Transcript Outline - 正修科技大學

 Properties of Some Typical Substrate Materials
MIC Manufacturing Technology
 Thin-Film Module
 Circuit is accomplished by a plate-through technique or an
etch-back technique.
 Thick-Film Module
1) Thick-film patterns are printed and fired on the ceramic substrate.
2) Printed circuit technique is used to etch the desired pattern in a plastic
substrate.
 Medium-Film Module

Above technologies are suitable for HMIC productions.
 Monolithic Technology

This technology is suitable for MMIC productions.
 Properties of Various Manufacturing Technology
Multi-Chip Modules (MCM)

MCM provides small, high precision interconnects among multiple ICs to
form a cost-effectively single module or package.

Four dominant types of MCM technologies:
1) MCM-L having a laminated PCB-like structure.
2) MCM-C based on co-fired ceramic structures similar to thick-film
modules.
3) MCM-D using deposited metals and dielectrics in a process very similar
to that used in semiconductor processing.
4) MCM-C/D having deposited layers on the MCM-C base

Advantages of an MCM over a PCB are :
1) Higher interconnect density.
2) Finer geometries enables direct chip connect.
3) Finer interconnect geometries enables chips placed closer together
and it results in shorter interconnect lengths.
 Comparison of MCM Technologies
Low Frequency Characters
of Microstrip Line
Microstrip Line
Microstrip line is the most popular type of planar transmission lines,
primarily because it can be fabricated by photolithographic processes and is
easily integrated with other passive and active RF devices.
When line length is an appreciable fraction of a wavelength (say 1/20th
or more), the electric requirements is often to realize a structure that
provides maximum signal, or power, transfer.
Example of a transistor amplifier input network
Microstrip components
 Transmission line
 Discontinuities
•Step
•Mitered bend
•Bondwire
•Via ground
The most important dimensional parameters are the microstrip width w,
height h (equal to the thickness of substrate), and the relative permittivity of
substrate r.
Useful feature of microstrip :
DC as well as AC signals may be transmitted.
Active devices and diodes may readily be incorporated.
In-circuit characterization of devices is straightforward to implement.
Line wavelength is reduced considerably (typically 1/3) from its free
space value, because of the substrate fields. Hence, distributed
component dimensions are relatively small.
The structure is quite rugged and can withstand moderately high
voltages and power levels.
Although microstrip has not a uniform dielectric filling, energe transmission
is quite closely resembles TEM; it’s usually referred to as ‘quasi-TEM’.
Electromagnetic Analysis Using Quasi-Static
Approach (Quasi-TEM Mode)

The statically derived results are quite accurate where frequency is below a
few GHz.

The static results can still be used in conjunction with frequency-dependent
functions in closed formula when frequency at higher frequency.
 Characteristic Impedance Z0
For air-filled microstrip lines,
L
Z0 
C
For low-loss microstrip lines,
L
1
1
Z 01 

 cL  L  2
C1 cC1
c C1
We can derive
1
Z
c
Z0 
 01 ;  eff 
C1
c CC1
 eff
Procedure for calculating the distributed capacitance:
Lapace's equat ion: t 2Vt0 ( x, y )  0
0

V
( x, y )
t
0
0
0
BCs for Vt , Et , Dt at y  0 and y  h
0
Gauss' s Law
Et  tVt0
Q
w
 
 0
 0  Q  D 0  ds  C  c  f ( ,  r )
c
 t
V0
h
Dt   Et
c
 C1  f (
w
,  r  1)
h
 Effective Dielectric Constant
 eff

c 2 C
w
( ) 
 g( ,  r )
vp
C1
h
For very wide lines, w / h >> 1
 eff   r
For very narrow lines, w / h << 1
 eff 
r  1
2
r
w
q 1
h
w
 r q  1/ 2
h
We can express eff as
 eff  1
 eff  1  q ( r  1)  q 
r 1
1   eff   r
where filling factor q represents the ratio of the EM fields inside the
substrate region, and its value is between ½ and 1. Another
approximate formula for q is
q
 r ( eff  1)
 eff ( r  1)
(provided by K.C. Gupta, et. al.)
r  1
 Planar Waveguide Model
(Parallel-Plate Model)
Z0 
h0
weff  eff
r
( )
where 0   0  0  120
 eff
 Conductor
Loss ac
In most microstrip designs with high r, conductor losses in the strip and
ground plane dominate over dielectric and radiation losses.
It’s a factors related to the metallic material composing the ground
plane and walls, among which are conductivity, skin effect, and surface
roughness.
Relationships:
h
ac 
, roughnessof thesubstratesurface.
w
ac oppositeto conductivity ( In idealized line,  )
T hestrip thicknessshould be greater than 3 ~ 5 timestheskin depth to minimizea c .
 Dielectric Loss ad
To minimize dielectric losses, high-quality low-loss dielectric substrate
like alumina, quartz, and sapphire are typically used in HMICs.
In MMICs, Si or GaAs substrates result in much larger dielectric losses
(approximately 0.04 dB/mm).
 Radiation
Loss ar
Radiation loss is major problem for open microstrip lines with low .
Lower  (5) is used when cost reduction is a priority, but it lead to
radiation loss increased.
The use of top cover and side walls can reduce radiation losses. Higher
 substrate can also reduce the radiation losses, and has a benefit in that
the package size decreases by approximately the square root of . This
benefit is an advantage at low frequency, but may be a problem at higher
frequencies due to tolerances.
 Formulations of Attenuation Constant
R
2R 1
R
ac 
 s
 s (Np/m)
2Z0
w 2Z0 wZ0
a
GZ0 1
 L 1  t an c
1
ad 
 (C  )
 (
) LC   LC t an c
2
2
 C 2

2
1
1
1
  t an c   0 0 eff t an c  k0  eff t an c (Np/m)
2
2
2
where k0   0 0   / c (free- space wavenumber)
However, the dielectric loss should occur in the substrate region only,
not the whole region. Therefore, ad should be modified as
 r ( eff  1) 1
1
a d  q  k0  eff tan  c 
 k 0  eff tan  c
2
 eff ( r  1) 2
k0 r ( eff  1) tan  c

(m -1 )
2  eff ( r  1)
How to evaluate attenuation constant
a
 Method 1 : in Chapter 2.14 ; a is calculated from RLCG values of material.
 Method 2: Perturbation method
a
Pl ( z  0)
where Pl is power loss per unit length of line,
2 P0
P0is the power on line at z=0 plane.
 Method 3:a is calculated from material parameters.
a  ac  ad  ar where ac is attenuation due to conductor loss
ad is attenuation due to dielectric loss
Rs dZ0
ac 
ar is attenuation due to radiation loss
2 Z  dl
0
k 2 tan
ad 
; unit : Np/m (for T E or T M waves)
2
k tan
ad 
; unit : Np/m (for T EM waves)
2
Combined Loss Effect : linearly combined quality factors (Q)
1
1
1
1



Q Qc Qd Qr
 Recommendations
1) Use a specific dimension ratio to achieve the desired characteristic
impedance. Following that, the strip width should be minimized to
decrease the overall dimension, as well as to suppress higher-order
modes. However, a smaller strip width leads to higher losses.
2) Power-handling capability in microstrip line is relatively low. To
increase peak power, the thickness of the substrate should be
maximized, and the edges of strip should be rounded ( EM fields
concentrate at the sharp edges of the strip).
3) The positive effects of decreasing substrate thickness are :
a) Compact circuit
b) Ease of integration
c) Less tendency to launch higher-order modes or radiation
d) The via holes drilled through dielectric substrate contributing
smaller parasitic inductances
However, thin substrate while maintaining a constant Z0 must narrow
the conductor width w, and it consequently lead to higher conductor
losses, lower Q-factor and the problem of fabrication tolerances.
4) Using higher  substrate can decrease microstrip circuit dimensions,
but increase losses due to higher loss tangent. Besides, narrowing
conductor line have higher ohmic losses. Therefore, it is a conflict
between the requirements of small dimensions and low loss. For
many applications, lower dielectric constant is preferred since losses
are reduced, conductor geometries are larger ( more producible), and
the cutoff frequency of the circuit increases.
5) For microwave device applications, microstrip generally offers the
smallest sizes and the easiest fabrication, but not offer the highest
electrical performance.
Design a
microstrip line by
the method of
“Approximate
Graphically-Based
Synthesis”
Example1: Design a 50 microstrip line on a FR4 substrate( r =4.5).
Solution
Assume eff = r =4.5
Z 01  Z 0  eff  50  4.5  106
From Zo1 curve  w/h=1.5
From q-curve

q=0.66
eff = 1+q (r +1)=1+0.66(4.5-1)=3.31
 2nd iteration
Z 01  Z 0  eff  50  3.31  91
eff = 3.31
From Zo1 curve  w/h=1.7
From q-curve

q=0.68
eff = 1+q (r +1)=1+0.68(4.5-1)=3.38
 3rd iteration
Stable result
w/h=1.88; eff =3.39
 Formulas for Quasi-TEM Design Calculations
Analysis procedure: Give w / h to find eff and Z0.
(provided by I.J. Bahl, et. al.)
r  1
r 1
 eff 

2
2 1  12 (h / w)
8h w
 60
ln(
 ),
 
w 4h
 eff
Z0  
120

,

  eff ( w h)  1.393  0.667 ln (1.444  w / h)
Synthesis procedure: Give Z0 to find w / h.
A

8e
w
1
h
w
1
h
w
2
 2A
w e  2
h

h 2 
r 1 
0.61  w
ln ( B  1)  0.39 
2
 B  1  ln (2 B  1) 
,



2 r 
r  h
 
Z0  r  1  r  1
0.11
377
where A 

(0.23 
), B 
60
2
r  1
r
2Z0  r
Example2: Calculate the width and length of a microstrip line for a 50
 Characteristic impedance and a 90° phase shift at 2.5 GHz. The
substrate thickness is h=0.127 cm, with eff =2.20.
Solution
Guess w/h>2
377
B
 7.985
2Z0  r
Matched


w 2
 1 
0.61
  B  1  ln (2 B  1)  r
ln
(
B

1
)

0
.
39

  3.081 with


h 
2 r 
 r 
guess
Then w=3.081h=0.391 (cm)
 eff 
r  1
2

r 1
2 1  12 (h / w)
 1.87
The line length, l, for a 90° phase shift is found as
  90  l   k l

k  2f
eff
0
90 ( / 180 )
l
 2.19 (cm)
 eff k0
0
c
 52.35 (m-1 )
Microstrip on an Dielectrically Anisotropic Substrate
 x
 0

 0
0
y 0 

0  z 
0
0
  0 0  9.4
 Sapphire   0  || 0    0 11.6 0 

 

9.4
0
 0 0     0
 iC f   yC y
; where
 req 
C f  Cy
0
C f  Ci   i 0 ( w / h ) denotesfringingcapacitnce
Cy   y 0 ( w / h ) denotesparallel- platecapacitnce
Empirical formula
1.21
;
2
1  0.39[log(10w / h)]
 0.5% accuracy throughouttherange
0.1  w / h  10
and
10  Z 0  100
 eff  12 
Curve  : i =10.6 ;
Curve : used req formula