵 = =
𝑖𝑥 1 + 𝑔𝑚 𝑍2
 Enforcing 𝑍𝐴 + 𝑍𝐵 = 0 yields
𝑍2 + 𝑍3
𝑍1 + =0 Impedance of oscillator circuit
1 + 𝑔𝑚 𝑍2
𝑍1 𝑍2
→ −𝑔𝑚 =1
𝑍1 + 𝑍2 + 𝑍3
Dept. of Telecoms Engineering 16
Dung Trinh, PhD HCMUT / 2019

2. LC Oscillator Analysis using KCL

 Example 2: A Hartley oscillator circuit having two individual inductors of 0.5mH each
(𝐿1 = 𝐿2 = 0.5𝑚𝐻), are design to resonate in parallel with a variable capacitor that can
be varied from 100pF to 500pF. Determine the upper and lower frequencies of
oscillation and also the Hartley oscillators bandwidth.
 Solution:
The upper frequency of oscillation:
1 1
𝑓𝐻 = = = 503 (𝑘𝐻𝑧)
−3 −3
2𝜋 𝐿1 + 𝐿2 𝐶1 2𝜋 0.5 × 10 + 0.5 × 10 100 × 10−12

The lower frequency of oscillation:

1 1
𝑓𝐿 = = = 225 (𝑘𝐻𝑧)
−3 −3
2𝜋 𝐿1 + 𝐿2 𝐶2 2𝜋 0.5 × 10 + 0.5 × 10 500 × 10 −12

The bandwidth of oscillation:

𝐵𝑊 = 𝑓𝐻 − 𝑓𝐿 = 503 − 225 = 278 (𝑘𝐻𝑧)

Dept. of Telecoms Engineering 17

Dung Trinh, PhD HCMUT / 2019

2. Hartley and Colpitts Oscillator

 The total capacitance of the tank circuit is split into two parts, 𝐶1 and 𝐶2 and
connected in series so that the net capacitance of the tank circuit is reduced. And
1 𝐿
thereby quality factor 𝑄 = 𝑅 of the tank circuit increases hence Colpitts has better
𝐶
frequency stability when compared to Hartley oscillator.

Hartley Oscillator Colpitts Oscillator

Hartley oscillator uses a tapped coil for It is easier to tap 2 capacitors in series as
the feedback for oscillation, and these well as being cheaper than to use a
tend to be more difficult to make. tapped coil for feedback purpose.
Capacitive tuning hence offers very low Circuit is economical and small in size due
wear and tear problem. to requirement of one inductor.
Circuit is bulky due to the present of two Inductive tuning offers very high wear and
inductors. tear problem.

Dept. of Telecoms Engineering 18

Dung Trinh, PhD HCMUT / 2019

2. Common-Collector Colpitts Oscillator

 Practical circuits for Colpitts-type oscillators are shown in following figures. We will
analyze the common-collector configuration.
 The small-signal equivalent circuit is shown where the transistor has been replaced
with its hybrid-pi model, and the finite Q of the inductor is modeled with a series
resistance, r.

CC Colpitts Oscillator CB Colpitts Oscillator CE Colpitts Oscillator

Dept. of Telecoms Engineering 19

Dung Trinh, PhD HCMUT / 2019

2. Common-Collector Colpitts Oscillator

 The small-signal equivalent circuit can be further simplified by combining some
elements as shown in following figure where
1
𝑍1 = 𝑟
𝑗𝜔 𝐶1 + 𝐶𝜋 𝜋
1
𝑍2 = 𝑅𝑒
𝑗𝜔 𝐶2 + 𝐶𝑜
1
𝑍3 = 𝑅1 𝑅2 𝑟 + 𝑗𝜔𝐿
𝑗𝜔𝐶𝜇
 In practice it is useful to choose 𝐶1 ≫ 𝐶𝜋 , 𝐶2 ≫ 𝐶𝑜 . This ensures that the external
components swamp the internal capacitances of the transistor, thereby minimizing the
circuit’s dependence on variations in the internal transistor capacitances.

Small signal equivalent circuit of CC Colpitts Oscillator Simplified small signal equivalent circuit.
Dept. of Telecoms Engineering 20
Dung Trinh, PhD HCMUT / 2019

2. Common-Collector Colpitts Oscillator

 It is also useful to choose 𝐶1 and 𝐶2 to be large enough so that 1 𝜔𝐶1 ≪ 𝑟𝜋 , 1 𝜔𝐶2 ≪
𝑅𝑒 . This causes 𝑍1 and 𝑍2 to be dominated by the capacitances external to the
transistor, thereby minimizing dependence on 𝑟𝜋 (which depends on bias current and
transistor 𝛽) and losses in 𝑟𝜋 and 𝑅𝑒 .
 In view of these considerations, in the following analysis we shall make the following
replacements: 𝐶1 + 𝐶𝜋 → 𝐶1 and 𝐶2 + 𝐶𝑜 → 𝐶2 .
 It is now useful to make some approximations in order to simplify the analysis.We
assume that the impedances of the capacitances 𝐶1 and 𝐶2 are small compared to 𝑟𝜋
and 𝑅𝑒 , respectively. In other words, define 𝑄1 = 𝜔𝐶1 𝑟𝜋 and 𝑄2 = 𝜔𝐶2 𝑅𝑒 . We
assume that 𝑄1 ≫ 1 and 𝑄2 ≫ 1.
 Then 𝑍1 and 𝑍2 can be transformed using a high-Q parallel to series transformation.
𝑟 1 1 1
𝑍1 ≃ 𝑄𝜋2 + 𝑗𝜔𝐶 = 𝜔2 𝐶 2𝑟 + 𝑗𝜔𝐶 , 𝑄1 ≫ 1
1 1 1 𝜋 1
𝑅 1 1 1
𝑍2 ≃ 𝑄𝑒2 + 𝑗𝜔𝐶 = 𝜔2𝐶 2𝑅 + 𝑗𝜔𝐶 , 𝑄2 ≫ 1
2 2 2 𝑒 2

𝑍3 ≃ 𝑟 + 𝑗𝜔𝐿

Dept. of Telecoms Engineering 21

Dung Trinh, PhD HCMUT / 2019

2. Common-Collector Colpitts Oscillator

 The condition for steady-state oscillation is
𝑍1 + 𝑍2 + 𝑍3 + 𝑔𝑚 𝑍1 𝑍2 = 0
 The real part of the equation is
1 1 1 1
2 2 + 2 2 + 𝑟 + 𝑔𝑚 4 2 2 − 2
=0
𝜔 𝐶1 𝑟𝜋 𝜔 𝐶2 𝑅𝑒 𝜔 𝐶1 𝐶2 𝑟𝜋 𝑅𝑒 𝜔 𝐶 𝐶
1 2

where 𝑟𝜋 = 𝛽 𝑔𝑚 .
 At sufficiently high frequencies, the term involving 𝜔−4 may be neglected provided
that 𝜔2 𝐶1 𝐶2 𝑟𝜋 𝑅𝑒 ≫ 1, or if 𝑄1 𝑄2 ≫ 1. The steady-state transconductance can be
written as
𝐶1
𝜔2 𝐶1 𝐶2 𝑟 + 𝐶 𝑅
2 𝑒
𝑔𝑚,𝑠𝑠 =
𝐶
1− 2
𝐶1 𝛽
 If 𝑅𝑒 is allowed to approach infinity (so that 𝑍2 becomes a pure reactance), and if 𝛽 ≫
𝐶2 2
𝐶1 , then 𝑔𝑚,𝑠𝑠 ≃ 𝜔 𝐶1 𝐶2 𝑟.
 In practical cases with 𝛽 ≫ 1 and 𝐶1 > 𝐶2 then
𝐶1 1
𝑔𝑚,𝑠𝑠 ≃ 𝜔2 𝐶1 𝐶2 𝑟 + and 𝜔0 = .
𝐶2 𝑅𝑒 𝐶 𝐶
𝐿𝐶 1+𝐶2
1 2
Dept. of Telecoms Engineering 22
Dung Trinh, PhD HCMUT / 2019

2. Loading Effect of Colpitts Oscillator

 The input impedance of the amplifier acts as a load on the resonant feedback circuit
and reduces the Q of the circuit.
 The resonant frequency of a parallel resonant circuit depends on the Q as:
1 𝑄2
𝑓0 =
2𝜋 𝐿𝐶𝑇 𝑄2 + 1
 A FET can be used in place of a BJT to minimize the loading effect of the transistor’s
input impedance because of the higher input impedance of FET.
 Also, when an external load is connected to the oscillator output, 𝑓𝑟 may decrease
because of a reduction in Q if the load resistance is too small.

Dept. of Telecoms Engineering 23

Dung Trinh, PhD HCMUT / 2019

2. Loading Effect of Colpitts Oscillator

 Example 3:
a. Determine the frequency of oscillation of the circuit shown below. Assume the
loading of the feedback circuit is neglectible.
b. Find the frequency of oscillation if the oscillator is loaded to a point where 𝑄 = 8.
 Solution
a. The frequency of oscillation is:
𝐶 𝐶
𝐶𝑇 = 𝐶 1+𝐶2 = 9.1 (𝑛𝐹)
1 2
1 1
𝑓𝑟 = = = 7.46(𝑘𝐻𝑧)
2𝜋 𝐿𝐶𝑇 2𝜋 50×10−3 ×9.1×10−9

b. The loaded frequency of oscillation is:

1 𝑄2
𝑓𝑟 = = 7.40 (𝑘𝐻𝑧)
2𝜋 𝐿𝐶𝑇 𝑄2 +1

Dept. of Telecoms Engineering 24

Dung Trinh, PhD HCMUT / 2019

2. Clapp Oscillator
 The Clapp oscillator is a variation of the Colpitts.
 The basic difference is an additional capacitor, C3 in series with the inductor in the
resonant feedback circuit.
 Since C3 is in series with C1 and C2 around the tank circuit, the total capacitance is
1 1 1
𝐶𝑇 = + +
𝐶1 𝐶2 𝐶3

 In Clapp oscillator, C3 is much smaller than C1

and C2. As a result CT is approximately equal
to C3 and the resonant frequency is controlled
by C3.
 Clapp provides a more accurate and stable
frequency of oscillation since C1 and C2 are
shunted by transistor and the stray capacitances
alter their values.

Dept. of Telecoms Engineering 25

Dung Trinh, PhD HCMUT / 2019

3. Wien Bridge Oscillator (RC Oscillator)

 Wien bridge oscillator is a low frequency
oscillator ranging from a few Hz to 1 MHz.
 Since the noninverting amplifier introduces
zero phase shift, the frequency-selective
feedback circuit must also introduce zero phase
shift to create the positive feedback condition.
 The loop gain is the product of the amplifier
gain and the feedback transfer function:
𝑅2 𝑍𝑝
𝑇 𝑠 = 1+
𝑅1 𝑍𝑝 + 𝑍𝑠
𝑅 1+𝑠𝑅𝐶
where 𝑍𝑝 = 1+𝑠𝑅𝐶 and 𝑍𝑠 = 𝑠𝐶
Wien bridge oscillator
 Then:
𝑅2 1
𝑇 𝑠 = 1+
𝑅1 3 + 𝑠𝑅𝐶 + 1/𝑠𝑅𝐶
 At the oscillation frequency, 𝜔0 :
𝑅2 1
𝑇 𝑗𝜔0 = 1 +
𝑅1 3 + 𝑗𝜔0 𝑅𝐶 + 1/𝑗𝜔0 𝑅𝐶
Dept. of Telecoms Engineering 26
Dung Trinh, PhD HCMUT / 2019

3. Wien Bridge Oscillator (RC Oscillator)

 Since 𝑇 𝑗𝜔0 must be real, the imaginary component of 𝑇 𝑗𝜔0 must be zero, then:
1 1
𝑗𝜔0 𝑅𝐶 + 𝑗𝜔 = 0 which gives: 𝜔0 = 𝑅𝐶
0 𝑅𝐶

 The magnitude condition is then:

𝑅2 1 𝑅2
1+ = 1 which gives: =2
𝑅1 3 𝑅1

 Example 4: Design a Wien-bridge oscillator to oscillate at 𝑓0 = 20𝑘𝐻𝑧.

 Solution:
1 1
𝑅𝐶 = = 3
= 7.96 × 10−6
2𝜋𝑓0 2𝜋 × 20 × 10
• A 10𝑘Ω resistor and 796pF capacitor satisfy the requirement. Since the amplifier
resistor ratio must be 𝑅2 𝑅1 = 2, therefore we can choose 𝑅1 = 10𝑘Ω, and 𝑅2 =
20𝑘Ω.
• Trade off: Standard-valued resistors 𝑅1 = 10𝑘Ω, and 𝑅2 = 20𝑘Ω. In place of the
ideal 796 pF capacitor, a standard-valued capacitor 𝐶 = 800𝑝𝐹 can be used. The
oscillation frequency would then be 𝑓0 = 19.9𝑘𝐻𝑧.

Dept. of Telecoms Engineering 27

Dung Trinh, PhD HCMUT / 2019

3. Phase Shift Oscillator (RC Oscillator)

 The phase-shifter consists of a negative gain amplifier (-K) with a third order RC
ladder network in the feedback.
 The inverting amplifier introduces a 1800 phase shift, which means that each RC
network must provide 600 of phase shift to produce the 1800 required of the frequency-
sensitive feedback network in order to produce positive feedback.
 The voltage followers in the circuit eliminate loading effects between each RC filter
stage.
 Note that the inverting terminal of op-amp A3 is at virtual ground; therefore, the RC
network between op-amps A2 and A3 functions exactly as the other two RC networks.

Phase shift oscillator

Dept. of Telecoms Engineering 28
Dung Trinh, PhD HCMUT / 2019

3. Phase Shift Oscillator (RC Oscillator)

 The transfer function of the first RC network:
𝑠𝑅𝐶
𝑣1 /𝑣𝐼 =
1 + 𝑠𝑅𝐶
 Since the RC networks are assumed to be identical, and since there is no loading effect
of one RC stage on another, we have:
3
𝑠𝑅𝐶
𝛽 𝑠 = 𝑣3 /𝑣𝐼 =
1 + 𝑠𝑅𝐶
 The amplifier gain 𝐴(𝑠) is:
𝑅2
𝐴 𝑠 =− 𝑅

Phase shift oscillator

Dept. of Telecoms Engineering 29
Dung Trinh, PhD HCMUT / 2019

3. Phase Shift Oscillator (RC Oscillator)

 The loop gain is then:
3
𝑅2 𝑠𝑅𝐶
𝑇 𝑠 =𝐴 𝑠 𝛽 𝑠 =−
𝑅 1 + 𝑠𝑅𝐶
 The condition for oscillation is that 𝑇 𝑗𝜔0 = 1 and the phase of 𝑇 𝑗𝜔0 must be
360 degree. We set 𝑠 = 𝑗𝜔:
3
𝑅2 𝑗𝜔𝑅𝐶 𝑅2 𝑗𝜔𝑅𝐶 𝜔𝑅𝐶 2
𝑇 𝑗𝜔 = − =
𝑅 1 + 𝑗𝜔𝑅𝐶 𝑅 1 − 3 𝜔𝑅𝐶 2 + 𝑗𝜔𝑅𝐶 3 − 𝜔𝑅𝐶 2
 Since the numerator is purely imaginary, the denominator must become purely
imaginary, or:
1 − 3 𝜔0 𝑅𝐶 2 = 0
which yields 𝜔0 = 1 3𝑅𝐶
.
 At the oscillation frequency, 𝑓0 :
𝑅2 1
𝑇 𝑗𝜔0 =
𝑅 8
 Consequently, the condition 𝑇 𝑗𝜔0 = 1 is satisfied when
𝑅2
=8
𝑅
Dept. of Telecoms Engineering 30
Dung Trinh, PhD HCMUT / 2019

3. Phase Shift Oscillator (RC Oscillator)

 The two voltage followers in the above circuit need not be included in a practical
phase-shift oscillator. The following figure shows a phase-shift oscillator without the
voltage-follower buffer stages.
 The three RC network stages and the inverting amplifier are still included. The loading
effect of each successive RC network complicates the analysis, but the same principle
of operation applies.
 The amplifier gain 𝐴(𝑠) is:
𝑅2
𝐴 𝑠 =−
𝑅
 The feedback gain is:
𝑠𝑅𝐶 3
𝛽 𝑠 =
𝑠𝑅𝐶 3 + 6 𝑠𝑅𝐶 2 + 5𝑠𝑅𝐶 + 1

Dept. of Telecoms Engineering 31

Dung Trinh, PhD HCMUT / 2019

3. Phase Shift Oscillator (RC Oscillator)

 The loop gain is then:
𝑅2 𝑠𝑅𝐶 3
𝑇 𝑠 =𝐴 𝑠 𝛽 𝑠 =−
𝑅 𝑠𝑅𝐶 3 + 6 𝑠𝑅𝐶 2 + 5𝑠𝑅𝐶 + 1
 The condition for oscillation is that 𝑇 𝑗𝜔0 = 1 and the phase of 𝑇 𝑗𝜔0 must be
360 degree. We set 𝑠 = 𝑗𝜔:
𝑅2 𝑗𝜔𝑅𝐶 3
𝑇 𝑗𝜔 = −
𝑅 𝑗𝜔𝑅𝐶 3 + 6 𝑗𝜔𝑅𝐶 2 + 5𝑗𝜔𝑅𝐶 + 1
 At oscillation frequency, 𝑗𝜔0
𝑅2 𝑗𝜔0 𝑅𝐶 3
− =1
𝑅 𝑗𝜔0 𝑅𝐶 3 + 6 𝑗𝜔0 𝑅𝐶 2 + 5𝑗𝜔0 𝑅𝐶 + 1
𝑅2
→ 6 𝜔0 𝑅𝐶 2 + 𝑗 𝜔0 𝑅𝐶 3 + 𝜔0 𝑅𝐶 3 − 5𝜔0 𝑅𝐶 = 1
𝑅

which yields
𝜔0 = 1 6𝑅𝐶.
𝑅2
= 29
𝑅

Dept. of Telecoms Engineering 32

Dung Trinh, PhD HCMUT / 2019

3. Phase Shift Oscillator (RC Oscillator)

 Example 5: Determine the value of capacitance C and the value of RD of the Phase-
shift oscillator shown, if the output frequency is 1 kHz.
 Solution:
1 1 1
𝑓0 = →𝐶= = = 6.5(𝑛𝐹)
2𝜋𝑅𝐶 6 2𝜋𝑓0 𝑅 6 2𝜋 × 103 × 104 6
• 𝐴(𝑗𝜔0 ) must be greater than 29. Choose 𝐴(𝑗𝜔0 ) = 40. Then
𝐴 = 𝑔𝑚 𝑅𝐿 = 𝑔𝑚 𝑅𝐷 𝑟𝑑 = 40
→ 𝑅𝑑 = 10 (𝑘Ω)

Dept. of Telecoms Engineering 33

Dung Trinh, PhD HCMUT / 2019

3. Phase Shift Oscillator (RC Oscillator)

 Example 6: Determine the value of Rf necessary or the circuit shown below to operate
as an oscillator. Determine the frequency of oscillation.
 Solution:
• The circuit oscillates if
𝑅𝑓
𝐴 = 𝑅 = 29 → 𝑅𝑓 = 29𝑅3 = 29 × 10𝑘Ω = 290(𝑘Ω)
3

• The frequency of oscillation:

1 1
𝑓0 = 2𝜋𝑅𝐶 = 2𝜋 = 6.5(𝑘𝐻𝑧)
6 6×104 ×10−9

Dept. of Telecoms Engineering 34

Dung Trinh, PhD HCMUT / 2019

4. Crystal Oscillator
 The most stable and accurate type of feedback oscillator uses a piezoelectric crystal in
the feedback loop to control the frequency.
 Quartz is one type of crystalline substance found in nature that exhibits a property
called the piezoelectric effect.
 When a changing mechanical stress is applied across the crystal to cause it to vibrate, a
voltage develops at the frequency of mechanical vibration.
 Conversely, when an AC voltage is applied across the crystal, it vibrates at the
frequency of the applied voltage.
 The greatest vibration occurs at the crystal’s natural resonant frequency, which is
determined by the physical dimensions and by the way the crystal is cut.

Dept. of Telecoms Engineering 35

Dung Trinh, PhD HCMUT / 2019

4. Crystal Oscillator
 A piezoelectric crystal, such as quartz, exhibits electromechanical-resonance
characteristics that are very stable (with time and temperature) and highly selective
(having very high Q factors).
 The resonance properties are characterized by
• A large inductance L (as high as hundreds of henrys),
• A very small series capacitance CS (as small as 0.0005 pF),
• A series resistance r representing a Q factor 𝜔0 𝐿 𝑟 that can be as high as a few
105.
• And a parallel capacitance Cp (a few pF).

Dept. of Telecoms Engineering 36

Dung Trinh, PhD HCMUT / 2019

4. Crystal Oscillator
 Example 7: Given a quartz crystal with the following parameters: 𝐿1 = 3𝑚𝐻, 𝑅1 =
6.8Ω, 𝐶1 = 0.0997𝑝𝐹 and 𝐶2 = 30𝑝𝐹. Compute the series, parallel resonant frequency
and the Q factor of the crystal.
 Solution:
• The series resonant frequency of the crystal:
1 1
𝑓𝑠 = = = 9.2(𝑀𝐻𝑧)
−3
2𝜋 𝐿1 𝐶1 2𝜋 3 × 10 × 0.0997 × 10 −12

• The parallel resonant frequency of the crystal:

1
𝑓𝑝 = = 9.23(𝑀𝐻𝑧)
𝐶1 𝐶2
2𝜋 𝐿1 𝐶 + 𝐶
1 2
• The Q factor of the crystal:
𝑋𝐿 2𝜋 × 9.2 × 106 × 3 × 10−3
𝑄= = = 25500
𝑟 6.8

Dept. of Telecoms Engineering 37

Dung Trinh, PhD HCMUT / 2019

4. Crystal Oscillator
 Since the Q factor is very high, we may express the crystal impedance as
2 1
1 1 𝑠 + 𝐿𝐶𝑠
𝑍 𝑠 = =
1 𝑠𝐶𝑝 2 𝐶𝑝 + 𝐶𝑠
𝑠𝐶𝑝 + 𝑠 + 𝐿𝐶 𝐶
𝑠𝐿 + 1 𝑠𝐶 𝑝 𝑠
𝑠
 The crystal has two resonance frequencies: a series resonance at 𝜔𝑠 and a parallel
resonance at 𝜔𝑝
𝜔𝑠 = 1
𝐿𝐶𝑠
𝜔𝑝 = 1
𝐶𝑠 𝐶𝑝
𝐿𝐶 +𝐶
𝑠 𝑝

 Then we can write:

1 𝜔2 − 𝜔𝑠2
𝑍 𝑗𝜔 = −𝑗
𝜔𝐶𝑝 𝜔 2 − 𝜔𝑝2
 We note that 𝜔𝑝 > 𝜔𝑠 . However, since 𝐶𝑝 ≫ 𝐶𝑠 , the two resonance frequencies are
very close. The parallel resonant frequency is usually at least 1 kHz higher than the
series resonant frequency.
Dept. of Telecoms Engineering 38
Dung Trinh, PhD HCMUT / 2019

4. Crystal Oscillator
 The crystal reactance is inductive over the very narrow frequency band between 𝜔𝑠
and 𝜔𝑝 . For a given crystal, this frequency band is well defined. Thus we may use the
crystal to replace the inductor of the Colpitts oscillator.
 The resulting circuit will oscillate at the resonance frequency of the crystal inductance
𝐶 𝐶
L with the series equivalent of 𝐶𝑠 and 𝐶𝑝 + 1 2 . Since 𝐶𝑠 is much smaller than the
𝐶1 +𝐶2
three other capacitances, it will be dominant and
𝜔0 ≃ 1 ≃ 𝜔𝑠
𝐿𝐶𝑠
 A popular Colpitts configuration (called the Pierce
oscillator) utilizing a CMOS inverter as an amplifier.
 Resistor Rf determines a dc operating point in the high-gain
region of the VTC of the CMOS inverter. Resistor R1
together with capacitor C1 provides a low-pass filter that
discourages the circuit from oscillating at a higher harmonic
of the crystal frequency.
 The extremely stable resonance characteristics and the very
high Q factors of quartz crystals result in oscillators with
very accurate and stable frequencies. Pierce oscillator
Dept. of Telecoms Engineering 39
Dung Trinh, PhD HCMUT / 2019

4. Providing clock to MCU using crystals

 Microprocessor and CPU needs stable clock input
for operation. Quartz crystal is widely used for these
purposes. Quartz crystal provides high accuracy and
stability compared to other RC or LC or RLC
oscillators.
 In general the clock frequency is used for
microcontroller or CPU is ranged from kHz to Mhz.
This clock frequency determines how fast the
processor can process data.
 To achieve this frequency a series crystal used with
two same value capacitors network is used across the
oscillator input of the respective MCU or CPU.
 A Crystal with two capacitor forms a network and
connected across Microcontroller unit or Central
processing unit via OSC1 and OSC2 input pin. The
capacitor value range from 10pF to 42 pF, anything
in between but 15pF, 22pF, 33pF is used widely.

Dept. of Telecoms Engineering 40

Dung Trinh, PhD HCMUT / 2019

5. Voltage Controlled Oscillator

 An oscillator whose output frequency can be controlled by a voltage signal. VCO is
usually implemented by tuning the feedback network by a variable capacitor.
 It is a very useful component in any system whose operating frequency needs to
change:
 FM radio, Frequency sweeping radars
 Frequency synthesizers, Spectrum analyzers
 Network analyzers
 All the specifications for fixed oscillators are still applicable for VCO.
 Tuning sensitivity: frequency change per volt of tuning voltage change.
 Sometimes referred to as “VCO gain”.
 Specified in Hz/V, or MHz/V
 A higher sensitivity means that you can cover a large frequency range with small
tuning voltage; it also means that the output frequency is more sensitive to the
noise in the tuning voltage.

Dept. of Telecoms Engineering 41

Dung Trinh, PhD HCMUT / 2019

5. Voltage Controlled Oscillator

 Sample VCO Datasheet

Dept. of Telecoms Engineering 42

Dung Trinh, PhD HCMUT / 2019

5. Voltage Controlled Oscillator

 Sample VCO Datasheet

Dept. of Telecoms Engineering 43

Dung Trinh, PhD HCMUT / 2019

5. Voltage Controlled Oscillator

 A general VCO can be treated as a black box with an input 𝑉𝑡𝑢𝑛𝑒 and a periodic
oscillating output 𝑉(𝑡). The output voltage 𝑉𝑜𝑢𝑡 𝑡 , differential or single ended, is
periodic:
𝑉𝑜𝑢𝑡 𝑡 = 𝑉0 sin 𝜔𝑐 𝑡 + 𝜑
 The angular carrier frequency 𝜔𝑐 𝑉𝑡𝑢𝑛𝑒 is dependent on the tuning voltage input
𝑉𝑡𝑢𝑛𝑒 . The transfer function of input voltage to output frequency:
𝜕𝑓𝑐
𝐾𝑉𝐶𝑂 =
𝜕𝑉𝑡𝑢𝑛𝑒
 Pushing: the transfer function from power suply to output frequency can be defined as
𝜕𝑓𝑐
𝐾𝑉𝑑𝑑 =
𝜕𝑉𝑑𝑑
 Pulling: VCO output frequency also varies with the load attached to its output:
𝜕𝑓𝑐
𝑃𝑢𝑙𝑙𝑖𝑛𝑔 =
𝜕𝑙𝑜𝑎𝑑

Dept. of Telecoms Engineering 44

Dung Trinh, PhD HCMUT / 2019

5. VCO – Phase Noise and Jitter

 As already stated, an ideal sinusoidal oscillator is described as
𝑉𝑜𝑢𝑡 𝑡 = 𝑉0 sin 𝜔𝑐 𝑡 + 𝜑
 A real oscillator is more generally given by:
𝑉𝑜𝑢𝑡 𝑡 = 𝑉0 𝑡 y 𝜔𝑐 𝑡 + 𝜑 𝑡
 The fluctuations introduced by 𝑉0 𝑡 and 𝜑 𝑡 – now functions in time - result in
sidebands close to 𝑓𝑐 , with symmetrical distribution around 𝑓𝑐 .
 The frequency fluctuations correspond to jitter in the time-domain, which is a random
perturbation of zero-crossings of a periodic signal.

Dept. of Telecoms Engineering 45

Dung Trinh, PhD HCMUT / 2019

5. VCO – Phase Noise and Jitter

 Frequency fluctuations are usually characterized by the single sideband noise spectral
density normalized to the carrier signal power. It is defined as
Ρ𝑠𝑖𝑑𝑒𝑏𝑎𝑛𝑑 𝑓𝑐 + Δ𝑓, 1𝐻𝑧
ℒ 𝑓𝑐 , Δ𝑓 = 10 log (𝑑𝐵𝑐/𝐻𝑧)
Ρ𝑐𝑎𝑟𝑟𝑖𝑒𝑟
• 𝑃𝑐𝑎𝑟𝑟𝑖𝑒𝑟 denotes the carrier signal power at the carrier frequency 𝑓𝑐 .
• 𝑃𝑠𝑖𝑑𝑒𝑏𝑎𝑛𝑑 denotes the single sideband power at the offset Δ𝑓 from the carrier 𝑓𝑐 at
a measurement bandwidth of 1Hz.

Jitter in time domain relates to phase noise in frequency domain

Dept. of Telecoms Engineering 46
Dung Trinh, PhD HCMUT / 2019

5. VCO – Phase Noise and Jitter

 The LO signal used for down-conversion has a noisy spectrum. Besides the wanted
signal with small power an unwanted signal with large power is present in an adjacent
channel (at a close-by frequency).
 After mixing with the LO the down-converted spectrum consists of two overlapping
spectra. The wanted signal suffers from significant noise due to in-band signal from
the down-conversion of the interferer by the LO sideband: the signal-to-noise ratio is
degraded.
 In order to be able to detect the signals from all channels while (stronger) interferers
may be present stringent phase noise specifications have to be met in wireless
communication systems.

Simplified receiver block diagram Effect of oscillator phase noise in a receiver

Dept. of Telecoms Engineering 47
Dung Trinh, PhD HCMUT / 2019

5. Phase Locked Loop (PLL)

 To achieve absolute synchronization of the LO signals, the VCO is engaged in a
phase-locked loop.
 PLL design is eased significantly with a VCO offering a linear dependence between
tuning voltage and frequency.
 At low offset frequencies, a 1/Δf3 behavior, also called flicker-noise behavior, is
observed.
 For medium offset frequencies the phase noise shows a 1/Δf2-dependence, also called
white-noise behavior, up to where the constant amplifier noise floor begins to
dominate.

Phase Locked Loop Noise spectra of the VCO and the resulting PLL output
Dept. of Telecoms Engineering 48
Dung Trinh, PhD HCMUT / 2019

6. Exercises
Exercise 1: Consider the voltage amplifier in following figure.
𝑉𝐶𝐶 = 12𝑉, 𝑅1 = 10𝑘Ω, 𝑅2 = 30𝑘Ω, 𝑅𝑒 = 1𝑘Ω
𝑅𝐶 = ∞ , 𝑅𝐿 = 1𝑘Ω , 𝐿 = 2𝜇𝐻 , 𝐶 = 50𝑝𝐹
Capacitors that are not labeled are assumed to be “short circuits” over the frequency range
of interest. The transistor’s & is large enough so that the bias point does not explicitly
depend on its value. You may neglect the transistor parameters 𝑟𝑥 , 𝑟𝜇 , 𝑟𝜇 , 𝑟𝑜 , and 𝐶𝑜 in your
analysis for parts 1b, 1c, and 1d.
a. Find the quiescent collector current, 𝐼𝐶𝑄 . Express
your result in mA.
b. Find the resonant frequency of the amplifier. The
voltage gain will be largest at this frequency. Express
your result in MHz.
c. Find the voltage gain at resonance.
d. Find the 3 dB bandwidth of the amplifier.
Express your result in MHz.

Dept. of Telecoms Engineering 49

Dung Trinh, PhD HCMUT / 2019

6. Exercises
Exercise 2: A quartz crystal resonator has the equivalent circuit
shown in following figure. Suppose it is known that a particular
crystal has Q = 50000, series resonant frequency of 5𝑀𝐻𝑧,
and parallel resonant frequency of 5.005𝑀𝐻𝑧. It is also known
that the parallel resonant frequency shifts by 1𝑘𝐻𝑧, if a 3𝑝𝐹
capacitor is placed in parallel with the crystal. Find the values
of the equivalent circuit elements for the crystal.
Exercise 3: Consider the AC equivalent circuit for an oscillator
circuit in following figure. The loop gain for this circuit can be
shown to be:
1
𝑔𝑚 2
𝜔 𝐶1 𝐶2
𝑇=
1 1 𝑗𝜔𝑅𝐿
+ +
𝑗𝜔𝐶1 𝑗𝜔𝐶2 𝑅 + 𝑗𝜔𝐿
It has been assumed that the transistor immittances could be
ignored in deriving the above equation. Find an expression for
the frequency of oscillation, 𝜔0 . To simplify interpretation of
your result, define 𝐶0 = 𝐶1 𝐶2 𝐶1 + 𝐶2 and express your
result in terms of R, L, and 𝐶0 .
Dept. of Telecoms Engineering 50
Dung Trinh, PhD HCMUT / 2019

6. Exercises
Exercise 4: Consider the circuit in following figure.
a. Find an expression for the frequency of
oscillation.
b. Find the minimum value of gm required to
make the circuit oscillate.
Hartley Oscillator

Exercise 5: Consider the circuit in following figure.

The amplifier has infinite input impedance and
voltage gain A.
a. Find an expression for the frequency of
oscillation (if it occurs).
b. Find an expression for the value of the voltage
gain (A) required to sustain steady-state oscillations. Wien Oscillator

Dept. of Telecoms Engineering 51

Dung Trinh, PhD HCMUT / 2019

6. Exercises
Exercise 6: The crystal oscillator circuit shown in following figure is called a “series
mode” oscillator, because it will oscillate very close to the series resonant frequency of the
crystal. Here the crystal grounds the base of the transistor at its series resonant frequency.
The circuit will also oscillate if the crystal is replaced by an “AC” short circuit (e.g., a
bypass capacitor). Use the negative resistance approach to study this oscillator. You can
also assume that R1 and R2 can be neglected and that Re is much larger than the reactance
of C2, so that it can be neglected as well. The coupling capacitors can be taken to be AC
short circuits.
a. First, remove the crystal and “look in” to the rest
of the circuit. Solve for the input impedance and show
that it is given by
𝑍1 𝑍𝐿 + 𝑍2 + 𝑔𝑚 𝑟𝜋 𝑍1 𝑍2
𝑍𝑖𝑛 = 𝑟𝜋 +
𝑍1 + 𝑍2 + 𝑍𝐿
1 1 𝑗𝜔𝐿𝑅
𝑍1 = 𝑗𝜔𝐶 , 𝑍2 = 𝑗𝜔𝐶 , 𝑍𝐿 = 𝑗𝜔𝐿+𝑅𝐿
1 2 𝐿

b. Solve for the frequency at which the circuit will oscillate (𝜔0 ), if an AC short is
connected from the base of the transistor to ground (instead of the crystal). You can
assume that 𝑅𝐿 ≫ 𝜔𝐿 at the frequency of oscillation.
c. What condition must be satisfied in order to guarantee that oscillations will start?
Dept. of Telecoms Engineering 52
Dung Trinh, PhD HCMUT / 2019

6. Exercises
Exercise 7: Consider the oscillator circuit shown in following figure. In this problem,
𝑉𝑐𝑐 = 3𝑉, 𝑅𝑏 = 33𝑘Ω, 𝑅𝑒 = 100𝑘Ω, 𝐶1 = 10𝑝𝐹, 𝐶2 = 22𝑝𝐹. Unlabeled capacitors are
either coupling or bypass elements and have negligibly small impedance at the frequencies
of interest.
a. Find an approximate value for the quiescent
collector current, 𝐼𝐶𝑄 . You may assume that 𝛽 = 100.
b. Find the 𝑔𝑚 for the transistor in this circuit.
c. Assume that the reactances of 𝐶1 and 𝐶2 are small
enough such that 𝑟𝜋 , 𝑅𝑒 , and 𝑅𝑏 can be ignored for
small-signal analysis. In this case, if the inductor is
removed the impedance 𝑍𝑖𝑛 will be
𝑔𝑚 1 Calculate the inductance, L, required to
𝑍𝑖𝑛 = − 2 + set the potential frequency of oscillation
𝜔 𝐶1 𝐶2 𝑗𝜔 𝐶1 𝐶2
𝐶1 + 𝐶2 in this circuit to 50𝑀𝐻𝑧.

d. Suppose a lossy inductor is available with inductance equal to the value that you
calculated in part (c). Calculate the minimum inductor Q (i.e., QL) required for
oscillations to be sustained in this circuit.
Dept. of Telecoms Engineering 53
Dung Trinh, PhD HCMUT / 2019

6. Exercises
Exercise 8: Consider the Colpitts oscillator shown in following figure.
a. Find an expression for the frequency of oscillation, 𝜔0 . You may assume that the loop
gain of the circuit is large enough to cause oscillation to start. You may neglect the
transistor immittances, so that your result will be in terms of only the four parameters L,
𝐶1 , 𝐶2 , 𝐶𝑉 .
b. Denote the minimum and maximum tuning-diode capacitances by 𝐶𝑉,𝑚𝑖𝑛 and 𝐶𝑉,𝑚𝑎𝑥
𝐶𝑉,𝑚𝑎𝑥
and the capacitor tuning ratio by 𝑟𝐶 = . Denote the minimum and maximum
𝐶𝑉,𝑚𝑖𝑛
frequencies of oscillation by 𝜔0,𝑚𝑖𝑛 and 𝜔0.𝑚𝑎𝑥 and the oscillator tuning ratio by 𝑟𝑜 =
𝜔0,𝑚𝑎𝑥 ′ , i.e. 𝐶 ′ = 𝐶1 𝐶2 . Find an
. Finally, denote the series combination of 𝐶1 and 𝐶2 by 𝐶
𝜔 0,𝑚𝑖𝑛 𝐶 +𝐶
1 2
expression for 𝑟𝑜 in terms of 𝑟𝐶 , 𝐶 ′, and 𝐶𝑉,𝑚𝑎𝑥 only.
c. Find a numerical value for the tuning ratio 𝑟𝑜 if 𝐶 ′ = 2𝐶𝑉,𝑚𝑎𝑥 and 𝑟𝐶 = 4.

A VCO
Dept. of Telecoms Engineering 54
Dung Trinh, PhD HCMUT / 2019

Q&A

Dept. of Telecoms Engineering 55

Dung Trinh, PhD HCMUT / 2019

Chapter 5
Mixers

Trinh Xuan Dung, PhD

[email protected]

Department of Telecommunications
Faculty of Electrical and Electronics Engineering
Ho Chi Minh city University of Technology
Dept. of Telecoms Engineering 1
Dung Trinh, PhD HCMUT / 2019

Contents

1. Introduction
2. Mixers based on nonlinearities
3. Mixers based on Switches
4. Mixer specifications
5. Exercises

Dept. of Telecoms Engineering 2

Dung Trinh, PhD HCMUT / 2019

1. Introduction - Heterodyne
 Heterodyning is a radio signal processing technique popularized by Canadian
inventor-engineer Reginald Fessenden in 1901, in which new frequencies are created
by combining or mixing two frequencies.
 The two frequencies are combined in a nonlinear signal-processing device such as a
vacuum tube, transistor, or diode, usually called a mixer. In the most common
application, two signals at frequencies f1 and f2 are mixed, creating two new signals,
one at the sum f1 + f2 of the two frequencies, and the other at the difference f1 − f2.
These new frequencies are called heterodynes (dị thể in Vietnamese). Typically only
one of the new frequencies is desired, and the other signal is filtered out of the output
of the mixer.

f1 + f2
f1
f1 − f2

f2
A Minicircuits Mixer A Marki Mixer Input and output signals of a mixer
Dept. of Telecoms Engineering 3
Dung Trinh, PhD HCMUT / 2019

1. Introduction – Image Frequency

 Consider 𝑣1 𝑡 = 𝑉1 cos 𝜔1 𝑡 and 𝑣2 𝑡 = 𝑉2 cos 𝜔2 𝑡. A multiplier can realize the
mixer function
1
𝑣𝑜 𝑡 = 𝑣1 𝑡 × 𝑣2 𝑡 = 𝑉1 𝑉2 cos 𝜔1 + 𝜔2 𝑡 + cos 𝜔1 − 𝜔2 𝑡
2
 Example: 𝑓𝑅𝐹 = 2450𝑀𝐻𝑧, 𝑓𝐿𝑂 = 2400𝑀𝐻𝑧. The IF frequency: 𝑓𝐼𝐹 = 50𝑀𝐻𝑧.
 Example: 𝑓𝑅𝐹 = 2450𝑀𝐻𝑧, 𝑓𝐿𝑂 = 2500𝑀𝐻𝑧. The IF frequency: 𝑓𝐼𝐹 = 50𝑀𝐻𝑧.
 Example: 𝑓𝐼𝐹 = 200𝑀𝐻𝑧, 𝑓𝐿𝑂 = 2400𝑀𝐻𝑧. The RF frequency: 𝑓𝑅𝐹 =
2200𝑀𝐻𝑧, 2600MHz.
𝑓𝐿𝑂

𝑓𝐼𝐹 𝑓𝑅𝐹

𝑓𝐿𝑂
𝑓𝑅𝐹 𝑓𝑅𝐹
𝑓𝐼𝐹 𝑓𝑅𝐹 𝑓𝐼𝐹

Dept. of Telecoms Engineering 4

Dung Trinh, PhD HCMUT / 2019

1. Introduction – Image Frequency

 Example 1: The IS-54 digital cellular telephone system uses a receive frequency band
of 869–894 MHz, with a first IF frequency of 87 MHz and a channel bandwidth of 30
kHz. What are the two possible ranges for the LO frequency? If the upper LO frequency
range is used, determine the image frequency range. Does the image frequency fall
within the receive passband?
 Solution:
The two possible LO frequency ranges are:
956 𝑡𝑜 981𝑀𝐻𝑧
𝑓𝐿𝑂 = 𝑓𝑅𝐹 ± 𝑓𝐼𝐹 = 869 𝑡𝑜 894 ± 87 =
782 𝑡𝑜 807𝑀𝐻𝑧
Using the 956 − 981𝑀𝐻𝑧 LO, the IF frequencies are
𝑓𝐼𝐹 = 𝑓𝑅𝐹 − 𝑓𝐿𝑂 = 869 𝑡𝑜 894 − 956 𝑡𝑜 981 = −87𝑀𝐻𝑧
The RF image frequency range is
𝑓𝐼𝑀 = 𝑓𝐿𝑂 − 𝑓𝐼𝐹 = 956 𝑡𝑜 981 + 87 = 1043 𝑡𝑜 1068𝑀𝐻𝑧
which is well outside the receive band.

Dept. of Telecoms Engineering 5

Dung Trinh, PhD HCMUT / 2019

1. Introduction – Image Frequency

 Signals at the image frequency will be down-converted to IF frequency and corrupted
the desired signal.
• Even if there is no other signal at the image frequency, noise can still be down-
converted, lowering your NF by 3 dB!
• An Image reject filter is usually put in front of the mixer to attenuate the signals
and noise at the image frequency.
 Block diagram with a more modern look (super-heterodyne receiver) is based on the
same principle.

Effect of Image Frequency A simplified superheterodyne receiver

Dept. of Telecoms Engineering 6
Dung Trinh, PhD HCMUT / 2019

1. Introduction – Mixer Implementations

 Fundamentally, mixing (heterodyning) can happen by two mechanisms
• Non-linear circuit: use device nonlinearity.
• Linear but time-variant circuit.
 Mixers can also be classified by:
• Passive mixers.
• Active mixers.
 The nonlinear mixer can be applicable at any frequency. It is the only approach
available at the upper mm-wave frequencies.
 When frequencies are low enough that good switches can be built, the switching
mixer is preferred because it generates fewer spurs. In some cases, sampling has been
substituted for switching.
 Most mixers in use today are of the switching type, whereby diodes or transistors are
used to switch the connection between the RF input and the IF output at a rate that is
controlled by the local oscillator.

Dept. of Telecoms Engineering 7

Dung Trinh, PhD HCMUT / 2019

2. Mixers based on Nonlinearities

 Consider a square-law device (could be a properly biased diode or transistor):
2
𝑖𝑜 𝑡 = 𝐾𝑣𝑖𝑛 (𝑡)
 The output voltage on the load: Spurious signals
2
𝑣𝑜 𝑡 = 𝑅𝐿 𝑖𝑜 𝑡 = 𝑅𝐿 𝐾𝑣𝑖𝑛 𝑡 = 𝑅𝐿 𝐾 𝑣𝑅𝐹 𝑡 + 𝑣𝐿𝑂 𝑡 2
1 2 1 2
𝑣𝑜 𝑡 = 𝑅𝐿 𝐾 𝑉𝑅𝐹 cos 2𝜔𝑅𝐹 𝑡 + 𝑉𝐿𝑂 cos 2𝜔𝐿𝑂 𝑡
2 2
1 2 1 2
𝑉 + 𝑉 + 𝑉𝑅𝐹 𝑉𝐿𝑂 cos 𝜔𝐿𝑂 + 𝜔𝑅𝐹 𝑡 + 𝑉𝑅𝐹 𝑉𝐿𝑂 cos 𝜔𝐿𝑂 − 𝜔𝑅𝐹 𝑡
2 𝑅𝐹 2 𝐿𝑂

DC Components Mixing signals

 In general cases, the output current will 𝑖𝑜
contain terms that are proportional to all
powers of the total input voltage: 𝑣𝑅𝐹 +
2 𝑣𝑖𝑛 𝑅𝐿
𝑖𝑜 𝑡 = 𝐼𝐷𝐶 + 𝑘1 𝑣𝑖𝑛 𝑡 + 𝑘2 𝑣𝑖𝑛 𝑡 +⋯
𝑣𝐿𝑂 -
 In general, this produces all possible
mixing products, including: 𝑓𝑅𝐹 , 𝑓𝐿𝑂 ,
𝑓𝑅𝐹 ± 𝑓𝐿𝑂 , 2𝑓𝑅𝐹 ± 𝑓𝐿𝑂 , 𝑓𝐿𝑂 ± 𝑓𝑅𝐹 , etc. Principle of mixers based on nonlinerities
Dept. of Telecoms Engineering 8
Dung Trinh, PhD HCMUT / 2019

2. Diode Mixers based on Nonlinearities

 Diodes typically exhibit exponential I-V characteristics, which, at small signals, can
be expanded to a polynomial.
𝑞𝑉𝑑 𝑞𝑣𝑑 𝑞𝑣𝑑 2 𝑞𝑣𝑑 3
𝑖𝑑 = 𝐼𝑆 𝑒 𝑛𝐾𝑇 − 1 ≃ 𝐼𝑆 + + +⋯
𝑛𝐾𝑇 𝑛𝐾𝑇 𝑛𝐾𝑇
 Mixing by non-linear circuit tend to generate a lot of spurious signals. We are mostly
interested in 𝑓𝐼𝐹 = 𝑓𝑅𝐹 ± 𝑓𝐿𝑂 , the rest of the frequencies are considered as undesired
intermodulation products. Isolations between the three ports are provided by filters.
 Schottky diodes are predominant in mixer applications, particularly at millimeter-
wave frequencies, because their fast transition time. Still, single ended diode mixers
are rarely used because of its bad isolation between the ports. Non-linear
component
V I VRF

VIF +-
RL
VLO +-
HSCH-9162 mmwave GaAs Schottky diode Schottky diode packaging A diode mixer

Dept. of Telecoms Engineering 9

Dung Trinh, PhD HCMUT / 2019

2. Diode Mixers based on Nonlinearities

 An implementation of an unbalanced diode mixer is shown in belowed figure. There are
a lot of spurious outputs generated. Ideally, we would like to see outputs only at 10MHz
and 210MHz.
 So, we prefer the switching type mixer when the RF and LO frequencies are low
enough that we can make decent switches.
 This takes us up through much of the mm-wave spectrum.

An implementation of a diode mixer Output signal of the mixer

Dept. of Telecoms Engineering 10

Dung Trinh, PhD HCMUT / 2019

2. BJT Mixers based on Nonlinearities

 In a practical implementation, a matching network would be used between the RF
source and the RF input in order to couple the maximum amount of RF signal to the
base of the transistor.
 The matching network would be designed to present a high impedance to the LO
signal. Similarly, a matching network would be used between the LO source and the
LO input, and this network would be designed to present a high impedance to the RF
source. In this way, the LO source is prevented from loading the RF source, and vice
versa. A simpler compromise is to supply the LO voltage through a small-value
coupling capacitor (CLO) such that the LO source does not significantly disturb the
matching for the RF signal. This circuit is attractive because of its simplicity, and it
has been used in inexpensive mass-produced receivers for consumer applications.
 In general, a single-ended BJT mixer will have a higher
noise figure than a mixer that employs an FET, or a
properly designed passive mixer based on a diode bridge.
 In addition, BJT mixers are subject to severe
intermodulation distortion. To avoid this problem, vRF
must be kept smaller than 10 mV. In general, BJT mixers
of this type have poor large signal characteristics.
A BJT mixer
Dept. of Telecoms Engineering 11
Dung Trinh, PhD HCMUT / 2019

2. FET Mixers based on Nonlinearities

2
𝑣𝑔𝑠
 If a FET is operated in saturation region, then: 𝑖𝐷 = 𝐼𝐷𝑆𝑆 1 − .
𝑉𝑝

 The FET mixers has two advantages over the BJT mixers:
• A much lower 3rd IMD, since there is no cubic term in the 𝑖𝐷 versus 𝑣𝑔𝑠
relationship.
• Much higher RF input voltages are usable, i.e., up to at least 100 mV.
 A disadvantage of the FET mixer is smaller conversion gain than the BJT circuit.
 A major drawback of the single-ended BJT and FET VDC

mixers is the presence of a local oscillator component as

L1
well as an RF component at the output of the mixer.
RF
 Although the LO and RF signals will be filtered by the
C1 IF
tuned output circuit, a relatively large LO signal will BPF

inevitably leak through the mixer to the IF stages.

LO Q1
 A large LO component at the input to the IF stages is
obviously undesirable, since it can cause overload at the IF BPF
R1
stages. It can also create undesirable effects such as gain
compression and possible generation of intermodulation
products in the IF stages. A FET mixer
Dept. of Telecoms Engineering 12
Dung Trinh, PhD HCMUT / 2019

1. Introduction – Image Frequency

 Example 2: A single-ended FET mixer is to be designed for a wireless local area
network receiver operating at 2.4 GHz. The parameters of the FET are 𝑅𝑑 = 300Ω,
𝑅𝑖 = 10Ω, 𝐶𝑔𝑠 = 0.3𝑝𝐹, and 𝑔1 = 10𝑚𝑆. Calculate the maximum possible conversion
gain. Note that the conversion gain of the FET mixer is defined as
𝑉𝐼𝐹 2
𝑅 2
𝑃𝐼𝐹−𝑎𝑣𝑎𝑖𝑙 𝑍𝐿 2 𝐿 4𝑅𝐿 𝑅𝑔 𝑉𝐼𝐹
𝐺𝑐 = = =
𝑃𝑅𝐹−𝑎𝑣𝑎𝑖𝑙 𝑉𝑅𝐹 2 𝑍𝐿 2 𝑉𝑅𝐹
4𝑅𝑔
 Solution:
The conversion gain is:
𝑔12 𝑅𝑑 10 × 10−3 2 × 300
𝐺𝐶 = 2 2 = = 36.6 = 15.6𝑑𝐵
4𝜔𝑅𝐹 𝐶𝑔𝑠 𝑅𝑖 4 × 2𝜋 2 × 2.4 × 109 2 × 10

Dept. of Telecoms Engineering 13

Dung Trinh, PhD HCMUT / 2019

2. Balanced Mixers based on Nonlinearities

 A balanced configuration can be used to effectively remove the LO and/or RF signals
from the output of the mixer.
 One type of balanced mixer often used in integrated circuits uses the differential-pair
as shown in figure belows.
 If the BJT is in active mode:
𝑣𝐵𝐸
𝑖𝐶 = 𝐼𝑆 𝑒 𝑉𝑇

 For the emitter-coupled pair with identical transistors:

𝑖𝐶1 𝑣𝐵𝐸1 −𝑣𝐵𝐸2
𝑉
𝑣𝑖
𝑉𝑇
=𝑒 𝑇 = 𝑒
𝑖𝐶2
 Since the emitters are connected to a constant current source, then
𝑖𝐸1 + 𝑖𝐸2 = 𝐼𝐸𝐸 or 𝑖𝐶1 + 𝑖𝐶2 = 𝛼𝐼𝐸𝐸
 Then:
𝛼𝐼𝐸𝐸
𝑖𝐶1 = −𝑣𝑖
1+𝑒 𝑉𝑇
𝛼𝐼𝐸𝐸
𝑖𝐶2 = 𝑣𝑖
1+𝑒 𝑉𝑇
A differential pair
Dept. of Telecoms Engineering 14
Dung Trinh, PhD HCMUT / 2019

2. Balanced Mixers based on Nonlinearities

 If the output is taken between the collectors of the two transistors as in following
figure, the output voltage will be proportional to the difference of the emitter currents:
𝑣𝑖 −𝑣𝑖
𝑒 2𝑉𝑇 −𝑒 2𝑉𝑇 𝑣𝑖
Δ𝑖𝐶 = 𝑖𝐶1 − 𝑖𝐶2 = 𝛼𝐼𝐸𝐸 𝑣𝑖 −𝑣𝑖 = 𝛼𝐼𝐸𝐸 tanh
2𝑉𝑇 2𝑉𝑇 2𝑉𝑇
𝑒 +𝑒
 If the input voltage is small, 𝑣𝑖 2𝑉𝑇 ≪ 1, then:
𝑣𝑖
Δ𝑖𝐶 ≃ 𝛼𝐼𝐸𝐸
2𝑉𝑇
 If a good signal, 𝑖𝑖2 , is added:
𝛼
Δ𝑖𝐶 ≃ 𝐼 + 𝑖𝑖2 𝑣𝑖
2𝑉𝑇 𝐸𝐸

Current difference Δ𝑖𝐶 A FET balanced mixer

Dept. of Telecoms Engineering 15
Dung Trinh, PhD HCMUT / 2019

2. Balanced Mixers based on Nonlinearities

 If 𝑣𝑖1 → 𝑣𝑅𝐹 and 𝑣𝑖2 → 𝑣𝐿𝑂 , the output voltage will proportional to 𝐾 + 𝑣𝐿𝑂 𝑣𝑅𝐹 .
Note that 𝑣𝑅𝐹 signal is present at the output, along with the desired signal.
 If 𝑣𝑖1 → 𝑣𝐿𝑂 and 𝑣𝑖2 → 𝑣𝑅𝐹 , the output voltage will proportional to 𝐾 + 𝑣𝑅𝐹 𝑣𝐿𝑂 .
Note that 𝑣𝐿𝑂 signal is present at the output, along with the desired signal.
 The unwanted term can be eliminated as follows:
• We construct two identical multipliers, and drive the differential
inputs of each amplifier with 𝑣𝑖1.
• Drive the tail current input of amplifier 1 with 𝑣𝑖2 and the tail-
current amplifier of multiplier 2 with −𝑣𝑖1 .
• The output of multiplier 1 is then proportional to 𝐾𝑣𝑖2 + 𝑣𝑖1 𝑣𝑖2 ,
the output of multiplier 2 is proportional to 𝐾𝑣𝑖2 − 𝑣𝑖1 𝑣𝑖2.
• The difference between the two outputs will contain only a term
proportional to 𝑣𝑖1 𝑣𝑖2 because the unwanted term will be
canceled out.
• Multipliers based on this principle are called Gilbert cell
multipliers, and are named after Barrie Gilbert who described a
practical, and precise, multiplier circuit based on the technique.
A FET balanced mixer
Dept. of Telecoms Engineering 16
Dung Trinh, PhD HCMUT / 2019

3. Mixers based on switches

 Most mixers used today are based on switches. Let 𝑣𝑅𝐹 𝑡 = 𝑉𝑅𝐹 cos 𝜔𝑅𝐹 𝑡, then:
∞
1 sin 𝑛𝜋 2
𝑣𝑜 𝑡 = 𝑣𝑅𝐹 𝑡 𝑝 𝑡 = 𝑉𝑅𝐹 cos 𝜔𝑅𝐹 𝑡 𝑉𝐿𝑂 + 𝑛𝜋 sin 𝑛𝜔𝐿𝑂 𝑡
2 2
𝑛=1
 The coefficient sin 𝑛𝜋 2 is zero if n is even, therefore the output signal has
components at the following frequencies: 𝑓𝑅𝐹 , 𝑓𝑅𝐹 ± 𝑛𝑓𝐿𝑂 , 𝑛 = 1,3, …
1 2 1 1
𝑣𝑜 𝑡 = 𝑉𝑅𝐹 cos 𝜔𝑅𝐹 𝑡 𝑉𝐿𝑂 + sin 𝜔𝐿𝑂 𝑡 + sin 3𝜔𝐿𝑂 𝑡 + sin 5𝜔𝐿𝑂 𝑡 + ⋯
2 𝜋 3 5
1 1 1
= 𝑉𝑅𝐹 𝑉𝐿𝑂 cos 𝜔𝑅𝐹 𝑡 + sin 𝜔𝑅𝐹 + 𝜔𝐿𝑂 𝑡 − sin 𝜔𝑅𝐹 − 𝜔𝐿𝑂 𝑡
2 𝜋 𝜋
Mixing products
RF feedthrough + 1 1
sin 𝜔𝑅𝐹 + 3𝜔𝐿𝑂 𝑡 − sin 𝜔𝑅𝐹 − 3𝜔𝐿𝑂 𝑡 + ⋯ Spurs
3𝜋 3𝜋
 A filter can be used to select the desired component which is usually, 𝑓𝑅𝐹 ± 𝑛𝑓𝐿𝑂 .

𝑣𝑅𝐹 𝑣𝑅𝐹
𝑅𝐿 𝑅𝐿
𝑣𝐿𝑂 𝑣𝐿𝑂

Dept. of Telecoms Engineering 17

Dung Trinh, PhD HCMUT / 2019

3. Mixers based on switches

 There are many ways to implement a switch that is controlled by the local oscillator
signal. Diodes are often used as switches.
 If a relatively large LO voltage is impressed on a diode, the diode will be forward
biased on positive excursions of the LO signal, and will be in a low-impedance state
for superposed, small RF signals.
 On negative excursions of the LO voltage the diode is driven into a high impedance
state and behaves like an open switch.
 Transistors can also be used as switches, with the LO controlling the bias current and
determining whether the device is biased into a low-impedance state, or is cutoff.

𝑣𝑅𝐹 𝑣𝑅𝐹
𝑅𝐿 𝑅𝐿
𝑣𝐿𝑂 𝑣𝐿𝑂

Dept. of Telecoms Engineering 18

Dung Trinh, PhD HCMUT / 2019

3. Single Balanced Mixers based on switches

 A balanced LO drive can reduce RF feed-through.
 If 𝑣𝑖1 → 𝑣𝐿𝑂 , 𝑣𝑖2 → 𝑣𝑅𝐹 and allow the amplitude of 𝑣𝐿𝑂 to be large compared to 𝑉𝑇 ;
𝑣
then the term tanh 2𝑉𝑖 will be driven to +1 on positive excursions of 𝑣𝐿𝑂 and to
𝑇
− 1 on the negative excursions. In this mode of operation, the transistors are driven as
switches.
 The output voltage of the circuit is:
∞
sin 𝑛𝜋 2
𝑣𝑜 𝑡 = − 𝐼𝐷𝐶 + 𝐼𝑅𝐹 cos 𝜔𝑅𝐹 𝑡 𝑅𝑉𝐿𝑂 2 𝑛𝜋 sin 𝑛𝜔𝐿𝑂 𝑡
𝑛=1 2
 The output signal has components at the following
frequencies: 𝑛𝑓𝐿𝑂 , 𝑓𝑅𝐹 ± 𝑛𝑓𝐿𝑂 , 𝑛 = 1,3, …

Principle of switching balanced mixer A FET balanced mixer

Dept. of Telecoms Engineering 19
Dung Trinh, PhD HCMUT / 2019

3. Single Balanced Mixers based on switches

 The DC term which was responsible for the RF feedthrough in the unbalanced mixer
does not appear in the double balanced mixer:
4 1 1
𝑣𝑜 𝑡 = − 𝐼𝐷𝐶 + 𝐼𝑅𝐹 cos 𝜔𝑅𝐹 𝑡 𝑅𝑉𝐿𝑂 sin 𝜔𝐿𝑂 𝑡 + sin 3𝜔𝐿𝑂 𝑡 + sin 5𝜔𝐿𝑂 𝑡 + ⋯
𝜋 3 5
 We still get LO feedthrough if we take a single-ended output or if there is a DC
current in the signal path. There is often DC present since the output of the trans-
conductance amplifier will have a DC current component.

Output signals of a single balanced mixer

Dept. of Telecoms Engineering 20
Dung Trinh, PhD HCMUT / 2019

3. Double Balanced Mixers based on switches

 Two differential switching pairs can be combined in a double-balanced Gilbert cell
arrangement to cancel the unwanted LO component. The currents are
𝑖1 𝑡 = 𝐼𝐷𝐶 + 𝑝(𝑡)𝑖𝑅𝐹 (𝑡)
𝑖2 𝑡 = 𝐼𝐷𝐶 − 𝑝(𝑡)𝑖𝑅𝐹 (𝑡)
 The output voltage is
∞
sin 𝑛𝜋 2
𝑣𝑜 𝑡 = 𝑅 𝑖2 𝑡 − 𝑖1 𝑡 = −2𝑅𝑖𝑅𝐹 𝑡 𝑝 𝑡 = −2𝑅𝐼𝑅𝐹 cos 𝜔𝑅𝐹 𝑡 2𝑉𝐿𝑂 𝑛𝜋 sin 𝑛𝜔𝐿𝑂 𝑡
𝑛=1 2
 Hence the output contains only components at the frequencies 𝑓𝑅𝐹 ± 𝑛𝑓𝐿𝑂 .

Double balanced BJT switching mixers Double balanced diode switching mixers Principle of double balanced mixers
Dept. of Telecoms Engineering 21
Dung Trinh, PhD HCMUT / 2019

3. Double Balanced Mixers based on switches

 Single balancing got rid of the RF feedthrough which was caused by the average DC
value of the switching function. Double balancing removes the LO feedthrough as
well, since the DC term cancels.
 In real mixers, there is always some imbalance. Transistors and baluns are never
perfectly matched or balanced. These non-idealities will produce some LO to IF or
RF to IF feedthrough (thus, isolation is not perfect). This is usually specified in terms
of a power ratio relative to the desired IF output power: dBc

Output signals of a double balanced mixer

Dept. of Telecoms Engineering 22
Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications

 Image Rejection.
 Convertion gain: voltage or power.
 Port to port Isolation: dBc.
 Large signal performance:
• Gain compression: P1dB.
• Intermodulation distortion (IMD): third order intercept point (IP3).
 Small signal performance:
• Noise figure.
 Operating range: Spurious free dynamic range (SFDR).

Dept. of Telecoms Engineering 23

Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – Conversion Gain (Loss)

 Conversion loss is defined to be
𝑃𝑖𝑛
𝐿𝐶 = 10 log
𝑃𝑜𝑢𝑡
where 𝑃𝑖𝑛 is the power delivered in to the mixer at the RF frequency, and 𝑃𝑜𝑢𝑡 is IF
output power delivered to the load at the desired output frequency.
 Consider the double-balanced diode-ring mixer and assume that the diode forward
resistance is small compared to the load resistance, the output of the mixer is:
∞
sin 𝑛𝜋 2
𝑣𝑜 𝑡 = −𝑉𝑅𝐹 cos 𝜔𝑅𝐹 𝑡 2𝑉𝐿𝑂 𝑛𝜋 sin 𝑛𝜔𝐿𝑂 𝑡
𝑛=1 2
 If a filter selects just one output component, the peak amplitude of the output voltage:
sin 𝑛𝜋 2
𝑉𝐼𝐹 = 𝑉𝑅𝐹 𝑉𝐿𝑂 𝑛𝜋
2
 The conversion loss, for any n when 𝑉𝐿𝑂 = 1, is therefore
𝑛𝜋
𝐿𝐶 = 20 log 2
sin 𝑛𝜋
2
 Active mixers may give gain; while passive mixers (e.g.
diode mixers) give loss. Conversion loss of harmonics
Dept. of Telecoms Engineering 24
Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications - Isolation

 Isolation (feed-through): insertion loss between ports (at the same frequency)
 Important ones: LO  RF feed-through, LO  IF feed-through.
• LO to RF leakage can be quite serious in direct conversion receiver architectures
because it will remix with the RF and produce a DC offset.
• Large LO to IF leakage can degrade the performance of a mixer post-amp if it is
located prior to IF filtering.
 Not so important ones:
• RF IF: typically too small compared with the IF signals
• IF  RF: typically too small compared with the LO  RF feed-through.
• IF  LO and RF  LO: typically too small to interfere with the LO circuit.

Dept. of Telecoms Engineering 25

Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – Gain Compression

 Conversion gain degrades at large input signal levels due to nonlinearity in the signal
path.
 Assume a simple nonlinear transfer function
3 3
𝑣𝑅𝐹 𝑡 = 𝑉𝑅𝐹 sin 𝜔𝑅𝐹 𝑡 − 𝑎3 𝑉𝑅𝐹 sin 𝜔𝑅𝐹 𝑡
3 3
3𝑎3 𝑉𝑅𝐹 𝑎3 𝑉𝑅𝐹
𝑣𝑅𝐹 𝑡 = 𝑉𝑅𝐹 1− sin 𝜔𝑅𝐹 𝑡 + sin 3𝜔𝑅𝐹 𝑡 + ⋯
4 4
 This distortion then gets mixed to the IF frequency.

Dept. of Telecoms Engineering 26

Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – IMD Distortion

 Intermodulation distortion occurs when two or more signals are present at the RF input
to the mixer. The LO input is provided as before. These two signals can interact with the
nonlinearities in the mixer signal path (RF to IF) to generate unwanted IMD products
(distortion) which then get mixed down to IF.
3
 Let’s consider the 3rd order nonlinearity, 𝑎3 𝑣𝑅𝐹 𝑡 = 𝑎3 𝑉𝑅𝐹1 cos 𝜔𝑅𝐹1 𝑡 +

3rd order IMD

 Third order products at 2𝑓𝑅𝐹1 − 𝑓𝑅𝐹2 and

2𝑓𝑅𝐹2 − 𝑓𝑅𝐹1 is generated at 899.98MHz
and 900.02MHz.
 Once multiplied with the LO frequency,
these IMD products may fall within the filter
bandwidth of the IF filter and thus cause
interference to a desired signal.
Dept. of Telecoms Engineering 27
Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – IMD Distortion

 In reality, the intercept power is 10 to 15 dBm higher than the P1dB gain compression
power.
 We have:
𝑃𝐼𝐹 − 𝑃𝐼𝑀𝐷
𝑂𝐼𝑃3 = 2 + 𝑃𝐼𝐹
𝑂𝐼𝑃3 = 𝐼𝐼𝑃3 + 𝑐𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛 𝑔𝑎𝑖𝑛

Dept. of Telecoms Engineering 28

Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – SFDR – Definition #1

 This figure shows the output of a hypothetical circuit with a single-tone input at 𝜔𝑖𝑛 .
In addition to the desired output at 𝜔𝑖𝑛 , there are several different frequency
components (spurs) in the output spectrum.
 The spurious-free dynamic range (SFDR) specification is defined as the ratio of the
desired signal amplitude to the largest spur over the bandwidth of interest.
 Since the SFDR definition specifies the maximum spur amplitude in reference to the
signal (or carrier) level, we express the SFDR in dBc (dB relative to the carrier).
 Sometimes, we prefer to express the maximum spur level in reference to the full scale
value of the circuit (dBFS).

Illustration of SFDR
Dept. of Telecoms Engineering 29
Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – SFDR – Definition #2

 Minimum Detectable Signal (MDS) power is determined by noise and corresponds to
a signal whose strength just equals the noise:
𝑀𝐷𝑆 = 10 log 𝑘𝑇𝐵 + 𝑁𝐹
 SFDR refers to the output power range where no third-order products are observed.
 The SFDR is a commonly used figure of merit to describe the dynamic range of an
RF system:
2
𝑆𝐹𝐷𝑅 = 𝑃𝑖𝑛,𝑚𝑎𝑥 − 𝑃𝑛𝑜𝑖𝑠𝑒 = 𝐼𝐼𝑃3 − 𝑀𝐷𝑆
3

Input MDS
Linearity and Dynamic Range are two of the most critical design considerations for practical devices
Dept. of Telecoms Engineering 30
Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – SFDR Definition in RF Design

 In RF design, there is another definition for the SFDR that is used to characterize the
linearity of a receiver.
 The SFDR is defined as the maximum tolerable signal power divided by the minimum
tolerable signal power. The minimum power is often referred to as the receiver
sensitivity 𝑃𝑠𝑒𝑛𝑠𝑖𝑡𝑖𝑣𝑖𝑡𝑦 and is given by
𝑃𝑠𝑒𝑛𝑠𝑖𝑡𝑖𝑣𝑖𝑡𝑦 = −174 + 10 log 𝐵 + 𝑁𝐹 + 𝑆𝑁𝑅𝑚𝑖𝑛 = 𝑀𝐷𝑆 + 𝑆𝑁𝑅𝑚𝑖𝑛
 SFDR is defined as
𝑆𝐹𝐷𝑅 = 𝑃𝑖𝑛,𝑚𝑎𝑥 − 𝑃𝑠𝑒𝑛𝑠𝑖𝑡𝑖𝑣𝑖𝑡𝑦 = 𝑃𝑖𝑛,𝑚𝑎𝑥 − 𝑀𝐷𝑆 − 𝑆𝑁𝑅𝑚𝑖𝑛
2
𝑆𝐹𝐷𝑅 = 𝐼𝐼𝑃3 − 𝑀𝐷𝑆 − 𝑆𝑁𝑅𝑚𝑖𝑛
3

Dept. of Telecoms Engineering 31

Dung Trinh, PhD HCMUT / 2019

1. Introduction – Image Frequency

 Example 3: A GSM receiver requires a minimum SNR of 12dB and has a channel
bandwidth of 200kHz. A wireless LAN receiver, on the other hand, specifies a
minimum SNR of 23dB and has a channel bandwidth of 20MHz. Compare the
sensitivities of these two systems if both have an NF of 7dB.
 Solution:
Sensitivity of GSM system is:
𝑃𝑠𝑒𝑛 = 10 log 𝑘𝑇𝐵 + 𝑁𝐹 + 𝑆𝑁𝑅 = −174 + 53 + 7 + 12 = −102 𝑑𝐵𝑚
Sensitivity of WLAN system is:
𝑃𝑠𝑒𝑛 = 10 log 𝑘𝑇𝐵 + 𝑁𝐹 + 𝑆𝑁𝑅 = −174 + 73 + 7 + 23 = −71 𝑑𝐵𝑚
The WLAN employs a much wider bandwidth and a more efficient modulation to
accommodate a data rate of 54 Mb/s. The GSM system handles a data rate of only 270 kb/s.

 Example 4: Compute the SFDR of GSM system if it has 𝑃𝐼𝐼𝑃3 = −15𝑑𝐵𝑚.

 Solution:
The SFDR of GSM system is:
2 2
𝑆𝐹𝐷𝑅 = 𝐼𝐼𝑃3 − 𝑀𝐷𝑆 − 𝑆𝑁𝑅𝑚𝑖𝑛 = −15 + 114 − 12 = 54𝑑𝐵
3 3
Dept. of Telecoms Engineering 32
Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – Examples

Internal LO amplifier allows a small

LO input

All ports are differential

Analog Devices ADL5801

Dept. of Telecoms Engineering 33

Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – Examples

Analog Devices ADL5801

Impedance Matching

Dept. of Telecoms Engineering 34

Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – Examples

Analog Devices ADL5801

Isolation

Dept. of Telecoms Engineering 35

Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – Examples

The level of a mixer is basically the required LO signal level

Minicircuits MAC-80H+
Dept. of Telecoms Engineering 36
Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – Examples

Minicircuits MAC-80H+
Dept. of Telecoms Engineering 37
Dung Trinh, PhD HCMUT / 2019

4. Mixer Specifications – Examples

Marki Mixers

Mmwave M9-0444LNV UBB M1-0008

Dept. of Telecoms Engineering 38
Dung Trinh, PhD HCMUT / 2019

5. Exercises

Exercise 1: Consider the mixer circuit shown below.

a. Assume that 𝑣𝐿𝑂 controls the state of the diodes at all times and that the diodes are ideal.
Derive an expression for the output voltage. Express your result in terms of a switching
function 𝑝(𝑡), and give the Fourier cosine series for your switching function.
b. What frequency components will appear at the output of this mixer?

Dept. of Telecoms Engineering 39

Dung Trinh, PhD HCMUT / 2019

5. Exercises

Exercise 2: Consider the mixer circuit shown below. The transformer is an ideal 3-
winding transformer, which has the property that the voltages across all 3 windings are
equal, with polarity corresponding to the dots shown next to each winding. The time-
varying RF and LO voltages are 𝑣𝑅𝐹 𝑡 = 𝑉𝑅𝐹 cos 𝜔𝑅𝐹 𝑡, and 𝑣𝐿𝑂 𝑡 = 𝑉𝐿𝑂 cos 𝜔𝐿𝑂 𝑡.
Assume that 𝑣𝐿𝑂 𝑡 controls the state of the diodes at all times, and that the diodes are
ideal switches.
a. Derive an expression for the IF voltage, 𝑣𝐼𝐹 𝑡 . Express your result in terms of a
switching function 𝑝(𝑡), and give the Fourier cosine series for your switching function.
b. List the frequencies of all components that will appear at the output of this mixer.

Dept. of Telecoms Engineering 40

Dung Trinh, PhD HCMUT / 2019

Q&A

Dept. of Telecoms Engineering 41