www.activefrance.

com
Prepared By :

TECHNICAL NOTES
No.

1 (10)

Neill Tucker
Project Title :

09:001
Date Rev

Stepped Impedance Matching

06/02/2010

File C:\ My documents \My Docs\ Stepped_Impedance_A.doc

TECHNICAL NOTES

Stepped and Tapered Impedance Matching Transformers

ABSTRACT This technical note outlines the theory used in Rfutils_M toolbox for MATLAB. This toolbox is basically an adjunct to the existing Rfutils toolbox and facilitates the design and evaluation of multi-step impedance matching transformers. The designs include n-section quarter-wave matches and splitters based on an approximation to the Chebyshev distribution, as well as tapered matches using the Klopfenstein taper. In addition to the impedance profiles, routines are included to calculate Microstrip realisations of the transformers and output these in Autocad DXF format.

EN/F AD 109 251 R1A C:\DOCUME~1\IBMUSE~1\MYDOCU~1\MYDOCS~1\Stepped_Impedance_A.doc

www.activefrance.com
Prepared By :

TECHNICAL NOTES
No.

2 (10)

Neill Tucker
Project Title :

09:001
Date Rev

Stepped Impedance Matching

06/02/2010

File C:\ My documents \My Docs\ Stepped_Impedance_A.doc

CONTENTS

ABSTRACT ................................................................................................................................................................................ 1 CONTENTS ................................................................................................................................................................................ 2 1. 2. INTRODUCTION.............................................................................................................................................................. 3 MULTI-SECTION MATCHES........................................................................................................................................ 4 2.1 Binomial .................................................................................................................................................................... 4 2.2 Chebyshev ................................................................................................................................................................... 5 2.3 Response Evaluation .................................................................................................................................................. 7 3. TAPERED MATCHES ..................................................................................................................................................... 8 3.1 Exponential ................................................................................................................................................................ 8 3.2 Triangular .................................................................................................................................................................. 8 3.3 Klopfenstein ................................................................................................................................................................ 9 3.4 Comparison .............................................................................................................................................................. 10 REFERENCES ......................................................................................................................................................................... 10

EN/F AD 109 251 R1A C:\DOCUME~1\IBMUSE~1\MYDOCU~1\MYDOCS~1\Stepped_Impedance_A.doc

www.activefrance.com
Prepared By :

TECHNICAL NOTES
No.

3 (10)

Neill Tucker
Project Title :

09:001
Date Rev

Stepped Impedance Matching 1. INTRODUCTION

06/02/2010

File C:\ My documents \My Docs\ Stepped_Impedance_A.doc

One of the commonest problems facing the Rf/Microwave engineer is matching, whether for maximum power transfer, minimum noise figure or optimum bandwidth, matching usually crops up somewhere in the design. Lumped element and stub type matching is the most common and very well documented, less well documented is stepped and tapered line matching. Lumped element and stub type matching can provide very effective and compact matching solutions. The price paid for this compactness is that the impedance must be moved 1 rapidly to the required value, requiring relatively high currents and voltages to exist in the matching components. High current densities result in ohmic losses in the conductors, while high field intensities result in dielectric losses. Bandwidth can be improved by using more complex networks with increased the component count, but accounting for strays can make a practical design increasingly difficult. Stepped and tapered line matching solutions are unlikely to be the most compact but are well suited to applications requiring wide bandwidth, low loss or high power operation. Typical applications include receiver front ends, transmitter outputs and general test and measurement. A variety of stepped and tapered line impedance transformer designs have been explored, the following sections outline the associated theory used to design and evaluate them.

Note 1 Relative to those on the matched line carrying the same power.

EN/F AD 109 251 R1A C:\DOCUME~1\IBMUSE~1\MYDOCU~1\MYDOCS~1\Stepped_Impedance_A.doc

www.activefrance.com
Prepared By :

TECHNICAL NOTES
No.

4 (10)

Neill Tucker
Project Title :

09:001
Date Rev

Stepped Impedance Matching 2. MULTI-SECTION MATCHES

06/02/2010

File C:\ My documents \My Docs\ Stepped_Impedance_A.doc

The simplest transmission line match is usually quoted as being of a single quarter-wave section of line, characteristic impedance Zmatch as given by equation 2-1

Z match = Z 0 Z load
Where : Z0 = Characteristic to be matched to Zload = Load impedance to be matched

Eq 2-1

If we take the example of matching a load impedance of 100ohms to a characteristic impedance of 50ohms, equation 2-1 gives us a matching impedance of 70.7ohms and a fractional bandwidth of 29%, see [1] for derivation and calculations. To increase the bandwidth of the transformer we need to add more quarter-wave sections, the trick is deciding what impedance to make them. 2.1 Binomial The Binomial or maximally flat design, as its name suggests, gives the flattest response in the band of operation for a given quality of match. The calculation of the appropriate transformer impedances is not that arduous and well documented in [1]. The essential calculations are summarised in equations 2.1-1 and 2.1-2 below. The binomial coefficients are given by :

CnN =

N! ( N n)!n!
N = Number of matching sections n = Section number

Eq 2.1-1

Where :

The coefficients are obtained by evaluating Eq 2-2 for values of n=0 to (N-1) The natural log of the impedances is then given by :

ZL ln(Z ( n +1) ) = ln(Z ( n ) ) + 2 N C nN ln Z 0

Eq 2.1-2

The values for ln(Z(n+1)) and hence Z(n+1), are obtained by evaluating Eq 2.1-2 for values of n= 0 to (N-1)

EN/F AD 109 251 R1A C:\DOCUME~1\IBMUSE~1\MYDOCU~1\MYDOCS~1\Stepped_Impedance_A.doc

www.activefrance.com
Prepared By :

TECHNICAL NOTES
No.

5 (10)

Neill Tucker
Project Title :

09:001
Date Rev

Stepped Impedance Matching 2.2 Chebyshev

06/02/2010

File C:\ My documents \My Docs\ Stepped_Impedance_A.doc

The Chebyshev design gives the best bandwidth for a given number of transformer sections, at the expense of ripple in the band of operation. The calculation of the appropriate transformer sections is not so easy in this case. The only methods I have found seem to involve a fair amount of algebra followed by comparing terms in the resulting expansions, a method that is not easy to automate. I must confess my resolve to dig through the maths was not that strong. Ultimately any design is probably going to need some tuning, either with a scalpel and copper tape or more likely in an em-simulator. So, do you spend all your time searching for mathematical perfection only to let an optimiser loose on it. Or, settle for a good first attempt, so at least your optimiser will not end up stuck on a false summit. Out of interest I decided to plot some of the tabulated Chebyshev design values for the 100ohm to 50ohm match, operating reflection coeff 0.05, for N=2,3..7 sections. Instead of plotting the section impedance values, the equivalent reflection coefficients for each section are plotted (y-axis). The x-axis is simply the impedance section number, figure 2.21 shows the resulting curves (solid, coloured).
N-section Transformer Reflection Coefficient (Zo=50 ZL=100) 0.35

0.3

0.25 Reflection Coefficient

0.2

0.15

0.1

0.05

4 5 Linear X-axis

Figure 2.2-1 Chebyshev curves plotted as reflection coefficient (reference chapprox.m in rfutils_m toolbox)

EN/F AD 109 251 R1A C:\DOCUME~1\IBMUSE~1\MYDOCU~1\MYDOCS~1\Stepped_Impedance_A.doc

www.activefrance.com
Prepared By :

TECHNICAL NOTES
No.

6 (10)

Neill Tucker
Project Title :

09:001
Date Rev

Stepped Impedance Matching

06/02/2010

File C:\ My documents \My Docs\ Stepped_Impedance_A.doc

Somewhat unsurprisingly, all the curves converge (if extrapolated) to a point at approximately (0.333,0.333), the reflection coefficient of the 100ohm load impedance measured in a 50ohm system. The other end of the curves finish at very approximately (N,0.333/(N+2)), where N is the number of sections. Effectively we have a series of curves whose start and finish points are determined purely by the reflection coefficient of the load impedance (referenced to Zo) and the number of sections in the transformer. With a little a fit by eye curves for N=2,3 and 4 can be approximated by a straight line. Curves for N=5,6 and 7 sections are approximated using an x-shifted cosine curve. For 8 or more sections a shifted and scaled (in x-axis) cosine curve is used. See figure 2.2-1 (dashed curves).

y = mx +c

for 2 N 4 for 5 N 7

Eq 2.2-1 Eq 2.2-2

T T y = cos (2 x + 5 N ) ld 0.95 + ld 12 2 2

Tld Tld y = cos ( x 1) ( N 1) 2 0.95 + 2

for 8 N

Eq 2.2-3

Z load Z 0 Z load + Z 0

Depending on the number of sections required (N), the appropriate curve is selected and the reflection coefficients for each section Tn is calculated. The impedance for each section is then calculated using equation 2.2-4, below.

Zn =

1 + Tn Z0 1 Tn

Eq 2.2-4

Whilst lacking the precision of the formal solution, the resulting transformer values will give operating band reflection coefficients of around 0.1 (-20dB) or better for any impedance ratios up to 4:1, for any number of sections. Bearing in mind the practical limitations of realising very high or low impedances, the degradation beyond the 4:1 ratio doesnt seem too much of a problem.

EN/F AD 109 251 R1A C:\DOCUME~1\IBMUSE~1\MYDOCU~1\MYDOCS~1\Stepped_Impedance_A.doc

www.activefrance.com
Prepared By :

TECHNICAL NOTES
No.

7 (10)

Neill Tucker
Project Title :

09:001
Date Rev

Stepped Impedance Matching 2.3 Response Evaluation

06/02/2010

File C:\ My documents \My Docs\ Stepped_Impedance_A.doc

Once the impedance for each section has been established, the frequency response is easily calculated by successively applying equation 2.3-1 at each frequency of interest, working from the load impedance back towards the input.

Z in = Z tr

Z L + Z tr tanh( l ) Z tr Z L tanh( l )

Eq 2.3-1

Where : = + j = LossdB/8.686 and = 2/

LossdB = Loss in dB per unit length along line l = Transformer line length (m) = Wavelength at the frequency of interest (m) ZL = Load impedance (ohms) Ztr = Transformer impedance (ohms)

Once the last transformer has been taken into account, Zin represents the matched input impedance as a function of frequency.

EN/F AD 109 251 R1A C:\DOCUME~1\IBMUSE~1\MYDOCU~1\MYDOCS~1\Stepped_Impedance_A.doc

www.activefrance.com
Prepared By :

TECHNICAL NOTES
No.

8 (10)

Neill Tucker
Project Title :

09:001
Date Rev

Stepped Impedance Matching 3. TAPERED MATCHES

06/02/2010

File C:\ My documents \My Docs\ Stepped_Impedance_A.doc

Although the stepped impedance transformer designs described in section 2 are normally sequences of quarter-wave sections (at frequency = f0), an increased number of shorter sections can also be used to approximate a taper. If we then look at the overall length of the taper instead of the individual sections, the frequency at which this becomes a halfwavelength is roughly the lowest frequency of operation. Since the response of these tapered lines can be calculated in the same way as outlined in section 2.3. Design can simply be a matter of setting the transformer lengths to be 0/100 and the number of sections to N to around 50, thus giving an overall length of roughly 0/2. Note that f0 is not the definitive lower cut-off frequency, it just defines a convenient wavelength to divide up into small sections and define the taper. Although the optimum length for tapered matches is around 0/2 (N=50), it does vary depending on taper type. The length of the taper can be altered in 0/100 increments by simply changing the number of sections used. 3.1 Exponential The Exponential taper gives the lowest cut-off frequency for a given length but at the expense of ripple in the operating band. The impedance as a function of distance (z) along the taper (length L) is given by equation 3.1-1 below.

Z ( z ) = Z 0 e az
Where : a =

for 0 z L

Eq 3.1-1

1 ZL ln L Z0

3.2

Triangular The Triangular taper has the highest cut-off frequency for a given length but has fewer ripples that die-off rapidily at higher frequencies. The impedance as a function of distance (z) along the taper (length L) is given by equation 3.1-1 below.

Z ( z ) = Z 0 e 2( z / L )

ln( Z L / Z 0 )

for 0 z L/2 for L/2 z L

Eq 3.1-1

Z ( z ) = Z 0 e ( 4 z / L2 z

/ L2 1)ln( Z L / Z 0 )

EN/F AD 109 251 R1A C:\DOCUME~1\IBMUSE~1\MYDOCU~1\MYDOCS~1\Stepped_Impedance_A.doc

www.activefrance.com
Prepared By :

TECHNICAL NOTES
No.

9 (10)

Neill Tucker
Project Title :

09:001
Date Rev

Stepped Impedance Matching 3.3 Klopfenstein

06/02/2010

File C:\ My documents \My Docs\ Stepped_Impedance_A.doc

Tapered matches could be based on either of the profiles already described, or indeed any profile you could imagine. However, the Klopfenstein taper is generally considered to offer the best performance in that for a fixed length, the ripple is minimum over the pass band. Or, for a given ripple value, it is the taper that can be realised in the shortest length. The natural log of the impedance as a function of distance (z) along the taper (length L) is given by equation 3.3-1 below.

ln Z ( z ) =

T 1 ln Z 0 Z L + load A 2 (2 z / L 1, A) 2 cosh A

for 0 z L

Eq 3.3-1

Where :

T A = a cosh load T ripple

Tload =

ZL Z0 1 ZL ln ZL + Z0 2 Z0

RdB / 20 and Tripple = 10

Although cited as approximately equivalent, I found that better results are obtained by using the log-quotient approximation for Tload. This is probably due to the mathematics of the derivation of equation 3.1-1 itself.

( x, A) =
Where : x=2z/L-1

I1 ( A 1 y 2 ) A 1 y2
and

dy

I1(x) is the modified Bessel function

The function Phi needs to be evaluated numerically, but note that the limit x is dependent on the position along the taper z. The optimum length for this taper design is 0.565 lambda.

EN/F AD 109 251 R1A C:\DOCUME~1\IBMUSE~1\MYDOCU~1\MYDOCS~1\Stepped_Impedance_A.doc

www.activefrance.com
Prepared By :

TECHNICAL NOTES
No.

10 (10)

Neill Tucker
Project Title :

09:001
Date Rev

Stepped Impedance Matching 3.4 Comparison

06/02/2010

File C:\ My documents \My Docs\ Stepped_Impedance_A.doc

The graphs in figure 3.4-1 shows the output from running the example file excomp1.m. The plots are for the Exponential, Triangular and Klopfenstein tapers implemented as half-wave structures at f0=1000MHz using 50 x 0/100 sections, used to match 100ohm to 50ohm lines.
Return Loss (Zo=50) 0 -5 -10 -15 Return Loss dB -20 -25 -30 -35 -40 -45
50 Impedance (Ohms) 100

Exponential Triangular Klopfenstein

Transformer Impedances 110

90

80

70

60

Exponential Triangular Klopfenstein 0 5 10 Zo 15 20 25 30 35 40 Matching Section Number Zload 45 50

-50

2000

4000 6000 Frequency MHz

8000

10000

Figure 3.4-1 Taper response comparisons

REFERENCES [1] Microwave Engineering 2nd Edition by David M. Pozar Published by Wiley ISBN 0-471-17096-8 [2] RF Circuit Design by Chris Bowick Published Newnes ISBN 0-7506-9946-9 [3] FastCAD32D A drafting package used to text the DXF export.

EN/F AD 109 251 R1A C:\DOCUME~1\IBMUSE~1\MYDOCU~1\MYDOCS~1\Stepped_Impedance_A.doc