Modelling parasitic effects of plastic encapsulated

Anuncio
47
By Juan Saiz-Ipiña1, Miguel Angel Solano1, Constantino Perez-Vega2 and Angel Mediavilla2 Jose-Maria Zamanillo2
Modelling parasitic effects of plastic
encapsulated microwave devices
This paper shows a new and accurate electrical modelling technique for low cost active plastic
microwave encapsulated devices. The method employs analytically derived expressions, and is based
on the analysis of measured S-parameters over an appropriate frequency range. The approach
produces a good fit between measured and simulated S-parameters for plastic packaged devices, as
demonstrated in the results for PHEMT devices up to 26.5GHz and for low-cost LNAs up to 6GHz.
ZUSAMMENFASSUNG
Low cost hermetical
plastic package
Bond wires
Modellierung parasitärer Effekte bei
Mikrowellen-Bauteilen in PlastikGehäusen (P.E.M.)
Dieser Beitrag zeigt eine neues und genaues
Verfahren für die elektrische Modellierung
von preisgünstigen Mirowellen-Bauteilen in
Plastik-Gehäusen. Die Methode benutzt
analytisch gewonnene Beziehungen und
basiert auf der Analyse von gemessenen SParametern über einen entsprechenden
Frequenzbereich. Diese Lösung gibt eine
gute Übereinstimmung zwischen
gemessenen und simulierten S-Parametern
für Bauteile in Plastik-Gehäusen,
demonstriert an den Ergebnissen für
PHEMTs bis zu 26.5 GHz und billigen LNAs
bis zu 6GHz.
SOMMAIRE
Modéliser les effets parasites des
dispositifs micro-ondes en boîtier
plastique (P.E.M.)
Cet article montre une technique nouvelle
et précise de modélisation électrique pour
des dispositifs micro-ondes en boîtier
plastique à prix réduit. La méthode emploie
des expressions évaluées analytiquement,
et est basée sur l’analyse des paramètresS mesurés dans une gamme de fréquence
appropriée. L’approche produit une bonne
correspondance entre les paramètres-S
mesurés et simulés pour des dispositifs en
boîtier plastique. Les résultats ont été
démontrés pour des dispositifs PHEMTS
jusqu’à 26.5GHz et pour les LNA à faible
coût jusqu’à 6GHz.
MMIC (die)
External pads
o meet the ever-increasing demands of
today’s communications systems, die
technologies such as silicon FET (including SiGe) and GaAs are being used, either separately or combined. In this way,
monolithic microwave GaAs ICs (MMICs)
are usually used in wireless, airborne and
space applications. A few years ago, only
two options were possible for the microwave
designers: hybrid circuitry using unpackaged
MMICs directly bonded to the circuit, or expensive ceramic hermetically packaged
MMICs. Advances in packaging technology
during the last decade have made it possible
for plastic packages to be used at RF and microwave frequencies, typically at L-band
(0.39-1.55GHz) and S-band (1.55-5.2GHz),
and different authors have studied these applications [1-2]. At higher frequencies, parasitic elements make the use of this type of
packaging difficult, and electromagnetic theory, including methods of moments or finite
differences, must be applied [3]. For space
applications, plastic packages appear as a
very interesting implementation option given
the important associated mass and size reduction, and several authors have studied the
effects of operating in space such as outgassing, thermal effects and reliability [4]. On
T
Figure 1(a): Schematic diagram of a SOT-23
plastic package, showing its internal parts
the other hand, the cost of these plastic packaged MMICs is considerably lower than that
of ceramic ones, because they use industrial
standard processes to manufacture them.
Figure 1 shows a schematic section of a standard hermetic plastic SOT-23 package showing its internal parts. Brown and Hiller [5]
have modelled this package for microwave
diodes up to 10GHz. However, designing
MMICs for use in such packages is difficult
due to the poor grounding characteristics of
the elevated paddle (MMIC ground), and
also to the lack of generally applicable circuit
models for the package. This difficulty is exacerbated for designs at frequencies higher
than 10GHz, and for packages with a great
number of pins. The purpose of this paper is
to show the accurate modelling of these effects, in order to use low cost plastic encapsulated microcircuits (PEM) at higher frequencies (up to Ku Band). The results shown
in this paper are focused on two types of plastic packaged CFY77-08/10 PHEMT devices
from the Infineon foundry. These transistors
are low noise (0.8 and 1dB respectively)
small signal devices, usually used for front-
Microwave Engineering ● May 2001● www.mwee.com
48 PACKAGE MODELLING
Packaged device
Packaged device
CIn-Out
Cpgd
Lpd
Rpd
Gate
Die
device
Cpg
Rpd
Lpd
LpIn
Drain
In
RpIn
Die
device
CpIn
Cpd
LpOut
Out
CpOut
Ground
Source
end amplifiers up to 20GHz and for DBS
down-converters in die form (without package). Both devices use the same plastic package, designated by the foundry as MW-4. Figure 1(b) shows the electrical diagram used
to model such package. The same technique
has been used for modelling commercial
plastic packaged LNAs from the Agilent
Technologies and Mini-Circuit foundries up
to 6GHz with very good results. The electrical model used for these amplifiers is shown
in Figure 1(c).
One of the main objectives in the characterisation of MMIC packages involves accurately determining the values of parasitic
elements - loss resistance, capacitance and
inductance for each pad - either theoretically
or experimentally. This data can be used to
describe an equivalent circuit that explains
the electrical behaviour of PEM devices. At
microwave frequencies these parasitic elements mask the electrical characteristics of
die devices and the connections between
package pins and chips begin to behave as
transmission lines rather than as simple
wire connections [6-7]. Parasitic capacitances, inductances and resistances, such as
pin-to-pin capacitance, as well as wire inductance and resistance, become significantly higher. For the different plastic packaged devices investigated here, parasitic elements are typically of the order of pF, nH,
and m for the capacitances, inductors and
resistances respectively.
This paper shows a new and accurate electrical small signal modelling technique called
DICOMPAK, developed by the authors for
low cost plastic encapsulated devices. In
order to summarise, system configurations,
as well as the hardware and software used to
make laboratory measurements possible, will
not be discussed. On the other hand, some
high performance tools (boxes, chip carriers
and boards) specially developed to measure
the devices under study have are discussed.
Experimental results for the measurements of
RpOut
scattering matrix parameters and using, if
necessary, optimisation algorithms to minimise the error function, it can be used to finally compute a unique set of parasitic values.
To make the electrical characterisation of
these parasitic elements possible, it is necessary to have the same type of component, either packaged or unpackaged dies, in order
to make the necessary measurements of the
device under both conditions.
Figure 1b (above left) and 1c (above right):
Electrical model used for the
characterisation of (a) CFY77-08/10, and
(b) Agilent MODAMP devices in MW-4
plastic packages
the different types of devices are also be presented.
The DICOMPAK technique
The DICOMPAK method uses the scattering parameters measured at different bias
points, under the hypothesis that the parasitic elements do not change with bias because of the plastic packages. This technique can be applied in two different forms
called DICOMPAK I and DICOMPAK II.
The intrinsic equivalent circuit at each bias
point of the active circuits has been computed using conventional techniques introduced by Dambrine [8] and later modified
by our group [9].
DICOMPAK II
The DICOMPAK II technique is very similar to the DICOMPAK I. The main difference between the two techniques is that
DICOMPAK II uses only the measured scattering matrix from the packaged device,
under the assumption that all parasitic elements are null at the first iteration. Afterwards, using an optimisation algorithm and
an iterative method, the plastic parasitic extrinsic elements are computed using the
variance of the intrinsic elements. Values of
the parasitic elements are updated in order
to minimise the error vector e using the sequential quadratic programming (SQP)
method [10]. A computer program using
the flowchart shown in Figure 2 has been
written using MATLAB [11]. At the moment a powerful version of this method is
being developed using Microsoft Visual
BASIC 6.0 [12].
DICOMPAK l
The DICOMPAK I algorithm is based on the
measurement of the scattering parameters at
two or three different bias points for the
same type of component, packaged and unpackaged die. It is assumed that both devices
are identical during the whole characterisation process, and that the plastic package
does not vary with the bias point. Parasitic elements at these bias points can be evaluated
by a conventional de-embedding process. By
comparison between modelled and measured
Microwave Engineering ● May 2001● www.mwee.com
Table 1: Results obtained applying the
DICOMPAK-I technique to the CFY77-08
PARASITIC ELEMENTS FOR CFY77-08 DEVICE
Resistances ()
Inductances (nH)
Capacitances (pF)
Rpg= 3.53
Lpg= 0.608
Cpg= 0.235
Rpd= 15.5
Lpd= 0.630
Cpd= 0.252
-
-
Cpgd= 0.00076
50 PACKAGE MODELLING
Extraction of parastic elements using DICOMPAK technique
[S]Unpak bias1[S]Unpak bias 2 [S]Unpak bias3
DICOMPAK-I
of them use the die model MSA-0600 manufactured by Agilent Technologies. The 7-element electrical model shown in Figure 1(c)
has been selected in order to characterise
the parasitic effects introduced by the plastic package all types of plastic and ceramic
packages of this family of amplifiers. Table
3 shows the results obtained by the application of the DICOMPAK techniques to
these devices. Figures 4(a) and 4(b) show
the comparison between modelled and measured scattering parameters for a plastic
package MSA-686 device. Figures 5(a) and
5(b) show the same comparison for a ceramic package MSA-670 device. In both
cases, a very good fit is observed.
Figure 2:
Flowchart of the
DICOMPAK technique.
DICOMPAK-II
Type of technique?
All parasitic elements
are set to zero at the
first iteration.
Compute initial estimation of
the values of plastic packaged
parastic elements for 3 bias
New values
of parastic
elements
[S]Unpak-modeled
Optimization
process
[S]Pak-modeled
No
[S]Pak-modeled=[S]Pak-measured?
Yes
End
The 7-element electrical model shown in
Figure 1(b) has been selected in order to
characterise the parasitic effects introduced
by the plastic package of devices presented
in this paper. Table 1 shows the results obtained by the application of the above mentioned techniques to a PHEMT device
CFY77-08 from Infineon.
Infineon CFY77 PHEMT devices
In order to validate the electrical model developed for the plastic package, measured
scattering parameters of the packaged device have been compared with measurements of the same device without packaging
and the addition of the electrical model extracted for the capsule. This comparison
offers good agreement for different devices.
As an example, Figure 3(a) and Figure 3(b)
show the results for the CFY77-08 device,
biased at Vds=2V, Vgs=-0.1V, and Id=15mA
up to 26.5GHz.
silicon bipolar MMIC process which utilises
self-alignment, ion implantation and gold
metallisation to achieve excellent uniformity and reliability. This characteristic is
enough for operation at L- and S-bands for
this kind of amplifier. In our study we have
measured different types of devices: die devices, three types of plastic packaged devices
- SOT-143, 85mil (2.16mm) model and
86mil (2.18mm) model - and a ceramic
packaged device - 70mil (1.79mm) package
model. This 70mil device has been studied
in order to evaluate the differences between
plastic and ceramic packaged versions, and
to verify if the DICOMPAK technique is
suitable for application to ceramic packaged devices. This type of amplifier with different plastic packages is commercialised by
Mini-Circuits. Therefore the electrical model
developed for this set of PEM devices is
valid for both brands of amplifiers. Table 2
shows the cross-reference between Agilent
part numbers and those of Mini-Circuits. All
Agilent Modamp and Mini-Circuits devices
The MODAMP MSA series amplifier is fabricated using a 10GHz fT and 25GHz fMAX
Table 2: Cross reference between Agilent
MODAMP and Mini-Ciruits devices, using
MSA-0600 die
Validation
AGILENT PART NUMBER
MINI-CIRCUITS PART NUMBER
Conclusions
A novel method of extracting the electrical
values of parasitic elements from plastic
packaged devices has been presented, along
with experimental results for commercial
plastic-packaged PHEMTs . The method
could easily be generalised for other MMIC
PEM devices, and has been tested for several different microwave packaged GaAs
MESFET and HEMT transistors. Furthermore, it has now been extended to MMIC
packaged devices with a large number of
pins.
As a conclusion of this work, several different models for plastic packaged devices
have been developed using the 7element custom model.
The models that have been developed
work very well up to 3GHz in terms of scattering parameters and figures of merit
(FOM), for the economy package SOT-143.
For 85/86mil plastic packages, the results are
very close to the equivalent 70mil ceramic
package up to 6GHz. For Infineon devices,
the model has excellent performance up to
the Ku band and beyond. As a consequence
of this study, it can be stated that the electrical model works reasonably well for AgilentMODAMP and Mini-Ciruits devices up to
6GHz, more than achieving the frequency
targets of their applications, which are usually centered at L- and S-bands. However, the
PACKAGE TYPE
PACKAGE OUTLINE
MSA-0685
MAR-6
Plastic
85mil
MSA-0686
MAR-6SM
Plastic
86mil
MSA-0611
VAM-6
Plastic
SOT-143
MSA-0670
-
Ceramic
70mil
MSA-0600
-
Die
None
Microwave Engineering ● May 2001● www.mwee.com
PACKAGE MODELLING 53
MODFET
MEAS
MODFET
EE
D
F F
F F
E
MEAS
BB
D
C
S11
S22
0.2
AA
C
0.5
1
2
S21
5
A
FF 1
EE D
D
A
2
C
C
3
4
5
6
B
Frequency [GHz]: 0.5-26.5
Frequency [GHz]: 0.5-26.5
study made here is not comprehensive
enough to develop a general purpose model
for the plastic packages studied. Furthermore, an extension of the model must be developed in the future to make possible the use
of the model at higher frequencies.
In summary, the possible solutions to
higher frequency plastic packages are:
1) To extract a new set of values for the
model by imposition of a best fitting in the
frequency band of interest between measured and modelled scattering parameters.
2) To develop a non-linear model for plastic packages in order to increase the validity bandwidth of the model, using the nonlinear techniques developed by us [13].
3) To develop a new model taking into account second order effects like the increase of temperature and moisture level
of the plastic package during operation.
4) To repeat this study for another family
of devices (GaAs MESFET, HEMTs, and
Figure 3a (above left) and 3b (above right):
Comparison between the measured and
modelled (a) S11 and S22, (b) S21 parameters
for the Infineon PHEMT transistor CFY77-08
GaAs MMICs amplifiers) with the same
packages, in order to generalise the models developed within this work, and in
order to see if the behaviour of these new
packaged devices is similar to the behaviour presented by the devices studied here.
Acknowledgments
This work has been supported partially by
ESA-ESTEC project contract No. 161487
and Tecnologías de Telecomunicaciones y
de la Información (TTI).
wireless communication applications."1998
Radio Frequency Integrated Circuits (RFIC)
Symposium 98. (1998 [RFIC]): 127-130.
[2] D.R.Green, J.P. Beccone and R.B. Crispell.
"Packaging Requirements for High Volume
Wireless Communications." 1992 MTT-S
International Microwave Symposium Digest
92.3 (1992 Vol. III [MWSYM]): pp.15011501.
[3] H.-Y. Lee. "Wideband Characterisation of a
Typical Bonding Wire for Microwave and
Millimeter-Wave Integrated Circuits." 1995
Transactions on Microwave Theory and
Techniques 43.1 (Jan. 1995 [T-MTT]): pp.63-68.
[4] B. Johnson and V. Verma. "Reability
assessment of fielded plastic and hermetically
packaged microelectronics". IEEE Transactions
on Reability, vol 45, No.1, March 1996.
References
[1] V. Krishnamurthy, E. Balch, K. Durocher, J.
Rose, R. Saia, D. Lester and D. Sherwood.
"Plastic microwave multi-chip modules for
Table 3: Results obtained applying the
DICOMPAK II technique to the AgilentMODAMP amplifier series
PARASITIC ELEMENTS OF MSA-600 SERIES
Model
Package
MSA-0611
SOT-143
MSA-0685
85mil
MSA-0686
86mil
Model
MSA-0670
Package
70mil
-
Plastic Package
Resistances ()
RpIN=3.38
RpOUT=13.13
RpIN=5.37
RpOUT=6.52
RpIN=4.70
RpOUT=7.64
-
Ceramic Package
Resistances ()
RpIN=1.78
RpOUT=1.90
-
Inductances (nH)
LpIN=0.934
LpOUT=0.731
LpIN=0.455
LpOUT=0.485
LpIN=1.059
LpOUT=1.047
-
Inductances (nH)
LpIN=0.328
LpOUT=0.327
Capacitances (pF)
CpIN=0.053
CpOUT=0.320
CpIN-OUT=0.264
CpIN=0.221
CpOUT=0.116
CpIN-OUT=0.024
CpIN=0.002
CpOUT=0.116
CpIN-OUT=0.192
Capacitances (pF)
CpIN=0.291
CpOUT=0.113
CpIN-OUT=0.182
Microwave Engineering ● May 2001● www.mwee.com
PACKAGE MODELLING 54
MODFET
MODFET
MEAS
MEAS
?
?
EE
D
B
D
?
O
DD E
E
OO
S11
DD
BB
0.8 S22 0.5 B
B
A1
FA
??
2
?
3
?
??
F
C
C
2D
4
6
8
10
Frequency [GHz]: 0.88-1.6
Frequency [GHz]: 0.88-1.6
[5] B. Brown and G. Hiller. "Circuit Models
for Plastic Packaged Microwave Diodes."
1996 MTT-S International Microwave
Symposium Digest 96.3 (1996 Vol. III
[MWSYM]): pp.1779-1782.
[6] S.-K. Yun and H.-Y. Lee. "Parasitic
Impedance Analysis of Double Bonding Wires
f or High-Frequency Integrated Circuit
Packaging." 1995 Microwave and Guided
Wave Letters 5.9 (Sep.1995 [MGWL]):
pp.296-298.
[7] C. Amrani, M. Drissi, V. Fouad Hanna
and J. Citerne. "Packaging and
Interconnection Mutual Coupling Effects in
Planar Structures and Discontinuities." 1993
MTT-S International Microwave Symposium
Digest 93.2 (1993 Vol II [MWSYM]):pp.
843-846.
[8] G. Dambrine, A.Cappy, F. Heliodore, E.
Playez. "A new method for determining the
FET small-signal equivalent circuit". IEEE
Transactions on MTT vol 36, nº 7, July 1988.
B
Figure 4a (above left) and 4b (above right):
Comparison between the measured and
modelled (a) S11 and S22, (b) S21
parameters for the Infineon PHEMT
transistor MSA-686
[9] J.M. Zamanillo, T. Fernández, Y. Newport,
A. Mediavilla, A.Tazón. "Wideband
Technique Models P-HEMT and GaAs
MESFETs". Microwaves & RF, pp. 60-68,Vol
35, No.2 Feb. 1996.
[10] J.M. Zamanillo, C. Pérez Vega and A.
Mediavilla "Modelling Parasitic Effects of
Plastic Encapsulated Microwave Devices".
Microwave Symposium 2000 (MS’2000)
Proceedings, pp.39-42, Tetouan , May 2000.
Figure 5a (below left) and 5b (below right):
Comparison between the measured and
modelled (a) S11 and S22, (b) S21
parameters for the Infineon PHEMT
transistor MSA-670
MODFET
MODFET
MEAS
MEAS
[11] Using MATLAB Version 5. By
Mathworks Inc. June 1997.
[12] Visual Basic™ 6.0 User´s Guide.
Microsoft Corpration 1998.
[13] T. Fernández, Y. Newport, J.M.
Zamanillo, A. Mediavilla, A. Tazón
"Extracting a Bias Dependent Large Signal
MESFET Model from Pulsed I/V
Measurements". IEEE Transactions on MTT
vol 44, pp.372-378, March 1996.
AUTHOR INFORMATION
Group of Electromagnetism
Communications Engineering Department1
Group of Microwaves and
Communication Systems2
ETSII y Telecommunication, University of
Cantabria, Av. de los Castros s/n 39005
Santander, SPAIN.
Tel:+34 942-201391 Fax:+34 942-201488
[email protected]
BB
?
?
?
DD
S110.8
S22
??
?
0.8
00
00
BB
DD 1
??
??
2
3
D
?2
??
4
6
8
10
?
Frequency [GHz]: 0.88-1.6
Frequency [GHz]: 0.88-1.6
Microwave Engineering ● May 2001● www.mwee.com
Descargar