Design of Mm-Wave Mic Frequency Multipliers and Mixers Using Simple Microstrip High-Pass Filters

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

DESIGN OF MM-WAVE MIC FREQUENCY MULTIPLIERS AND

MIXERS USING SIMPLE MICROSTRIP HIGH-PASS FILTERS

Jerzy Chramiec*, Marek Kitliński#, Andrzej Bochenek+, Bogdan Janiczak#


Abstract: - General requirements on semiconductor device embedding impedances in mm-wave MIC frequency
multipliers and converters have been discussed. Preferred filter types in different circuit paths have then been
defined showing the utility of high-pass filters. Microstrip with short-circuited edge has been introduced as a
high-pass filter realizable at mm-wave frequencies and successfully employed in a 20 – 40 GHz
subharmonically pumped mixer.

Introduction
Parameters of mm-wave Schottky diode MIC frequency multipliers and mixers are strongly affected by the diode
parasitics, circuit losses and the transmission line discontinuities. The chosen topology should be as simple as
possible, especially in the case of broadband circuits. Technological limitations due to very high operation
frequencies should be taken into account and are the next constraint limiting the designer’s choice of circuit
components. The aim of this paper is to discuss the general principles governing the choice of single
semiconductor device multiplier or mixer circuit topology and to show the advantages resulting from the use of a
simple high-pass filters in form of microstrip with short-circuited edge.

Requirements on filters used in mm-wave frequency multipliers and mixers


The design of mm-wave MIC frequency converters starts with the choice of a suitable dielectric substrate of
certain thickness h and relative permittivity εr. A widely accepted formula [1] governing this choice:
c 2
fc = arctg ε r , fmax << fc (1)
2π h ε r − 1
implies the use of low permittivity, thin enough substrates, preferably with low dielectric loss at mm-wave
frequencies. The involved frequency range up to fmax should in principle include not only the frequencies of
primary interest but also higher-order spectral components that should be controlled at the design stage. The
increasing effect of losses as well as complex and diode input and output impedances that are fast-varying with
frequency call for as simple circuits as possible. For this reason single beam-lead semiconductor device circuits
(including single Schottky barrier diode and aniparallel diode pair) are often prefferred at very high frequencies.
Fig. 1 presents typical functional scheme of single device frequency multiplier. Series connection of devices
has been assumed because of its compatibility with microstrip circuit technology.

Zi(f) Zid(f) Zio(f) Zo(f)

f1 , P1 n f1 , Pn
Input filtering Output filtering
& matching network & matching network

Fig. 1. Typical functional scheme of n-th order single device frequency multiplier .

The scheme of Fig. 1 permits to fromulate some convenient requirements on the device embedding network
impedances. They are relatively simple in the case of diode frequency multipliers because of simple frequency
spectra involving the input signal frequency band , corresponding desired output frequency band and frequency
bands of most important non-used harmonics [2] (i.e. the third harmonic in frequency doublers, second and
fourth harmonic in frequency triplers etc.). Loss of power at the non-used harmonics should be avoided so the
minimum requirement is a dominantly reactive load seen by the device at these frequencies.
Open circuit (including the effect of diode parasitic elemnts) provides a nearly optimum load [2]. However, due
___________________________
* Gdynia Maritime University, 81-87 Morska St., 81-225 Gdynia, Poland; [email protected]
# Gdansk University of Technology, 11/12 Narutowicza St., 80-952 Gdansk, Poland; [email protected]
+ Wroclaw University of Technology, 27 Wybrzeże Wyspiańskiego St., 50-370 Wroclaw, Poland;
[email protected]
to the postulated circuit simplicity and strong influence of diode parasitics it may be difficult to achieve at mm-
wave frequencies. For this reason in what follows conditions for non-used harmonics have been omitted. The
following desired conditions may be then written:

Zi(0) = Zo(0) = 0 (DC path in the case of single diode) (2)


|Zo(f1)| << |Zid (f1)| or Zo(f1) ≈ -Im (Zid (f1) (3)
|Zi(nf1)| << |Zod (nf1)| or Zi(nf1) ≈ -Im (Zod (nf1)) (4)
Zi(f1) = Z*id (f1 ) (large signal input matching ) (5)
Zo(nf1 ) = Zopt (optimum large signal load impedance seen by the device) (6)

Conditions (3), (4) are not absolutely necessary but they may simplify the matching problem by limiting the
matched impedance to the diode’s impedance only or even by removing (at least partly) the diode’s reactance at
the input and/or output frequency.
Fig. 2 shows an example of a single semiconductor device frequency down-converter functional scheme where
the local oscillator (LO) diplexer has been placed on the output signal side. Actually the choice depends on the
signal (fs). local oscillator (fLO) and output (f0) frequency bands and should simplfy the design of diplexer and
filters.

Zi(f) Zid(f) Zio(f) Zo(f) fLO, PLO

fs, Ps LO/f0 diplekser , f0 , P0


Input filtering output filtering
& matching network & matching network

Fig. 2. Example of single device down-coverter functional scheme with LO diplekser on the output signal side.

Because of a much more rich frequency spectrum the requirements on the semiconductor device embedding
impedances become more complex. Assuming the frequency conversion relationship

f 0 = ± f s m nf LO (7)

the following essential conditions may be formulated: condition (2) should be retained at DC, condition (3) at fs,
condition (4) at fLO and f0,, condition (5) at fs (but it applies to small-signal impedance of a pumped device),
condition (6) at fLO (optimum source impedance) and f0 (optimum pumped device’s small signal load).
Additional conditions should be met at the image frequency band. They depend however on the converter type
(single or double sideband) and to which part of converter thisfrequency band should be assigned.
Now, we may consider the use of low-pas, high-pass and band-pass filters having in mind the above impedance
conditions. MIC band-pass filters (BPF) with relative bandwidth exceeding several percent must use coupled
microstrip lines and may be not manufacturable due to the required gap widths when realized on thin, low
permittivity substrates. Moreover, being essentially built with the use of quarter-wave or half-wave line sections
at the center frequency they may exhibit the desired impedance conditions such as (3), (4) only far away from
their pass-band. The DC path will be provided only by filters with short-circuited input element.
Microstrip low-pass filters (LPF) with capacitive input element [3] are a convenient choice as the input and
output filters of Fig. 1 and Fig.2, respectively. They are realizable at millimeter-wave frequencies. Their stop-
band reactance has capacitive character and may fulfil the inequalities (3) or (4). The required DC path may be
insured by a high input impedance quarter-wave shorted stub inserted at any convenient place. Problem of
spurious pass-bands may be solved by cascade-connection of filters with increasing cut-off frequencies.
Microstrip high-pass filters (HPF) with inductive input element [3] may be used in the output and input network
of Fig. 1 and Fig. 2, respectively. They provide the required DC path and have inductive reactance in the stop
band which may be employed as a part of the matching circuit. However, practical implementation of these
filters at mm-wave frequencies may be difficult if not impossible. Fig. 3 shows the most useful microstrip high-
pass filter configuration [3] with series interdigital capacitors and parallel inductors realised as short-circuited
microstrip stubs. Practical limitations of this structure at mm-wave frequencies result from interdigital capacitor
parasitic elements (series inductance, parallel capcitances) [4] as well as from the realizability problems of
parallel-connected inductive stubs whose equivalent inductance may be of the same order as the inductance of
grounding via-holes. Coupling between the neigbouring stubs may be also a problem in the case of very short
interdigital capacitors.
Fig. 3. Topology of microstrip five-element high-pass filter with inductive elements at the inputs.

High-pass filters using microstrip with short-circuited edge


The concept of HPFs employing microstrip line with short-circuited edge (MSCE) was advanced by one of the
authors long ago [5] and used in some circuits up to the X-band. Recently modelling of such filters with the use
of electromagnetic analysis program has been successfully performed [6] prompting us to move to much higher
frequencies and thin low permittivity substrates. The obtained results justified a further effort – application of
MSCEs to mm-wave frequency converters.
Fig. 4a recalls the MSCE principle. Because of the short-circuited edge the fundamental electromagnetic field
mode is of the HE type [7], being in fact quite similar to the fundamental TE10 field mode in one-half of an
equivalent rectangular waveguide shown in Fig. 4b. Due to the effect of the MSCE fringing field the equivalent
waveguide width ww is somewhat larger than twice the MSCE’s physical width. Permittivity of dielectric filling
the equivalent waveguide equals the MSCE effective permittivity. Kompa’s experiments on higher-order
microstrip field modes [7] have proved the validity of such an approach. Formulae permitting the evaluation of
MSCE cut-off frequency, fundamental mode wavelength, wave impedance, voltage-power defined characteristic
impedance at the driving point position x (see Fig. 4) and the equivalent circuit below the cut-off frequency may
be found elsewhere [8]. It should be said however that now-a-days they are if indicative nature only as all the
required information is provided by electromagnetic analysis of the structure of Fig. 4a which includes
discontinuity, radiation, higher-order modes and the enclosure effects.
l
(a) l (b)
wc
w x ww
x x
εr h εeff

Fig. 4. a) MSCE topology; b) equivalent rectangular waveguide.

The advantages of MSCE structure in the proposed applications are evident: very simple topology and
technology, DC path, low input reactance at frequencies much lower than the cut-off frequency, inductive input
reactance below the cut-off frequency. However, the pertinent question was: “Can MSCE operate at mm-wave
frequencies”?

MSCE application example: broadband subharmonically pumped mixer


In view to extend the frequency range of a 21 GHz spectrum analyzer a 20 – 40 GHz mixer pumped
subharmonically at 15 GHz has been designed following the general scheme of Fig. 2. Fig. 5 presents the
adopted mixer architecture defining the filters used to separate the most important frequency bands. The chosen
substrate has the parameters h = 0.254 mm, εr = 2.2, leading to fc = 357 GHz. This prevents from the surface
wave excitation with a very good safety margin but excludes the use of broadband BPF in the input path.

fLO 15 GHz

BPF 15 GHz
fs LPF HPF LPF 1 resonator LPF f0
fc = 40GHz MSCE fc = 15.5 GHz fc = 11 GHz
20 - 40 5 elements Pass-band 3 elements 5 elements 0 - 10
GHz from 20 GHz GHz

Fig. 5. Architecture of 20 – 40 GHz mixer.

M/A-COM GaAs MAE 2039 beam-lead antiparallel diodes have been used in experimental circuits. Modelling
of the mixer has been performed using Agilent’s ADS simulator [9]. The required MSCE has been modelled
and optimised employing the Sonnet software [10]. The designed MSCE (w = 2.4 mm, l = 6 mm) has been
experimentally tested. Fig. 6 presents the results. They include the effects of two 1.85 mm coaxial-to-microstrip
transitions and are very satisfactory, proving that MSCE may be used at mm-wave frequencies. It should be
Fig. 6. Measured parameters of MSCE used in 20 – 40 GHz mixer. Triangles - |S11|, diamonds - |S21|.

noted the program enables the analysis of MSCEs with any width and position of the connecting lines (wc in
Fig. 4), e.g. representing the beam-lead device lead or the first microstrip in the 40-GHz LPF of Fig. 5. Possible
modification of MSCE parameters by a local dielectric overlay has been also demonstrated. The MSCE
scattering matrix obtained in such a way is inserted in the complete mixer circuit created under ADS and
simulated using HB engine. As the experimental MSCE evaluation has been possible up to 40 GHz only, its
actual behaviour at higher frequencies has been estimated by the modelling. Good measurement results on the
experimental mixer have been obtained (Fig. 7 and 8). The aforementioned possibility of MSCE parameters
modification with a dielectric overlay has been used to experimentally improve the input SWR above 30 GHz
lowering somewhat the MSCE’s cut-off frequency. This effect is visible on the input SWR characteristic.
50,00 20,00
15,00
Lc - dB

30,00
SWR

10,00
10,00 5,00
-10,00 0,00
0,00 10,00 20,00 30,00 40,00 20,00 25,00 30,00 35,00 40,00
Frequency - GHz Frequency - GHz

Fig. 7. Input (triangles) and output (diamonds) SWR of Fig. 8. Conversion loss of the experimental mixer.
the experimental mixer. PLO = 4 mW. PLO = 4 mW.

Conclusions
Some rules facilitating the design of single semiconductor device mm-wave frequency multipliers and mixers
have been presented. Microstrip with short-circuited edge (MSCE) has been used as a high-pass filter in the
experimental mixer for the 20 – 40 GHz band. MSCE has been also employed in another component, not
described in this paper – a 40–50 GHz frequency doubler. Both circuits have for the first time demonstrated
practical utility of the MSCE concept at mm-wave frequencies.

Acknowledgement
This work has been supported by the State Commitee for Scientific Research (KBN) under grant No
3T11D00626.

References
[1] Bahl I. Lumped elements for RF and microwave circuits, Artech House, Inc., 2003.
[2] Faber M.T., Chramiec J., Adamski M.E., Microwave and millimeter-wave diode frequency multipliers, Artech House,
Inc., 1995.
[3] Jia Sheng Hong, Lancaster M.J , Microstrip filters for RF/microwave applications, J.Wiley & Sons, Inc., 2001.
[4] Esfandiari R., Maki D.W., Siracusa M., Design of inerdigital capatitors and their application to GaAs monolithic filters,
IEEE Trans. Microwave Theory Tech.,. Vol. MTT-31,1983, pp. 57–64.
[5] Chramiec J., A high-pass waveguide applicable to microwave integrated circuits, Proc. 26th Int.Sci.Coll. (Ilmenau), Vol.
A3, 1981, pp. 91–93.
[6] Chramiec J., Kieda R., Kitliński M., Full-wave simulation of microstrip line with short-circuited edge, Proc. of 15th Int.
Conf. MIKON, 2004, Vol. 1, pp. 222-224.
[7] Kompa G., S-matrix computation of microstrip discontinuities with a planar waveguide model, A.E.Ü., Vol. 30, 1976,
pp. 58–70.
[8] Chramiec J., Microwave diode frequency multipliers and converters (in Polish), Gdansk University of Technology
Scientific Papers, Electronics, No 83, 1996.
[9] Agilent Technologies, Advanced Design System 2003 A.
[10] Sonnet Software, v. 9.52, 2003, www.sonnetusa.com

You might also like