Vous êtes sur la page 1sur 40

Advanced Theory and Technology of Devices Politecnico di Torino

Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008


1. Introduction to RF and microwave integrated circuits
1) Define RF, microwaves and millimeter waves from the standpoint of frequency allocation.

Type Frequency Range Wavelength Range


RF Few MHz and 1 GHz Order of 1 m
Microwaves 1 GHz to 30 GHz 30 cm to 1 cm
Millimeter 30 GHz to 100 (300) GHz Below 1 cm
The extension to 300 GHz for the millimeter waves is usually cut off to 100 GHz since no use are done above that
frequency value.

2) Suppose that a radar has to be designed to detect objects of the average size of 1 cm: is an RF operating
frequency adequate to this? Explain why/why not. Suggest if the case a more suitable frequency range.

RF frequencies are not adequate to this purpose since the dimension of the object you want to detect has to be much
bigger than the wavelengths used (in this case, λ is around 1 m).
A suitable choice could be the usage of millimeter waves.

3) A dielectric medium has εr=9. Evaluate the free-space wavelength at 10 GHz and the wavelength in the dielectric
medium.

In the free-space exists the relationship among the wavelength, the frequency and the light velocity:

From this equation, it is possible to compute the free-space wavelength:

The wavelength in the dielectric medium can be computed by means of the following formula:

4) Identify the L and the K bands (frequency limits). In what frequency band are GSM cellphones operating?

Following the IEEE Standard, the L band is allocated between 1 GHz and 2 GHz while the K band between 18 GHz and
27 GHz.
GSM cellphones are operating in the L band.

5) Explain why signals cannot conveniently be transmitted in baseband through a Herzian channel, but rather they
have to be upconverted through analog or digital modulations. Assume as an example e.g. a hi-fi signal with
frequency between DC and 20 KHz.

Since a baseband signal implies a great wavelength and an efficient antenna must have a dimension comparable to it,
it is not convenient to transmit the signal at those frequencies because you need enormous antennas.
For the example we get:
c0 3 *108
λ= = = 15km
f 20 *10 3
As you can see, such an antenna is not reasonable.
Even if the antenna can be realized, it is not convenient to transmit baseband signals because such waves are very
little attenuated and will be spread worldwide.

6) Explain why transmitting the human voice in baseband through a portable phone would be for many reasons
unpractical.

1
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
The main reason for which transmitting human voice in baseband on a portable phone would be unpractical is the size
of the antenna (which has to be comparable to the wavelength).

7) In a cellular system each cell exploits the same frequency channels (e.g. around 2 GHz), they are allocated to
different users. Explain why there is no interference between nearby cells.

The same frequencies can be reused since the Herzian channel provides enough attenuation between radio-base
stations.
This is possible for RF higher than 1 GHz and microwaves.

8) Describe a basic RF RX/TX (receiver/transmitter) scheme.

This is the scheme of a basic RF transceiver:

The modulation/demodulation blocks are used to swap between BB/RF and RF/BB.
Amplification towards the antenna provides the power necessary to transmit the information while the one towards
the BB section increases the received signal level (normally low noise amplification LNA).
The duplexer switches between RX and TX mode.
Filters are used to select multi-channel bands, channels, to reject image frequencies and to cutoff noise.

9) List the basic RX section building blocks, starting from the antenna down to the downconversion mixer.

The building blocks are the following:


• Antenna
• Duplexer
• Band select filter
• LNA: low noise amplifier
• Image rejection filter: only in heterodyne
• LO: local oscillator
• Mixer

10) List the basic TX section building blocks, starting from the upconversion mixer up to the antenna.

The building blocks are the following:


• Mixer
• LO: local oscillator
• Band pass filter
• PA: power amplifier
• Duplexer

2
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
• Antenna

11) What is the difference between a homodyne and a heterodyne receiver?

Homodyne receiver:
Also named direct conversion, it directly converts RF to baseband.
Heterodyne receiver:
Before converting to baseband, it downconverts the RF to intermediate frequencies (IF) at least one time.

12) What is the difference between a low noise, high gain and maximum power amplifier?

The scheme of the three amplifiers can be the same but their purpose is different.

13) What are the typical features of planar vs waveguide microwave circuits?

Planar circuits are characterized by low cost and small size; moreover, the approach used in designing them is the
same as low-frequency ICs or hybrid circuits.
Waveguide microwave circuits imply large size and cost.

14) Explain the differences between a hybrid and a monolithic microwave circuit.

MMIC (monolithic) HIC (hybrid)


Cheap in large quantities Simple circuits can be cheap
Very good reproducibility Poor reproducibility: device placement - bond wires
Small and light Larger and heavier
Reliable Mostly “glued” together and so reliabilities suffers
Less parasitic – more BW and higher frequency The best transistors are always available
Space is premium – must be as small as possible Substrate is cheap – µ-strip to be used abundantly
Long turn-around-time (3 Months) Can be very fast (1 week), making redesigns easy

15) List in order of increasing frequency range the following semiconductors: indium phosphide, silicon, gallium
arsenide.

Material Frequencies
Silicon (Si) 5 – 10 GHz
Gallium Arsenide (GaAs) Up to 50 GHz
Indium Phosphide (InP) Up to millimeter waves

16) Explain the difference between a lumped and a distributed-parameter circuit. Why distributed elements cannot
be integrated in an RF circuit?

Lumped Distributed
Elements small vs wavelength (<λ/8) Based on transmission line approach (typ. >λ/8)
Typ. integrated on Si (up to few GHz), GaAs (up to 50 Typ. hybrid; the circuit is made on a dielectric substrate
GHz), InP (up to 100 GHz) with thin film technique

Monolithic integration of distributed components is not feasible at RF for their large layout.

17) Quote a few microwave field-effect or bipolar transistor with the related semiconductor material.

Field-effect transistors Bipolar transistors


MOSFETs – Si BJTs – Si
MESFETs and HEMTs – GaAs HBTs – SiGe
HEMTs - InP HBTs - GaAs and InP

3
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
18) Explain the difference between analog large-signal and small-signal models. Clarify what model is linear and
what model is nonlinear.

Small-signal model is linear and it does not take in account nonlinearities such as harmonic generation,
intermodulation products and it is reasonable only for low input levels.
Large-signal model is nonlinear and it considers all the nonlinear effects.

19) Explain the difference between a memoriless model and a model with memory.

In a memoriless model (non dispersive) the transfer function has no dependence on frequency, so it is constant (e.g.
resistors).
A model with memory (dispersive) has a transfer function that depends on frequency (e.g. capacitors and inductors).

4
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
2. Passive components for RF and microwave circuits; Distributed elements;
Lumped elements
1) Explain the difference between a TEM and a quasi-TEM transmission line.

TEM transmission line Quasi-TEM transmission line


Homogeneous cross-section, minimum two conductors, E Non-homogeneous cross section, E and H fields
and H fields in the cross section transversal only at low frequencies
Frequency independent propagation parameters Slightly dispersive propagation parameters
Low frequency (RC) dispersion in lossy lines: Low frequency (RC) dispersion in lossy lines:
c0 c0
v f ≈ vg = v f ≈ vg =
εr ε eff

2) A lossy transmission line has per-unit-length parameters L, C, R, G. Express the characteristic impedance and
complex propagation constant of the line in terms of the parameters for the general case and in the high-frequency
approximation. Identify, in the high-frequency approximation, the propagation constant and the attenuation.

General case:
jω L + R
Z∞ = γ = α + jβ = ( jωL + R)( jωC + G )
jωC + G
High-frequency approximation:
L R GZ 0
Z∞ = γ = α + jβ ≈ + + jω LC
C 2Z 0 2
R GZ 0
α = αc + αd = + β = ω LC
2Z 0 2
It is reasonable to neglect the dielectric attenuation α d term since it is much smaller than the conductor attenuation
αc .

3) A lossless quasi-TEM line has a 50 Ω impedance and an effective permittivity εeff=2. Evaluate the per-unit-length
parameters L, C. Compute the guided wavelength at 10 GHz.

The phase velocity for a quasi-TEM line can be obtained by:

From the line theory, we can derive a system useful to compute L, C:

In order to resolve the system, it’s a good way multiplying and dividing the two equations:

To compute the guided wavelength you have to compute the free-space one:

5
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

4) A lossy quasi-TEM line has a 50 Ω impedance. The dielectric attenuation is 0.1 dB/cm while the conductor
attenuation is 1 dB/cm at 1 GHz. Evaluate the per-unit-length parameters R, G. Estimate their values and the
resulting dielectric and conductor attenuation at 10 GHz. Assuming an effective permittivity εeff=7, evaluate the
total loss over 1 guided wavelength at 10 GHz.

First of all, you need to compute attenuation values in natural form:

After that, it is possible to compute R, G from the equations that relate each attenuation to each parameter:

At 10 GHz, the value of R increases of a factor equal to the square of 10 with respect to 1 GHz while the value of G
increases of a factor equal to 10 with respect to 1 GHz:

The total attenuation can be obtained by:


α = α c + α d = 3.16 + 1 = 4.16dB / cm
To compute the guided wavelength you have to compute the free-space one:

The total attenuation over 1 guided wavelength is equal to:


α = 4.16 *1.13 = 4.7dB
5
5) The conductivity of a 2 µm thick conductor is σ = 1 x 10 S/m. Evaluate the frequency at which the skin-effect
penetration depth is equal to the conductor thickness.

The skin depth can be obtained by the following formula:

6) A lossless transmission line with 50 Ω characteristic impedance and 5 mm guided wavelength is closed on ZL = 50
+ j50 Ω. Compute the input impedance for a 2.5 and 1.5 mm long line.

The reflection coefficient can be computed by:

6
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
7) A lossless line is infinitely long. Is the input impedance always equal to the characteristic impedance? Explain.

The expression of the input impedance of a transmission line is equal to:


Z L cosh(γl ) + Z 0 sinh(γl )
Z in = Z 0
Z 0 cosh(γl ) + Z L sinh(γl )
For an infinitely long transmission line, the input impedance is always equal to the characteristic impedance of the line
since:
 Z cosh(γl ) + Z 0 sinh(γl )   cosh(γl ) Z L + Z 0 tanh(γl )  Z L + Z0
lim Z in = lim  Z 0 L  = lim  Z 0 *  = Z0 = Z0
l →∞ l →∞  Z 0 cosh(γl ) + Z L sinh(γl )  l →∞  cosh(γl ) Z 0 + Z L tanh(γl )  Z0 + Z L

8) A quasi-TEM line has a per-unit-length capacitance of 5pF/mm and an in-vacuo capacitance of 2pF/mm. What is
the effective permittivity?

9) Sketch the cross section of a microstrip and of a coplanar waveguide.

10) A microstrip on 0.5 mm thick alumina substrate has a strip width of 0.5 mm. What is (approximately) the
characteristic impedance?

The characteristic impedance of a microstrip depends on the material used as substrate (in this case alumina, εr=10)
and the ratio W/h (in this case equal to 1).
From the following graph which relates Z0 to those parameters, you can say that the characteristic impedance is
approximately 50 Ω.

11) Sketch the attenuation of a microstrip and of a coplanar waveguide as a function of the strip width.

Microstrip attenuation:

7
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

Coplanar waveguide attenuation:

12) Sketch the behavior of the attenuation of a transmission line as a function of frequency.

13) Sketch a strip, a loop and a spiral inductor. What usually limits the frequency range on which integrated RF and
microwave inductors can operate?

Strip inductor:

8
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

Loop inductor:

Spiral inductor:

Frequency range on which integrated RF and microwave inductors can operate is limited by the dimension of the
inductor since the bigger the inductor the lower the resonant frequency.

14) List some possible uses of inductors in integrated RF circuits.

• Bias T
• LNA
• Matching
• Series inductive feedback

15) Sketch an interdigitated and a MIM capacitor.

Interdigitated capacitor:

MIM capacitor:

2
16) In a MIM capacitor the dielectric is 100 nm thick, width permittivity equal to 2. What is the capacitance per mm
area?

9
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

C ε 0 ε r 8.854 * 10 −12 * 2
= = −9
= 177.08µF / m 2
A d 100 * 10
C 177.08 * 10 −6
= 6
= 177.08 pF / mm 2
A 10
17) What are chip inductors, capacitors and resistors? Are they used in hybrid or integrated implementations?

They are external components and they are used only in hybrid implementations (normally they are mounted with
surface mount techniques).

18) What is a coaxial-to-microstrip transition?

A coaxial-to-microstrip transition is a connection between a coaxial cable and a microstrip circuit. It can be performed
in various ways:
• Solder
• Wire
• Band
• Without any soldering, just with a contact spring
These techniques are used to limit losses.

19) What are the main differences between a coplanar and a microstrip circuit layout?

Coplanar circuit layout Microstrip circuit layout


Ground plane available on the surface Ground plane on the bottom plane
Easy short circuit stubs Ground plane reachable only through vias
Used only for integrated circuits For MMIC and HIC
More bulky Smaller layout
Difficult heat dissipation Easier heat dissipation

3. Scattering parameters, gains, stability

1. The open-circuit voltage of a real generator is V0 = 10 V and the internal impedance is ZG = 50+j50. What is the
generator available power? What is the load impedance yielding power match to the generator?

V02RMS 10 2
Pav = = = 0.5W
4 RG 4 * 50
Z L = Z G* = (50 − j 50)Ω
2. Suppose the normalization impedance is 50 Ω. Locate on a Smith chart (approximately) the following
impedances: ZL1 = 50 Ω; ZL2 = 50 + j50 Ω; ZL3 = 50 - j50 Ω; ZL4 = 100 Ω; ZL5 = 25 Ω; ZL6 = 0 Ω; ZL7 = ∞ Ω.

10
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

ZL2

ZL6 ZL5 ZL1 ZL4 ZL7

ZL3

3. A transmission line with length equal to λg/4 is loaded with a 100 Ω impedance. The characteristic impedance is
50 Ω. Locate on the Smith chart the load impedance and input impedance. What is Zin? This structure is called a
quarter-wavelength transformer.

11
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

Zin ZL

The input impedance is equal to the admittance of the impedance put as a load.

4. A resistive two-port has the following impedance matrix:

Sketch a possible structure (implementing the above impedance matrix) and evaluate the scattering matrix (assume
the normalization impedance R0 = R).

12
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

1 1
 
S = (Z − R0 I )(Z + R0 I ) =4 4
−1

 1 1

4 4
5. A reactive two-port has the following impedance matrix:

Evaluate the scattering matrix assuming R0 = X and check that the properties of the S-matrix of a lossless two-port
are verified.

 0 .4 + i 0 .8 0 .4 − i 0 .2 
S = (Z − R0 I )(Z + R0 I ) = 
−1

 0 .4 − i 0 .2 0 .4 + i 0 .8 
To verify the lossless property, you have to use the following formula:
ST S* = I
In this case, the condition is verified.

6. A real generator has internal impedance ZG = 50-j50 Ω and open circuit voltage V0 = 10 V. Assuming R0 = 50 Ω
derive the power wave equivalent circuit (ΓG and b0).

R0 50
b0 = V0 = 10 = 0.5657 + i0.2828 [W1/2 ]
Z G + R0 50 − i50 + 50
Z − R0 50 − i50 − 50
ΓG = G = = 0.2 - i0.4
Z G + R0 50 − i50 + 50
7. A load exhibits a reflection coefficient Γ equal in magnitude to -10 dB. What part (in %) of the incident power is
reflected?
The reflection coefficient is obtained by means of the following equation:

13
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
10
V− −
Γ = + = 10 20
V
Since the relationship between power and voltage is a square proportional:
P−
ΓP = = Γ 2 = 10 −1 = 0.1 ⇒ 10%
P+
1/2
8. A real generator with ΓG = 0.2 and b0 = 1 W is connected to a load with ΓL = 0.5. Evaluate the power delivered
to the load and the maximum available power of the generator.

2
2 1 − ΓL 1 − 0.5 2
PL = b0 2
= = 0.93W
1 − ΓG ΓL (1 − 0.2 * 0.5) 2
2
b0 1
Pav = 2
= = 1.04W
1 − ΓG 1 − 0.2 2
9. A loaded two-port has the following characteristics: Pin = 10 mW; Pav,in = 20 mW; PL = 100 mW; Pav,out = 300 mW.
Evaluate the two-port gains Gop, Gav, Gt.

PL 100 * 10 −3
Gop = = = 10
Pin 10 * 10 −3
Pav ,out 300 * 10 −3
G av = = = 15
Pav ,in 20 * 10 −3
PL 100 * 10 −3
Gt = = =5
Pav ,in 20 * 10 −3

10. A two-port has the following scattering matrix (R0 = 50 Ω):

Evaluate the two-port MAG. Is the two-port unilateral?

MAG =
S 21
S12
(
K − K 2 −1 = 0 )
The device is unilateral from port 2 to port 1 (null forward).

11. A two-port has the following scattering matrix (R0 = 50 Ω):

Compute the input and output reflection coefficients when the two-port is loaded on 100 Ω.

Z L − R0 100 − 50 1
ΓL = = =
Z L + R0 100 + 50 3
1
10 * 0.01 *
b S S Γ 3 = 0.134
Γin = 1 = S11 + 12 21 L = 0.1 +
a1 1 − S 22 ΓL 1
1 − 0.1 *
3
Supposing a generator with Z G = 100Ω on port one, you get:

14
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
Z G − R0 100 − 50 1
ΓG = = =
Z G + R0 100 + 50 3
1
10 * 0.01 *
b S S Γ 3 = 0.134
Γout = 2 = S 22 + 12 21 G = 0.1 +
a2 1 − S11ΓG 1
1 − 0.1 *
3
12. Maximum power transfer between generator and load through a two-port implies simultaneous power
matching at both ports. Is this condition always possible?

This condition can be achieved only if the stability factor is equal to or larger than one.

13. A two-port has K = 2; S21 = 15(1+j) and S12 = 0.1: Evaluate the two-port MAG and MSG. Assume the two-port is
unconditionally stable.

MAG =
S 21
S12
(
K − K 2 −1 = )
15(1 + j )
0.1
(
2 − 4 − 1 = 56.84 )
S 21 15(1 + j )
MSG = = = 212.13
S12 0.1

14. A two-port has K = 2.5; |∆S| = 1.5. Is the two-port unconditionally stable?

No because even if K is larger than one, the module of ∆S has to be lower than one.
Sometimes, for practical purposes, K > 1 can be a sufficient condition, but theoretically both the equations have to be
respected.

15. Discuss the stability (according to the one- and two-parameter criteria) of the two-port with scattering matrix:

Two-parameter criteria:
The stability can be evaluated by means of K and ∆S:
∆ S = S11 S 22 − S12 S 21 = 1.00005 > 1 ⇒ condition _ not _ respected
2 2 2
1 − S 22 − S11 + ∆ S
K= = 0.99 < 1 ⇒ condition _ not _ respected
2 S 21 S12
One-parameter criteria:
2
1 − S11
µ1 = = 0.8678 < 1 ⇒ condition _ not _ respected
S 22 − S11* ∆ S + S12 S 21
The two-port is not unconditionally stable.

16. Discuss the stability (according to the one- and two-parameter criteria) of the unilateral two-port with
scattering matrix:

The two-port is not unconditionally stable because |S11| is higher than one.

17. Is a unilateral device always unconditionally stable?

15
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
A unilateral device is always unconditionally stable if |S11| and |S22| are lower than one.

18. Suppose a device in unconditionally stable above f0 and potentially unstable below f0. Qualitatively sketch the
behavior of the device MAG and MSG as a function of frequency.

Gain

MSG MAG

f0 f
unstable stable

19. Consider two passive two-ports, one reactive (lossless), the other resistive (lossy). What kind of property do we
expect from their stability factors?

If the two-port is reactive (lossless) you get a borderline condition where K Is equal to one.
If the two-port is resistive (lossy) you cannot say anything a priori, since the module of ∆S could be greater or smaller
than one.

4. Active RF and microwave components

1) Sketch the equilibrium band diagram (Ec, Ev, EF ) of an intrinsic semiconductor assuming Nc = Nv. Repeat for an n-
doped and p-doped semiconductor. Repeat for a semiconductor under high injection conditions (sketch the position
of the quasi-Fermi levels).

16
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
The following sketches are the equilibrium band diagram of an intrinsic semiconductor assuming NC=NV, n-
doped and p-doped (it is not possible to have degenerate cases at thermodynamic equilibrium):

EF

The following sketches are referred to high injection condition:

2) Quote some compound semiconductors: wide-gap, narrow-gap, general purpose. What are wide-gap
semiconductors for?

Wide-gap Narrow-gap General purpose


GaN GaSb GaAs
SiC CdTe SiGe
HgTe Si
InP

Wide-gap semiconductors (Eg > 2eV) are used for high power applications.

3) Explain the composition of the AlGaAs and InGaAsP alloys.

(GaAs )1− x + ( AlAs )x = Al x Ga1− x As


α + β +γ +δ =1
(GaAs)α + (InP)β + (InAs)γ + (GaP) δ = Inβ +γ Gaα +δ Asα +γ Pβ +δ = In x Ga1− x As y P1− y

4) Explain the purpose of compound semiconductor alloys.

17
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
Semiconductor alloys are useful to optoelectronics (specific wavelengths emittance) or to carry out some energy band
engineering useful to specific devices.

5) Suppose you want to grow a lattice-matched layer of InGaAs on an InP substrate. What is the In composition?
Repeat for InAlAs.

In order to obtain lattice matching between InGaAs and InP substrate, you have to move along the previous graph
until you reach the following composition:
Ga 0.47 In0.53 As
The same thing for InAlAs and InP:
Al 0.48 In0.52 As

6) Explain the meaning of lattice-matched and pseudomorphic heterostructures.

Lattice-matched means that both the layer and the substrate have very similar lattice constant. This is necessary to
grow alloys on substrates.
Pseudomorphic heterostructure is obtained by forcing the layer to assume the same reticular structure of the
substrate.

7) List the following substrates in order of availability and maturity (one of the materials is not an available
substrate today!): GaAs, GaN, SiC, Si, InP.

Si, GaAs, InP, GaN (not yet mature), SiC (not yet available).

18
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
8) Explain why a GaAs substrate is semi-insulating, while a Si substrate is not.

Insulating characteristics are determined by the carriers quantity. Non-doped silicon has a much greater number of
carriers with respect to GaAs (which has a higher energy gap).

9) Sketch the bandstructure (energy-momentum relation) of a direct-bandgap vs. an indirect-bandgap


semiconductor.

10) Qualitatively explain why the velocity-field characteristic of electrons in a compound semiconductor first
increases, then decreases and finally saturates.

The trend shown in the graph above for InP and GaAs (compound semiconductors, direct bandgap) is explainable
using the following graph:

19
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
First of all, applying low electric field, electrons reach the Γ zone that is characterized by carriers high speed. Then, as
the electric field increases, electrons tend to move to the X and L zones which are characterized by lower speed
7
(saturation value 10 cm/s).

11) Define a heterostructure and sketch the related band diagram.

A heterostructure is obtained by joining together two different materials with different energy gaps. This produces a
discontinuity in energetic levels as shown below:

12) Explain how a quantum well can be exploited as the conducting channel of a field-effect transistor.

Since the quantum well can be seen as a confining structure for carriers, it is used as a channel that only allows
carriers’ movement along the source-drain direction (FET transistor).

The picture above shows the energetic barriers that block the carriers’ movement along one direction.

13) Sketch the structure of a GaAs MESFET and the DC characteristics of the device.

20
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

14) Sketch the structure of a GaAs-based PHEMT and the DC characteristics of the device.

21
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
15) Sketch the structure of an InP-based PHEMT and the DC characteristics of the device.

16) List in order of performance (cutoff frequency) the following devices: InP-based PHEMT, GaAs-based
conventional HEMT, Si-based MOSFET, GaAs-based MESFET, GaAs-based PHEMT.

Si-based MOSFET, GaAs-based MESFET, GaAs-based conventional HEMT, GaAs-based PHEMT, InP-based PHEMT.

17) What is the heterostructure bipolar transistor? Why are the frequency performances of this device better than
the ones of a conventional bipolar?

HBTs use a narrowgap base and collector to allow base doping to be larger than emitter doping without
compromising the emitter efficiency.

Increasing the base doping lowers the base resistance, leading to an increase of the maximum frequency of
oscillation.
High cutoff frequencies and low transit time are obtained through thin epitaxial bases. As a consequence,
very narrow emitter is needed to decrease the base resistance.

22
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

18) In a SiGe HBT-based circuit, where do you find Ge?

19) Sketch the cross-section of a GaN-based HEMT. Quote a few advantages of widegap semiconductors.

Advantages of widegap semiconductors are:


• High breakdown field
• High thermal conductivity
• Wide bandgap
• High electron velocity

23
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

5. Small- and large-signal RF FET circuit models


1) Explain what are physics-based models, circuit models, black-box models.

Physics-based models solve the transport and Poisson equations at a microscopic level in a way that could be
analytically-approximately or numerically-exactly.

Circuit models are approximate but efficient and they can be fitted on measured data or based on analytical PB
models.

Black-box models are mathematical models entirely derived from measured data, such as S parameters.

24
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
2) Sketch a MESFET small-signal equivalent circuit and separate the extrinsic parasitics from the intrinsic model.

3) Sketch the behaviour of the scattering parameters of a microwave FET.

S11 and S22 are reflection coefficients, so they can be drawn in the Smith chart:

S12 and S21 are transmission coefficients, so they can be drawn in the polar diagram:

25
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

4) Define the cutoff frequency of a FET and evaluate it from the intrinsic FET parameters.

The cutoff frequency is defined when the short-circuit current gain is equal to 1:

CGD is typically negligible in this analysis.

The cutoff frequency depends on the length of the gate in the way that it grows with device scaling down:

5) Define the maximum oscillation frequency of a FET.

The maximum oscillation frequency of a FET is defined when the MAG is equal to 1.
In practice, beyond fmax the device cannot operate and according to the application, the suggested maximum
frequency can be lower or much lower than fT.

6) Derive the fmax of a unilateral FET from the intrinsic parameters.

26
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
Considering the circuit power matched, thus and , it is possible to compute powers
introduced in the circuit:

7) What is the typical behaviour of the MAG as a function of frequency?

The MAG decreases as the square of frequency:

8) Explain how the FET equivalent circuit can be extracted from measured S-parameters.

The FET equivalent circuit parameters can be computed from the Y intrinsic parameters, which can be derived from
scattering parameter measurements of the whole device through parasitic de-embedding. Parasitic resistances and
inductances can be directly measured on the device in which the drain-to-source bias is set to zero and the gate is in
direct bias.
The de-embedding procedure is the following:

9) Sketch the large-signal equivalent circuit of a FET in DC and explain the relationship of the output generator to
the device transconductance and output resistance.

27
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

10) Explain how a nonlinear capacitor can be derived from small-signal measurements of the same.

A nonlinear capacitance is defined by a nonlinear charge-voltage relationship; in small-signal conditions this


relationship simply defines the capacitance as a function of the DC voltage.
For example, for input capacitor:

11) Sketch a quadratic Curtice model and justify the presence of diodes in this model.

12) A FET has RDS → ∞; VT0 = -2 V, drain current at vGS = 0 equal to IDSS = 100 mA, output conductance δID/δVDS = 100
mS for vGS = 0; vDS → 0. Evaluate the values of the parameters of the quadratic Curtice model β; α; VT0; λ.

The model of the FET is:

The λ parameter is the effect of the output resistance RDS. Since RDS → ∞, λ is equal to 0;

For vDS → 0, the hyperbolic tangent tends to its argument, so:

28
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

13. Sketch some DC and DC pulsed VI curves exhibiting low-frequency dispersion effects.

Low-frequency dispersion effect makes the DC values RDS and gm quite different from the RF small-signal ones,
producing a difference in real DC and pulsed DC curves as follows:

14. To what device does the Chalmers model apply?

It has been developed for HEMTs which exploit different approximations of components.

6. Linear amplifiers
1) Explain the difference among a maximum gain, a low-noise and a power amplifier.

The difference among a maximum gain, a low-noise and a power amplifier is the purpose of each device: the first one
wants to maximize the gain, the second one to minimize the noise and the last one to maximize the power.

2) Define a narrowband, a wideband and an ultrawideband amplifier.

A narrowband amplifier is an amplifier that has a bandwidth which is lower than 10% of the operating frequency; a
wideband one has a bandwidth higher than 10% but lower than 40% of the operating frequency; an ultrawideband
amplifier has an even larger bandwidth.

3) Explain the design flow of a narrowband maximum gain RF amplifier.

At the beginning, you have to choose the active device to use. Usually, the cutoff frequency of the device is chosen
higher than two or four times the operating frequency.
Moreover, you have to choose a suitable bias point and design the bias network.
After that, you have to make the active device unconditionally stable: this condition is not a must inside the
bandwidth of interest; it is a must at other frequencies, in particular low frequencies, if you do not want oscillations.
Then, you have to evaluate the optimum loads for maximum gain and so design the matching networks in order to
reach the complex conjugate matching.
In the end, you have to generate the circuit schematic and the layout.

4) Sketch the maximum gain, low-noise, maximum power working point of a FET.

29
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

5) Explain the operation and structure of a bias-T.

The bias-T is a component that has the role of decoupling the bias network from the RF circuit.
This device is a tripole with two inputs (RF IN and DC IN) and one output (RF+DC OUT), as shown in the following
schematic:

In practice, the DC (from 0 to f) and RF (form F to infinity).


If f and F are low the design requires very high C and L values, which imply difficulties because the larger the
component the lower the operating frequency.

6) Sketch some possible FET stabilization circuits and explain why those circuits include reactive elements together
with resistive ones.

The stabilization can be done using different configurations of passive elements made of resistances and reactive
elements:
• Series/parallel input resistors: this configuration is not good for low noise amplifiers since they introduce
noise at input.
• Series/parallel output resistors.
• Feedback resistors: usually non convenient, unless for feedback amplifiers.
Reactive elements (with a proper optimization of K) are used since they permit to obtain strong low-frequency
stabilization and lighter in-band stabilization.
Configurations are shown in the following picture:

30
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

7) Explain why the narrowband design strategy with reactive matching sections cannot be entirely extended to
wideband design.

Reactive components have a strongly different behavior as the frequency changes.


Moreover, since they do not introduce any power dissipation (they only exchange power), they don’t attenuate the
backward wave.

8) Justify why a wideband amplifier with input reactive matching cannot have at the same time flat gain and input
matching.

To have wideband flat gain you need to mismatch the transistor.


However, if the input network is reactive, mismatch at low frequency is not compatible with low input reflection
coefficient, unless some dissipative elements are introduced directly or indirectly in the input matching network.

9) Sketch a balanced amplifier and explain why this circuit is able to provide flat gain and good input matching.

The flat gain condition is introduced by the couplers.


For the first coupler, port 1 is matched if the two devices have the same input impedance: in this condition, the input
matching is achieved. The output matching is obtained if the output impedance of the devices are the same.

At the input and output ports the two devices are closed on the reference impedance R0. The device is described by
the S parameters on R0.

31
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

10) What is a directional coupler? Imagine that an ideal 3dB, 90° coupler is fed with a 100 mW signal. What is the
power on the coupled and the transmission port? What is the power on the insulated port? What is the phase
difference between the coupled and transmission ports?

A directional coupler is a 4-port device which introduces a power split and a phase shift between two ports.
PS
PT = = 50mW
2
P
PC = S = 50mW
2
PI = 0
Φ = 90°

11) Sketch the layout of a Lange coupler and of a branch-line coupler and indicate the centerband dimensions.

Lange coupler:

Branch-line coupler:

32
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

12) Is the hybrid ring a 90° or 180° coupler?

13) A Wilkinson divider on 50 Ω loads operates at 10 GHz. Assuming εeff = 4 evaluate the lengths and characteristic
impedance of the divider arms.

14) Justify the fact that a balanced amplifier has the same gain of a single stage but twice as the maximum power.

The balanced amplifier has the same gain of a single stage one for it performs a power splitting in order to have the
same gain.
It has twice the maximum power because it contains two amplifiers instead of one (as in the single stage amplifier).

15) Explain the purpose of an RF parallel / series feedback amplifier in terms of gain flatness and input and output
matching.

First of all, the introduction of an RF feedback permits to set input and output matching just modifying the feedback
elements.

33
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

With feedback you get S11 equal to S22 and so you can obtain simultaneously input and output matching.
In order to obtain S11= S22=0 (input and output match) you have to set:
R02 1
R1 = −
R2 g m

Meanwhile you get the other scattering parameters equal to:


R0 R0 − R2
S12 = S 21 =
R0 + R2 R0
Usually, to have R1=0 and so avoiding series feedback, gm is designed to be:
R2
gm =
R02
The parallel feedback resistance is equal to:
(
R2 = R0 1 + S 21 f (0) )
where S 21 f (0) is the desired gain.
By choosing the components in this way, you get a flat gain all over the bandwidth.

By adding an inductor to the parallel feedback section, you obtain a wider bandwidth.

34
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
16) We want to design 10 dB amplifier with parallel / series feedback. What is the minimum device |S21|?
( )
R2 = R0 1 + S 21 f (0) = 11R0
R0 − R2
S 21 = = 10
R0
17) Explain the operation of a distributed amplifier. What happens if the phase velocities on the input and output
transmission lines are different? What limits the amplifier bandwidth?

DAMPs allow to increase the gain-bandwidth product (GBP) of conventional amplifiers by means of distributed
structures made of two transmission lines having the same phase velocity:
gm
f T = AV (0) f B =
2πC GS

The operation of a DAMP is explained in the following picture:

Apart from DC, the power will be split into half in both loads.
If there is velocity mismatch the bandwidth is shrank. Moreover, the bandwidth is also affected by losses.

7. Noise modeling and low-noise amplifier design


1) Explain why noise causes power transfer in a circuit even if it is a zero-average random signal.

Since noise has a nonzero mean square value and a nonzero power spectrum, it causes power transfer within a circuit
anyway.
2
2) A voltage noise source has a power spectrum of 1 (nV) /Hz . Assuming a bandwidth of 500 MHz, evaluate the
mean square value of the noise voltage and the noise available power on 50 Ω.Evaluate the noise available power
spectral density of the generator.

v n,rms = S v B = 10 −18 * 5 *10 8 = 22.36µV

35
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
S V B 10 −18 5 * 10 8
Pav = = = 2.5 pW
4R 4 * 50

3) Sketch the series and parallel equivalent circuits for a noisy one-port and for a two-port.

4) Explain the physical cause of electrical noise from microscopic fluctuations.

The electrical noise is caused by velocity fluctuations and population fluctuations of carriers. The former is due to
interactions with crystal lattice, while the latter (in semiconductors) is due to generation and recombination
phenomena.
Velocity fluctuation produces a thermal noise that is a white noise; population fluctuation generates a colored noise.
In RF applications white noise dominates.

5) A resistor with R = 1 kΩ operates with a bandwidth of 5 GHz. Evaluate the power spectral density at 300 K.
Evaluate the spectral density of the resistor noise voltage and the r.m.s. noise voltage value over the specified
bandwidth.

p n ( f ) = k B T = 4.14 * 10 −21
−17 V2
S v = 4k B TR = 1.656 * 10
Hz
v n,rms = S v B = 1.656 * 10 −17 * 5 * 10 9 = 287.75µV

6) Sketch the PRC noise model for a FET.

36
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

7) In the circuit in Fig. 1, assume Z1 = 50 + j50 Ω; ZL = 50 - j50 Ω; the two noise generators are the thermal noise
(Nyquist law) generators of the two impedances, respectively (i.e. in1 is associated to Z1; en2 to ZL). Assuming a 1
GHz bandwidth, evaluate at 300 K the total power on the load.

SVL =
ZL
Z1 + Z L
2
(Z 1
2
)
S in1 + S en 2 + Z 1 S en* 2in1 + Z 1* S en 2in1 = 2500(70.7 * 4k B Tℜ[Y ] + 4k B Tℜ[ Z ])

Z = 50Ω Y = 0.02 S
V2
(
SVL = 2500 70.7 * 4 * 4.14 * 10 − 21
* 0.02 + 4 * 4.14 * 10 − 21
)
* 50 = 2.12 * 10 −15

Hz
VL = SVL B = 2.12 *10 −15 *10 9 = 1.46mV
V L2
PL = = 10.6nW
4R
8) Define the noise figure of a two-port.

Pnd , L noise _ available _ power _ spectral _ density _ on _ load


NF = '
=
P nd , L noise _ available _ power _ spectral _ density _ on _ load _ with _ noiseless _ two _ port

37
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008
The noise figure can also be shown to admit the alternative system definition involving the input and output signal
over noise ratios:
S
N G
NF =
S
N L
9) In the circuit in Fig. 2 compute the minimum noise figure and optimum generator impedance assuming Z1 = 10 Ω;
gm = 500 mS. The two (uncorrelated) noise generators e1 and i1 are white, with spectral density equal to 100
2 2
(pV) /Hz and 100 (pA) /Hz, respectively. The system bandwidth is 100 MHz.

Assuming T = 300 K:
S en S in
Rn = = 0.6Ω Gn = = 0.6S
4k B T 4k B T

 R12 Gn  Gn 2R1Gn
NFmin 
= 1 + 2 Rn + + = 2.02
 g 2
Z
2  2
g Z
2
g 2
Z
2
 m 1  m 1 m 1

2
2
g m2 Z 1 Rn
Rsopt = R +1 = 11.18Ω X sopt = − X 1 = 0
Gn
10) What is the behaviour of the noise figure with respect to the minimum when the generator impedance is varied
with respect to the optimum value?

The noise figure increases as you get far from the optimum value of the generator impedance:
2
4 g Z G − Z Gopt
NF = NFmin + n 2
RG Z Gopt
11) Discuss minimum noise vs. maximum gain amplifier design.

In LNA design the source impedance is the optimum impedance and the load is power matched. In this way you get
minimum noise but a gain lower than the maximum available gain (MAG).
The maximum gain amplifier design implies power match at input and output and so the gain is equal to the MAG.

38
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

12) In a minimum noise amplifier, what is the associated gain Gass? Can the associated gain be larger than the
MAG?

The optimum source impedance for noise does not generally yield input power match and therefore the
corresponding power gain, called associated gain Gass, is lower than the MAG, which is an upper bound for gains (this
implies that Gass cannot be larger than MAG).

13) Two amplifiers are cascaded (50 Ω design) with Gav;1 = 10 dB, Gav;2 = 20 dB, NF1 = 1 dB, NF2 = 6 dB. Evaluate
the total noise figure according to the Friis formula.

NF2 − 1 3.98 − 1
NFTOT = 1 + ( NF1 − 1) + = 1 + (1.26 − 1) + = 1.558
G av ,1 10
14) A resistive attenuator designed on 50 Ω has 3 dB loss. What is the noise figure?
S
N G 1
NF = = =2
S 1
N L
2
15) An inductive series feedback amplifier must be designed on 50 Ω at 10 GHz. Assuming CGS =0.2 pF and gm = 200
mS, evaluate LS and LG. Explain the purpose of the inductive series feedback LNA design vs. the conventional LNA
approach through input noise matching.
C GS R0
LS = = 50 pH
gm

1
LG = 2
− LS = 1.22nH
ω C GS

The conventional LNA design strategy aims at the minimum noise figure by means of optimum generator impedance
and output power matching network. This may imply a too low associated gain. The inductive series feedback uses a
lossless component at the input in order to obtain input power matching and a larger gain (with respect to the
previous case). However, the noise figure is slightly different from the minimum one.

39
Advanced Theory and Technology of Devices Politecnico di Torino
Campione Salvatore 145781 – Bosio Matteo 148451 A.Y. 2007/2008

40