You are on page 1of 5

EQUATIONS FOR MICROSTRIP CIRCUIT DESIGN

rA

Erik 0. Hammerstad

ABSTRACT
A method for determining the accuracy of computed impedance and waVelength
data formicrostrip is proposed. It is shownthat the standard equations of
Wheeler and Schneider have rather large errors. Revisedequations formicrostrip impedance and wavelength are given both for analysis and synthesis
with accuracy better than 1%. A simple, accurate equation for the microstrip open circuit is presented. Experimental dataon microstripT-junctions
are compared with existing theories. Anew accurate equation is given for the
reference plane displacement in the stub arm. Corrections are proposed inthe
existing equations for the other parameters of the equivalent circuit.

INTRODUCTION
During the last few years a large number of papers on computerized methods
of microstrip impedance calculation have been presented. Tobe of practical
value to a microstrip circuit designer, theymust show better accuracy than
that of the standard equations of Wheeler [1] and Schneider [2], and it is
therefore somewhat surprising that this aspect is usually treated very superficially if at all. The computerized methods alsohave the disadvantage
of being cumbersome to incorporate into a CAD package allthough manual design may be done with tables. For optimization, equations are absolutely
neccessary as the computerized methods are much too time-consuming.
The accuracy needed in equations for microstrip impedance and effective
dielectric constant should be set in accordance with typical variations in
substrate parameters anid manufacturing tolerances, and 1% as a maximum relative error seems to be a reasonable goal for both. Currently available
equations do not have that accuracy.

DETERMINATION OF ACCURACY
For the impedance of the single-dielectric microstrip an accurate but implicit and quite complex solution exists, as for example given in [2]. A

carefully constructed computer program for this solution may have relative
errors in the order of 10-5[3]. As the single-dielectric impedance, Z , the
microstrip impedance, Z, and effective dielectric constant, Ee' are related
by
z
2
ee

(1)

(Z o/Z)
0()

this program may be used in estimating errors of other computations. In a


very common method, that using the Green's function, the main errors are
due to the quantization of the charge distribution on the microstrip and the
programming of the method may easily be done so that the errors due to the dielectric substrate are negligible [ 4]. The above program is then suff icient
for determination of it's computational accuracy. Thiswas shown in a program
written by Maesel [4] which was checkedwiththis method. Relativeerrors in
Z were found tobe below 0. 1%. This accuracy is due to the division of the microstrip into substrips of unequal width and with a linearly varying charge
distribution over the substrips while often used coarser approximation of the
charge distribution may give errors of 1%.
The author is on leave of absence fromELAB, N7034Trondheim- NTH, Norway,withthe
Div. of Network Theory, Chalmers Univ. of Techn.,Fack,S40220 Gothenburg 5, Sweden

268

ACCURACY OF CURRENT EQUATIONS


Wheeler has given sets of equations for determining microstrip impedance.,
both for analysis (Z = f(w/h))and for synthesis (w/h = g(Z)), while
Schneider has given equations for single-dielectric impedance and effective
dielectric constant. Relative errors in impedance of these equations are
shown in Table 1 and of effective dielectric constant in Table 2. (With
Wheeler's synthesis equation, the single-dielectric analysis equations
with (1) is used to determine Ee and a constant crossover between the equations for wide and narrow strips of w/h = 2 is assumed for both analysis
and synthesi.s.)
NEW EQUATIONS

New equations with accuracy fulfilling the 1% goal have been obtained by
modifying those of Wheeler and Schneider. Schneider's equations for 5e are

E= yf +1i+ (c
Ee =2 8 r +
(r

1) FJ

(2)
(2)

1/2

F = (1 + 10 h/w)

(3)

Making F independent of er is an approximation, but for Er < 16 it is good


enough and the following equation gives relative errors in Ce not larger
than 1% for 0.05 < w/h < 20 and c < 16

(1

+ 12

(1

12

h/w) 1/2

+ o.o4(1 - w/h)

h/w) /w/h

w/h < 1

(1

>

(It is estimated that the errors outside this range are not larger than 2%
when (3) is used for w/h < 0.05, however, strips that narrow will hardly
be used in practice).
Impedance equations for analysis are best constructed with those for singledielectric together with the se equation. The error inSchneider's equation
for w/h > 1 is approximately three times that claimed and efforts to improve
on this showed that the best result was obtained by modifying Wheeler's

ZV'F-'=

z ln(8 h/w + 0.25 w/h)

no

{w/h

1.393

0.6671n(w/h +1.444)}

w/h

<

, w/h

>

The maximum relative errors in ZO is then 0.4% and approximately 0.8% in Z.


The same accuracy was obtained by modifying the constants of Wheeler's syn-

thesis equations:

| 8/{exp(A)
w/h
<

exp(-A)}

~ ~ ~~c-1
r

I- ITIB-I-ln(2B-1 )+

where
A =
and

pi

\/2(E r

1)Z+

2s_

, w/h < 2

{ln(B-1)+039-0.61/si
rr
J

0+1 .23 + 0.11/, B


(
O11s

Er

wth

>

(6)

2VP",Z

no = 376.73 ohm the free-space wave impedance.

These equations, together with those for dispersion (which may increase the
errors to 2%) and loss (including the effect of surface roughness)[31],
form a complete and accurate set for the microstrip line parameters.
Their relative errors are shown in Table 3.

269

MICROSTRIP DISCONTINUITIES
Present methods of calculating the equivalent circuits of microstrip discontinuities are only adequate if an electrostatic approximation is sufficient, such as for the open circuit and gap, but not for those which must
be described by a dynamic theory such as the T-junction [5]. The stripline
theories which have been proposed, are not any better, but may be useful
as an indication of the functional relationships to use in a curve-fitting
to experimental data on the microstrip discontinuities.

MICROSTRIP OPEN CIRCUIT


Silvester and Benedek [6] have, in addition to the usual tiny graphs, presented their calculated data for excess capacitance of the microstrip open
circuit as polynomial equations for six different values of r . It has been
found possible to reduce these equations to one valid for all values of Er
by transforming the capacitance to a line extension, Ae (the maximum error
isless than 5% of h for er = 1 and 1% of h for other values of sr
A
ce + 0.3
*w/h +0.262r
C - 0.258
w/h + 0.813
h
0
e
MICROSTRIP T-JUNCTION
The microstrip T-junction with nomenclature and the preferred equivalent
circuit is shown in Fig. 1. Stripline equations for a lumpedelement circuit
may be transformed to the preferred one as shown in [7], where they and
curves for the waveguide T are compared with experimental results. Both
theories show large errors compared with the measurements, except for the
transformer turns ratio, and are therefore not very suitable for direct
application to microstrip. The frequency dependence of the T-equations is
given by the factor 2D1/X which cannot be much larger than 0.3 due to radiation [3], where
(8)
D = p h/Z
0

As the stub arm reference displacement is the most important one in terms
of it's effect on circuit performance, it has been extensively investigated
experimentally. T-junctions on 1/1i6" thick Rexolite 2200 substrate (e =2.62)
were measured for 2D1 / = 0. 1 and 0.3 with values of Z1 from 30 to 130 ohm.
The results are shown in Fig. 2 together with results for 1/8" Rexolite [3]
and alumina [5].A curve fit to these data, based on the stripline equation
gave the following result with an estimated accuracy of better than 10%of h

(9)

d2' I/D1

{0.076+0.2(2D1 /X)2+0.663exp(-1.71Z1/Z2)-0.172ln(Z1/Z2)}Z1/Z2

An equation for the main line reference displacement (which is quite small)

d/D2

0.05

2Z1/Z2

(lo)

where the constant (which is rather arbitrary in the theory) has been adjusted to take into account stripline data and one microstrip result for 1/8"
Rexolite. Measurements by Easter indicate that the constant might be halved.
2

2DZ

sin( 2
Tr

12

2D1

d2'\2

is a good approximation to the stripline theory and experimental data for


the transformer turns ratio and also to results for one T measured by
Easter. An equation for B is given in [3], but it's accuracy is not better
270

than order of magnitude, and B may rather be neglected for Z /Z < 1.2.
1 2
ACKNOWLEDGMENT

Some of this work is reported in [3] which was supported by the Royal Norwegian Council for Scientific and Industrial Research. The author isgrateful for the support of E. Myrseth and the great help.given by F. Bekkadal
in producing that report. He would also like to thank Prof. E. F. Bolinder
for making the continuation of this work possible.
REFERENCES
[1] H. A. Wheeler: "Transmission Line Properties of Parallel Strips Separated by a Dielectric Sheet", IEEE Trans. MTT-13, pp 172-185.
[2] M. V. Schneider: "Microstrip Lines for Microwave Integrated Circuits"
BSTJ 48, pp 1421-1444.
[3] E. 0. Hammerstad: "A Microstrip Handbook", ELAB Report STF44 A74169,
Trondheim, Feb. 1975.
[4] M. Maesel: "A Theoretical and Experimental Investigation of Coupled

Microstrip Lines", ELAB Report TE-1.68, Trondheim, Apr. 1971.

B. Easter: "The Equivalent Circuit of Some Microstrip Discontinuities3',


School of Electronic Engineering Science, University College of North
Wales, 1974.
[6] P. Silvester and P. Benedek: "Equivalent Capacitance of Microstrip
Open Circuits", IEEE Trans. MTT-20, pp 511-516.
[7] A. G. Franco and A. A. Oliner: "Symmetric Strip Transmission Line Tee
Junction", IEEE Trans. MTT-10, pp 118-124.

[5]

12 1

Fu

jc

ei

Figure 1. Microstrip T-junction and preferred equivalent circuit.

II

d2

hi=i52mm
h=3.1 1 mr

fr2.62

0. fr=2.62

0,2
U

+F;
U

/f_tI

0.5

+N-

1.5

Figure 2. Microstrip T-jiunction

-0.2

[31

h=0.66mm t 5]

2Di/A =0.1

Er9.8

stub arm ref erence plane di splacement, measured results a nd proposed equation.

271

2.5

Zi /Z2

Schneiaer

Eq.
1.0

0. 1

0. 1
0.0
0.0

0.2
0.5

0.1
0.0
o.4

0.5

1
10

0.7

Wheeler

(an.)

Wheeler

3.78

9.6

0.7
0.8
0.8

0.3

o.6
o.6
o.6
o.4

0.1

0.0

-0.3

0.1 -o. 6

0.1

0.1

2.62

2.62

3.78

9.6

1.0

2.62

3.78

9.6

1.0

0.7

0.8
o.4

0.8

0.1
0.1

0.1

-0.6

0.0
0.0
0.0

0.2
0.2
0.3

o.6 o.4
0.1 0.5
0.1 0.5

-1 . 1

0. 1

0.2

1.2

1.4

0.5
0.0
-0.2
-0.3
0.0

0.3
-O. 1

-0.2

-0.5
-0.5

-0.2
0. 1

0.4

0.2

-1 .0

-0.3
0.2

0.2

0.1

-1 .0
0.5 0.0
0.3 0.2

0.2
0.2
0.3

0.3
1.3
-1.5
-0.3
0.0

1.4

-1.2
0. 1
0.2

(syn. )

0.1
1.0
-0. 1

0.5

-0.7 -0.9 -1.4


0.1 -0.2 -0.2 -0.5

Table 1. Relative errors in the impedance equations of [1] and [2].

Schneider

Eq.

W/h

0.1

0.2

0.5
1
2

10

Wheeler (an.)

2.62 3.78 9.6 _j2.62 3.78 9.6

2.62 3.78 9.6


-1 .5

-0.9
0.1

o.6

0.8
0.5
0.5

Wheeler (syn.)

-1.6 -1 .6 -0.2 -0.3 -0.3 -0.2 -0.3 -0.3


-0.2 -0.3 -0.4
-o. 8 -o.4 -0.2 -0.3
-0.5
o.4 1.3 -o. 4 -0.5 -o.4 -0.3
1.2 2.3 -0.2 -0.3 -0.4 0.0 -0.1 -0.2
1.1
3.5 3.7 3.9
1.2 2.3 2.2 2.6 3.2
0.0 1.2 1.7

0.8 0.8

0.9

o.8

1.6
1.3

1.2
0.5

1.8

1.4
o.6

1.3

0. 8

1.5

0.7

0.8

1.7
0.9

Table 2. Relative errors in effective dielectric constant as


determined from the impedance equations [1] (see text)
and [2].

Z-analysis

Eq.

w/h r

w/h-synthesis

9.u~~~~~~~~

1.0

2.62
6 2.62 3.7u 9.6
9.6{
1.0 2.62
|
3.78 9.j
3.78
9.6

0.1

0.1

0.2

0.0

0.2
0.2
0.2

0.2
0.3
0.0

-0.2

-o. 6

o.4 o.6
0.0
o.4

o.4

0.1

0.5

0.0

0.1

-0.3
2

10

0.0

0.3

-0.

0.1

0.1

0.3
0.2

-0. 1
-0.3
-0.3
-0.2

1
0.2
0.

o.6 0.5

0.5
0.5
0.3

0.1
0.1

0.1

1.0
1

-0.

0.

Table 3. Relative errors of the

new

o.6

0.5
0.3
-0.

1
1

o.4

0.5
0.5
0.3
-o. 6

0.1

0.1

1
0.0

0.2

0.3

0.3

-0.

0.1
0.1

1
1

0.5

-0.

equations.

272

-0.

-0.5
-0.3

-0.3

-0.3

0.0

o.6
o.8

-0. 1

0.2

-o.4

-0.

0.7

-o .4

-0.2

0.5

-0.7

-0.7

-0.3

-o. 6

-o. 6

You might also like