81
J. Prasad SPara 2
This process
change should
have improved
Ft and Fmax !
I didn’t see any change.
Actually Fmax was lower!
Microwave
Test Engineer
Process
Engineer
VP
Technology
VP
Engineering
J. Prasad SPara 3
1.1THz Test System from Cascade Microtech/Keysight
International Microwave Symposium, San Francisco, May 22-27, 2016
J. Prasad SPara 4
What is High Frequency Characterization?
Small Signal
Characterization
Large Signal
Characterization
fT
fmax
NFmin
Rn
Zs opt
GA
Re
Rb
Rs
Rd
Pin vs Pout
PAE
P1dB
OIP3
IIP3
Load Pull
Source Pull
Linear Measurements Non-Linear Measurements
This talk
J. Prasad SPara 5
OUTLINE
• Introduction
• S-parameters
• Smith Chart
• Vector Network Analyzer
• Calibration
• De-embedding
• Examples of measured data
• Gain and Stability
• Ft and Fmax from S-parameters
• Mason’s Gain
• Transistor Specmanship
• Parameter extraction/wafer maps
• Conclusion
J. Prasad SPara 6
2
mT
gf
C C
max
'8
T
bb
ff
r C
21
2 2 2 2
---------------- ------------ ------
DCAB Ee jE e ee c jC
T DE n n m sat
e b c
WN W W W Wr C r r r C
f N D D v v
6
7
thermionic emission velocity
5 10 cm/sec for Si.2 *
saturation velocity
10 cm/sec for Si.
m
sat
kTv
m
v
max
'8
T
bb
ff
r C
2
2jE m
n
WC C g
D
jCC C
BIPOLAR FIGURES OF MERIT
Approximate Exact
J. Prasad SPara 7
h21 (BJT)
0
10
20
30
40
50
0.01 0.1 1 10 100
Freq (GHz)
h2
1 B
JT
(d
B)
h21 BJT angle
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
0.01 0.1 1 10 100
Freq (GHz)
h2
1 B
JT
an
gle
(d
eg
)
21( )
1
100
50 GHz
BJT
T
DC
T
hf
jf
f
50
Ft VARIATION WITH FREQUENCY (Bipolar)
Microwave Engineers
find Ft by plotting
h21 vs frequency!
20log(h21)
20dB/dec
J. Prasad SPara 8
2 ( ) 1 ( )
mT
D Sgs gd gd m D S p
o
gf
R RC C C g R R C
r
max8
T
G gd
ff
R C
2
mT
gs gb gd
gf
C C C
max2 ( ) 2
T
ds G S T G gd
ff
g R R f R C
MOS FIGURES OF MERIT
: parasitic cappC 2
mT
gs
gf
C
& are small in satgd gbC C
Approximate Exact
J. Prasad SPara 9
h21 MOS
0
10
20
30
40
50
60
70
80
0.01 0.1 1 10 100
Freq (GHz)
h2
1 M
OS
(d
B)
h21 MOS angle
-100
-80
-60
-40
-20
0
0.01 0.1 1 10 100
Freq (GHz)
H2
1 M
OS
an
gle
21( )
1
50 GHz
MOS
T
T
hf
jf
f
Note that this
goes to infinity!
Ft VARIATION WITH FREQUENCY (MOS)
Phase is
constant
-90
J. Prasad SPara 10
S- PARAMETERS
J. Prasad SPara 11
• At HF, difficult to measure currents and voltages
• Difficult to create open and shorts
• Everything behaves like Transmission lines with reflections
• S-parameters are very easy to understand and use
• S-parameters exist for any network
• Can easily relate to gain, loss, reflection and power
• Can predict the performance of cascaded networks
• From S-parameters, one can convert to Z, Y or H parameters
• Needed for SPICE model parameter extraction
• Some CAD programs need S-parameters for circuit design
Why S-Parameters?
J. Prasad SPara 12
Two Port Parameters
1 11 1 12 2
2 21 1 22 2
1 11 1 12 2
2 21 1 22 2
V z I z I
V z I z I
I y V y V
I y V y V
1 11 1 12 2
2 21 1 22 2
1 11 1 12 2
2 21 1 22 2
1 2 2
1 2 2
V h I h V
I h I h V
I g V g I
V g V g I
V AV BI
I CV DI
Device
I1 I2
V1 V2
+ + _ _
1 2
J. Prasad SPara 13
Device
1 2
a1
b1
a2
b2
1 11 1 12 2
2 21 1 22 2
b S a S a
b S a S a
2
2
1
1
111
1 0
221
1 0
112
2 0
222
2 0
a
a
a
a
bS
a
bS
a
bS
a
bS
a
Input reflection coefficient with output terminated in Zo
Output reflection coefficient with input terminated in Zo
Forward transmission coefficient with output terminated in Zo
Reverse transmission coefficient with input terminated in Zo
S-Parameters Defined
J. Prasad SPara 14
Device
1 2 ZO
ZO
Z1 Z2
+ +
_ _ V1 V2
VS
+
_
S-Parameters in terms of impedances and voltages
ZO : Characteristic
impedance (50W)
1 011
1 0
2 022
2 0
221
112
2
2
S
S
Z ZS
Z Z
Z ZS
Z Z
VS
V
VS
V
G.Gonzalez, Microwave Transistor Amplifiers,
Prentice Hall 1984
J. Prasad SPara 15
6dB Pad
1 2 ZO
ZO
Z1 Z2
+ +
_ _ V1 V2
VS
+
_
S-Parameters for a 6dB pad
S =
0
0
0.5
0.5
1 011
1 0
2 022
2 0
221
112
2
2
S
S
Z ZS
Z Z
Z ZS
Z Z
VS
V
VS
V
J. Prasad SPara 16
1 2 ZO
ZO
Z1 Z2
+ +
_ _ V1 V2
VS
+
_
6 dB pad
16.6 16.6
66.9
Z = 83.5 66.9
66.9 83.5
S =
0 0.5
0.5 0
Looking at Z - parameters
one can not quickly infer
this is a 6dB Pad!
50
50
Z-Parameters for a 6dB pad
J. Prasad SPara 17
1 11 1 12 2
2 21 1 22 2
b S a S a
b S a S a
12 2111
221
Lin
L
S SS
S
SIGNAL FLOW GRAPHS
a1 b2
b1 a2
S11
S12
S22
S21 a1 b2
b1 a2
S11
S12
S22
S21
L
in
a1 b2
b1 a2
S11
S12
S22
S21
out
S
12 2122
111
Sout
S
S SS
S
J. Prasad SPara 18
INTRODUCTION TO SMITH CHART
Smith Chart Story
J. Prasad SPara 19
0
0
0
0
LL
L
SS
S
Z Z
Z Z
Z Z
Z Z
Mapping of resistances – Smith Chart-1
0 50W o o
0 +1 -1
25 100 200
ZO = 50W
Pure resistances map along the
x-axis between -1 and + 1
Convert all impedances to
reflection coefficient and plot it.
That is Smith Chart!
0 ∞
𝑗∞
−𝑗∞
J. Prasad SPara 20
Radius=1 0
0
0
0
LL
L
SS
S
Z Z
Z Z
Z Z
Z Z
Mapping of reactances – Smith Chart-2
Inductive
Capacitive
Note that magnitude is always 1 but angle varies. Pure reactance maps along the circumference of a unit circle.
ZO = 50W
o
o
o
o
o
o
o
( )
0 1180
j10 1157.4
j25 1126.9
j50 1 90
j100 1 53.1
j200 1 28.1
j
1 0
L LZ W
y
-j200
x 0 50
j50
-j50
o o
j10
j25 j100
j200
-j10
-j25 -j100
J. Prasad SPara 21
R=const
circle
X=const
circle
Unit circle:
Radius=1
y 0
0
0
0
LL
L
SS
S
Z Z
Z Z
Z Z
Z Z
Mapping of impedances – Smith Chart-3
Inductive
Capacitive
All impedances (R+jX) with R>0 will map inside a unit circle.
If R is negative (R< 0), it will map outside the unit circle.
ZO = 50W
o
o
( )
50+j50 0.45 63.4
50 50 0.45 -63.4
L LZ
j
W
x
0
50
j50
-j50
o o 63.4o
50+j50
0.45
J. Prasad SPara 22
The Glorious Smith Chart
J. Prasad SPara 23
How to measure S-parameters?
- The Vector Network Analyzer
J. Prasad SPara 24
E 8361C PNA Series NETWORK ANALYZER 10MHz – 67GHz
You will also need a 4155 Semiconductor Parameter Analyzer for biasing.
J. Prasad SPara 25
VECTOR NETWORK ANALYZER BLOCK DIAGRAM
J. Prasad SPara 26
What is Calibration?
J. Prasad SPara 27
Short the leads and adjust the ZERO OHMS pot
so that the meter reads zero. We have zeroed out
resistance of the test leads.
CALIBRATION
J. Prasad SPara 28
Short the leads and write down the reading R1.
Connect the resistor Rx and take the reading R2.
Unknown resistor Rx = R2 – R1
CALIBRATION
J. Prasad SPara 29
4155/4156
Parameter
Analyzer
I-V curves Measure & gen
I-V curves Output file
Multimeter Measure
0.1W resistor Calibrate Output data Measure
1 term error
correction: short
Capacitance
Meter 4980
Capacitance
Measurement Calibrate Output data Measure
2 term error
correction: short and open
J. Prasad SPara 30
CALIBRATION
4980A CAPACITANCE METER (20Hz – 2MHz)
Short the leads and write down the series resistance Rs.
Open the leads and write down the stray capacitance Cp.
The instrument does the correction for the series resistance.
Then it subtracts the stray capacitance from the measured data.
J. Prasad SPara 31
8361C
Network Analyzer Calibrate Measure
3 term error correction:
short, open and load
1-port
S-parameter S-parameter
data
One-Port Measurement using Network Analyzer
J. Prasad SPara 32
One-Port Error Correction
J. Prasad SPara 33
8361C
Network Analyzer Calibrate Measure
3 term error correction:
short, open and load
8361C
Network Analyzer
2-port [S]
package Calibrate Measure
12 term error correction:
short , open, load and thru
1-port
S-parameter
S-parameter
data
S-parameter
data
One / Two-Port Measurement using Network Analyzer
J. Prasad SPara 34
TWO PORT CALIBRATION
J. Prasad SPara 35
J. Prasad SPara 36
J. Prasad SPara 37
w
f
Phase
(rad)
o
Group delay
1
360
gt
f
f
w
RFin RFout
S S
S S
G D
THRU CALIBRATION FOR GROUP DELAY
delayed
signal
J. Prasad SPara 38
J. Prasad SPara 39
THRU
REFLECT
LINE
TRL CALIBRATION
At low frequencies, the lines become long. So, we need different TRL structures for different frequency bands for wide band characterization. Use SOLT for lower freq.
Glenn Engen, Cletus Hoer, MTT-27 (12), Dec 1979
J. Prasad SPara 40
RF MICROWAVE PROBES (CASCADE MICROTECH)
J. Prasad SPara 41
CALIBRATION USING IMPEDANCE STANDARD SUBSTRATES
The purpose of Cal
is to bring the reference
plane to the probe tips
J. Prasad SPara 42
0
50
j50
-j50
o o
BEFORE CALIBRATION AFTER CALIBRATION
0
50
j50
-j50
o o
J. Prasad SPara 43
DE- EMBEDDING
J. Prasad SPara 44
8361C
Network Analyzer Calibrate Measure
3 term error correction:
short, open and load
8361C
Network Analyzer
2-port [S]
Package Calibrate Measure
12 term error correction:
short , open, load and thru
1-port
S-parameter
S-parameter
data
S-parameter
data
8361C
Network Analyzer
On-wafer
2-port [S] Calibrate Measure
12 term error correction:
short , open, load and thru
[S] data
of device De-embed
J. Prasad SPara 45
S S
S S
G D
Cpad
100-600 fF
S S
S S
G D
Cdevice
1- 6 fF
Pad capacitance far exceeds
Single device capacitance
A device array alleviates
this problem to some extent
100
devices
DUT
Y3
Y1 Y2
G D
S S
Device
J. Prasad SPara 46
DUT
Y3
Y1 Y2
G D
S S
Device
Y3
Y1 Y2
G D
S S
OPEN
1-Step De-Embedding
open open
device device
DUT device open
DUT DUT
S y
S y
y y y
y S
Measure on- wafer OPEN
Measure DEVICE
Use the equations on the left
DE- EMBEDS PAD CAPACITANCE ONLY !
J. Prasad SPara 47
S S
S S
G D DUT
Y3
Y1 Y2
Z2 Z1
Z3
G D
S S
Device
2- Step De - Embedding
DE- EMBEDS
• Pad Capacitance
• Series Impedance
J. Prasad SPara 48
DUT
Y3
Y1 Y2
Z2 Z1
Z3
G D
S S
Device
Y3
Y1 Y2
Z2 Z1
Z3
G D
S S
OPEN
Y3
Y1 Y2
Z2 Z1
Z3
G D
S S
THRU
Y3
Y1 Y2
Z2 Z1
Z3
G D
S S
SHORT
2- Step De - Embedding
J. Prasad SPara 49
_ _
_ _ _ _
_ _
open open
short short
device device
dev no pad device open
dev no pad dev no pad
DUT dev no pad short
DUT DUT
S y
S z
S y
y y y
y z
z z z
z S
DUT with Pad capacitance
and series elements
DUT
Y3
Y1 Y2
Z2 Z1
Z3
G D
S S
Device
Whoa!
2- Step De - Embedding
J. Prasad SPara 50
RF Characteristics of 0.18-m CMOS Transistors: Kwangseok Han, Jeong-hu Han, Minkyu
Je and Hyungcheol Shin Department of Electrical Engineering and Computer Science,
Korea Advanced Institute of Science and Technology, Taejon 305-701
High Frequency Performance of 0.18um CMOS
Measurement: 0.5-50GHz
Ft=50GHz, Fmax=45GHz
J. Prasad SPara 51
IMPERFECT on-wafer SHORT
J. Prasad SPara 52
E E
E E
B C
Layout with minimum inductance and reflections
Small octagonal pads
to reduce capacitance
and reflections
J. Prasad SPara 53
Definitions of Gain and Stability
J. Prasad SPara 54
DEFINITIONS OF GAIN
J. Prasad SPara 55
Limiting Factors of RF Performance Improvement as Down-scaling to 65-nm Node MOSFETs
H. L. Kaoa*, B. S. Lina, C. C. Liaob, M. H. Chenc, C. H. Wuc, and Albert Chinb
a Dept. of Electronic Engineering, Chang Gung Univ., Tao-Yuan, Taiwan, ROC
b Nano-Sci. Tech. Ctr, EE. Dept., Nat’l Chiao-Tung Univ., UST, Hsinchu, Taiwan, ROC
c Dept. of MicroElectronics Engineering, Chung Hua Univ., Hsinchu, Taiwan, ROC
Ft and Fmax of 65nm CMOS
J. Prasad SPara 56
Device
1 2 ZS
ZL
in
+ +
_ _ V1 V
2
VS
+
_
out
CIRCUIT FOR MEASURING MAXIMUM AVAILABLE GAIN & Fmax
We vary ZS and ZL so as to provide a simultaneous conjugate match.
This maximizes the input power and delivers maximum output power to the load.
This will give us MAG.
If we do this at each frequency, we can generate a plot of MAG vs frequency.
From this plot, we can determine the frequency at which the power gain will
become unity. This is Fmax.
J. Prasad SPara 57
11
12
21
22
0.65 95
0.04 40
5.00 115
0.80 35
S
S
S
S
12 2111
221
Lin
L
S SS
S
IN = 1 circle
53o
j50
-j50
50 0
unstable
THE EFFECT OF LOAD IMPEDANCE
J. Prasad SPara 58
11
12
21
22
0.65 95
0.04 40
5.00 115
0.80 35
S
S
S
S
OUT = 1 circle
127o
j50
-j50
50 0
unstable
12 2122
111
Sout
S
S SS
S
THE EFFECT OF SOURCE IMPEDANCE
J. Prasad SPara 59
j50
-j50
50 0
IN >1 j50
-j50
50 0
unstable
IN >1
𝐾 > 1 𝑎𝑛𝑑 ∆ < 1
For unconditional stability
conditionally stable
unconditionally stable
K<1 MAG undefined Simultaneous conjugate
match not possible
K>1 MAG defined Simultaneous
conjugate match possible
J.M.Rolett, IRE Trans CT, CT-9(1), pp 29-32, Mar 1962, W.Ku, Proc IEEE 54(11), pp 1617-1618, Nov 1966
OUTPUT STABILITY CIRCLE & UNCONDITIONAL STABILITY
= Stability factor
J. Prasad SPara 60
GAIN EQUATIONS in S- Domain
h21
Use only for K>1
Use only for K<1
Use for all K
Does not depend on K Mason’s Gain
Maximum Available Gain
Maximum Stable Gain
Current Gain
J. Prasad SPara 61
Determining Ft and Fmax from
Measured S-parameter Data
J. Prasad SPara 62
[S]DUT over
frequency
[S] g [h]
Plot 20 log (h21)
vs frequency
extrapolate
h21 g 0dB
fT
[S]DUT over
frequency
K > 1 and
D <1 ?
MSG
Plot 10 log (MSG)
vs frequency
MAG
Plot 10 log (MAG)
vs frequency
extrapolate
MAG g 0dB
fmax
Yes No
Determining Fmax when the device is stable
K>1 over some frequency range
J. Prasad SPara 63
[S]DUT over
frequency
MSG
Plot 10 log (MSG)
vs frequency
MAG
Plot 10 log (MAG)
vs frequency
extrapolate
MAG g 0dB
fmax
Yes No K > 1 and
D <1 ?
[S]DUT over
frequency
Calculate U
Plot 10 log (U) vs
frequency
extrapolate
U g 0dB
fmax
K > 1 &
D< 1 for
all f ?
No
Stable Unstable Unstable
Determining Fmax when
the device is unstable K<1
throughout all frequencies
J. Prasad SPara 64
HF performance of 3x10 HBT Ic=10mA Vce=2V
0
5
10
15
20
25
30
35
40
0.1 1.0 10.0 100.0
Frequency (GHz)
Ga
in (
dB
)
h21(dB)
MSG/MAG
U(dB)
MAG
MSG
h21
U
Ft = 54GHz
Fmax= 50GHz
Fu= 72GHz
slope:
20dB/decade
J. Prasad SPara 65
Mason’s Gain Explained
J. Prasad SPara 66
If Fmax is a Figure of Merit, is it Unique?
How can I find out whether a device is Active or Passive?
J. Prasad SPara 67
Method of unilateralizing a Two Port Network
J. Prasad SPara 68
POWER GAIN CALCULATIONS IN DIFFERENT CONFIGURATIONS
J. Prasad SPara 69
MASON’S INVARIANT or UNILATERAL GAIN U
S.J.Mason, IRE Tran CT, CT-1 (2), pp 20-25, June 1954
M.S.Gupta, T-MTT, 40(5), pp 864-879, 1992
J. Prasad SPara 70
Transistor Specmanship!
J. Prasad SPara 71
HF performance of 3x10 HBT Ic=10mA Vce=2V
0
5
10
15
20
25
30
35
40
0.1 1.0 10.0 100.0
Frequency (GHz)
Ga
in (
dB
)
h21(dB)
MSG/MAG
U(dB)
MAG
MSG
h21
U
Ft = 54GHz
Fmax= 50GHz
Fu= 72GHz
slope:
20dB/decade
J. Prasad SPara 72
70GHz Fmax Silicon Bipolar Transistor
Mamoru Ugajin, Jun-ichi Kodate, Yoshiji Kobay ashi, Shinsuke
Konaka, and Tetsushi Sakai NTT LSI Laboratories 3-1,
Morinosato Wakamiya, Atsugi-Shi, Kanagawa, 243-01 Japan
IEDM95
J. Prasad SPara 73
450GHz Ft and Fmax InP/InGaAs HBT
N. Kashio et al., TED 61(10), Oct 2014 p 3423
J. Prasad SPara 74
High Frequency Performance of 0.12um SiGe HBT
J. Prasad SPara 75
Other uses of S-parameter Measurements:
Parameter Extraction for SPICE
J. Prasad SPara 76
H11 vs Frequency
0
50
100
150
200
250
300
350
400
0.1 1.0 10.0 100.0
Frequency (GHz)
Re
(h
11
) o
hm
s
E
B’
C +
_
r C
CB
bi ' bb r
'm b eg v
Base resistance extraction from h11
h11
J. Prasad SPara 77
Wafer Map of Base Resistance and Fmax
S.J. Prasad, BCTM 1992, pp 204-207
J. Prasad SPara 78
Wafer Map of Emitter resistance and Ft
S.J. Prasad, BCTM 1992, pp 204-207
J. Prasad SPara 79
Wafer Map of Collector Capacitance
S.J. Prasad, GaAs IC Symposium 1992, pp 271- 274
J. Prasad SPara 80
ENGINEERING
It is a great profession.
There is the fascination of watching a figment of imagination
emerge through the aid of science to a plan on paper.
Then it moves to realization in stone or metal or energy.
Then it brings jobs and homes to men.
Then it elevates the standards of living and adds comforts to life.
That is the engineer’s high privilege.
J. Prasad SPara 81
ENGINEERING - II
The great ability of the engineer compared to men of other professions
Is that his works are out in the open where all can see them.
His acts, step by step are in hard substances.
He cannot bury his mistakes in the grave like doctors;
He cannot argue them into thin air like the lawyers;
He cannot cover his failures with trees and wines like the architects;
He cannot screen his shortcomings by blaming the opponents
like the politicians ;
The engineer simply cannot deny he did it.
If his works does not work he is damned.
-- Herbert Hoover
