- Parent Directory - MIF File -
![]()
A set of 6 chips has been designed at UC
Berkeley in the Micro Linear 1.5
m
BiCMOS tile array technology. These chips contain individual
devices and building-block analog and digital circuits in a
modern BiCMOS process; they are the foundation for a one-semester
junior-level laboratory at Berkeley. The laboratory supports a
junior-level, core microelectronics course using the new text Microelectronics:
An Integrated Approach, by R. T. Howe and C. G. Sodini
(Prentice Hall, 1996). However, experiments based on the devices
and circuits could be used in device physics courses or analog IC
design courses. There are several circuits, such as two BiCMOS
operational amplifiers on Lab Chip 6, which are not used in the
Berkeley lab. The laboratory manual used at Berkeley is being
placed in the public domain for the benefit of faculty, who are
free to duplicate or build upon the experiments. The text of the
experiments in various formats will be made available through the
book's web site at www.prenhall.com.
A brief history of this project is provided on
p. 4, with acknowledgments to the many contributors at Micro
Linear and at UC Berkeley since its inception in early 1992.
All chips are packaged in 28-pin, 600 mil-wide
DIP (Dual In-line Package) and have the same power (VCC - pin 28)
and ground (VSS - pin 14) pins. For manufacturing purposes, Pins
1 and 2 on each chip are reserved for the die identification. The
chip set has been carefully designed to meet customary
electrostatic discharge (ESD) specifications; however, it is
recommended to handle the chips with ground straps and proper
precautions. The chips have proven to be robust over several
semesters of use at UC Berkeley.
The first chip supports experiments that
introduce students to the concepts of IC fabrication technology,
layout, and sheet resistance. A common way to realize resistance
on integrated circuit is to use the polysilicon layer. Two
polysilicon resistors (RP1-RP2 and RP3-RP4) are included on this
chip. RP1 corresponds to pin 21 on the pinout for Lab Chip 1 (see
p. 12). Students are to measure the resistances of these two poly
resistors and extract the sheet resistance. A diffused resistor
is also included (RB1-RB2), with the well contact brought out as
(RBW). The effect of the depletion width on the sheet resistance
can be studied by varying the well voltage.
With the increasing size of silicon die, metal
interconnects may contribute a significant amount of resistance.
A long metal runner (pin 13 - pin 15) is included for students to
gain an appreciation for the differences between metal and
polysilicon sheet resistances. Capacitors are the other common IC
passive component. In this technology, capacitors are realized by
overlapping polysilicon and the diffusion layer, with gate oxide
as the dielectric. A single capacitor (CP-CN) is designed in Lab
Chip 1.
In order to support device characterization
experiments, one or more devices of each transistor type (NMOS,
PMOS, npn, and pnp) are included. These devices are located on
Lab Chips 1 and 2. By placing transistors in series, three NMOS
channel lengths are realized on Lab Chip 1 (1.5
m for NMOS1, 3
m for NMOS2, and 9
m for NMOS3). Because the chip set is
fabricated with 1.5
m BiCMOS
process, the shortest devices exhibit short-channel effect. Since
this is an n-well process, the well for the NMOS transistors is
connected to the substrate, which is hard-wired to ground. One
lateral pnp is also included on the first chip for device
characterization.
Lab Chip 2 includes a diode and an RC
delay line. There are three PMOS transistors (PMOS1,2, and 3 with
channel lengths 1.5, 3, and 9
m),
which share a common n-well that is brought out as NWELL (pin
12). Two types of vertical npn transistors, one a medium power
device, are included. CMOS inverters that exhibit different
propagation delays are also designed in Lab Chip 2.
A two-input NOR gate and two-input NAND gate
are included in Lab Chip 3, along with a ring oscillator. The
ring oscillator consists of 143 stages with buffers attached to
the output to drive parasitic and external capacitances. The
free- running frequency is roughly 10-15 MHz. A a three-input
dynamic logic gate is present on this chip, which is driven by an
external clock.
Three single-stage amplifiers are included on
this chip. The first two are common-emitter amplifiers (with and
without emitter degeneration) that require an external resistor
to supply collector current. By connecting a resistor between VCC
and the CE-BIAS pin, the base-emitter voltage is set by an
on-chip current mirror. In order to minimize the background
required for this basic lab, students are not shown the details
of the transistor bias. The third amplifier is a self-biased
DC-coupled common-emitter with a current-source supply (active
load). An external low-pass filter is needed to implement the
feedback bias scheme.
This chip has several MOS amplifiers. A
self-biased common-source amplifier which is very similar to the
bipolar version on Lab Chip 3 is designed with MOS transistors.
Due to the low transconductance of the MOS transistors, we are
able to place simple common-source amplifiers with current-source
supplies (active loads) with both NMOS (CSAL pin labels) and PMOS
(PCSAL pin labels) (both with NMOS and PMOS) on this chip. The
gain can be found from the transfer function using general lab
test equipment.
Buffers are important in digital circuits for
driving capacitive loads; they also play a key role in analog
circuits. Lab Chip 4 includes both MOS (common drain/source
follower) and bipolar (common collector/emitter follower) forms
with internal current mirror biasing. By connecting a bias
resistor between VCC and the biasing pin (i.e., EF_BIAS or
SF_BIAS), a bias current is set up for the amplifier. With
properly chosen input DC value, the circuit can easily be biased
in the linear range.
Cascoded amplifiers are very useful for their
large gain-bandwidth product. A self-biased cascode amplifier
(SBCASBJT pin labels) is implemented with bipolar devices on this
chip. By applying the same low pass filter technique (connected
between pins SBCASBJT_LP1 and _LP2) with DC feedback as used on
Lab Chip 3, this circuit can be biased in the high-gain region.
In analog integrated-circuit design, currents
sources based on mirrored devices play a central role. Several
current sources are included in the Lab Chips, with SBSOURCE
being a simple pnp current mirror implemented in Lab Chip 4.
Two additional current sources are placed on
this chip. CBSOURCE implements base-current compensation current
source to demonstrate the effect of base current in pnp
transistors. In order to achieve a high output resistance, a
cascode PMOS current source (CASMSOURCE) is designed. An NMOS
current sink (to VSS) and a cascoded npn current sink (CASBSINK)
are included on Lab Chip 5. A "totem pole" voltage
source (TOTEM) is included in this chip, which demonstrates that
several voltages can be generated as a function of an external
resistor.
A self-biased MOS cascode amplifier (SBCASMOS
pin labels) complements the bipolar cascode on Lab Chip 4. This
circuit requires an external low-pass filter to set a stable
operating point using DC feedback.
Differential amplifiers are an essential
component in analog IC design. A two-stage MOS differential
amplifier (MOSDP pins) using resistors to supply the drain
currents is included in Lab Chip 5. An input can be applied
either differentially or single-endedly to the input pads and
outputs can be taken either at the output of the first stage or
the second stage to observe the gain after each stage.
This chip includes a bipolar differential
amplifier with external resistor loads (BFTDPM) and a
differential amplifier using emitter-degenerated bipolar input
transistors (BJTDPE). Due to the high transconductance of the
bipolar transistor, only a single-stage differential pair is
designed for the bipolar implementation. The differential
amplifier affords an opportunity to measure the matching of
nearby and distant npn bipolar transistors, by measuring the
input offset voltage. BJTDM (M for matched=nearby input
transistors) and BJTDU (U for unmatched=distant input transistor)
are nominally identical circuits on Lab Chip 6 that designed to
investigate this effect.
Two operational amplifiers that are designed
for internal applications in VLSI systems are included on Lab
Chip 6. OPAMP1 is a simple two-stage BiCMOS op amp that has a
common-emitter second gain stage, in order to illustrate
inter-stage loading effects. OPAMP2 has a common-collector
voltage buffer incorporated into the second gain stage in order
to achieve higher gain. In order to minimize the systematic input
offset voltage, the input-stage's current mirror supply is
cascoded.
This chip set is used in thirteen one-week
experiments in the laboratory accompanying EE 105 at UC Berkeley.
The devices are characterized fully using HP Semiconductor
Parameter Analyzers. Not all of the circuits are measured due to
time limitations. In particular, the op amps on Lab Chip 6 are
not used in EE 105. The experience of using three revisions of
the Lab Chips, starting in Fall 1993 has been positive. Students
are able to study devices from a modern integrated circuit
process. The Lab Chips include "bare," accessible
versions of the core IC digital and analog building circuit
building blocks that are introduced in the lectures. Therefore,
the lab experiments serve to reinforce the lectures and vice
versa.
The Lab Manual for EE 105 is attached. Since
the final release of the Lab Chips is the third version used in
the Lab, there may be cases where the pin labels were not
changed. In addition, there are certainly typographical errors
and much room for improvement. They are provided with the hope
that they will provide a platform for others to build upon and
improve. I would appreciate any feedback on errors and new
experiments based on the Lab Chips.
A Brief History of the Project
EE 105 is intended to introduce the basic
principles of analog and digital integrated circuit
design, using both MOS and bipolar transistors. Since the focus
in on chip-level rather than board- or system-level design, the
laboratory experiments should emphasize device and circuit
phenomena and trade-offs that are encountered in the
"internal" IC environment. It was clear in 1991 that
the laboratory was not supporting the goals of the course. The
experiments dated back to the 1970s and were typically focussed
on board-level design using simple (and obsolete) ICs. Students
were not impressed with the level of technology in the laboratory
for this core course.
In order to achieve a more effective
laboratory, it became clear that the only route was to design
special ICs that contain MOSFET, bipolar transistor, and passive
test devices, along with the integrated-circuit building-blocks
that form the core of the circuit-design material in EE 105. In
January 1992, a meeting on this concept with Prof. Paul R. Gray,
then the EECS Department Chairman at Berkeley, led to his
initiating discussions with Dr. Jim McCreary, then Vice President
of Engineering at Micro Linear, an IC company based in San Jose,
California.
Micro Linear is a leader in high-performance
mixed analog/digital ICs that achieve system-level integration
for applications in the automotive, hard-disk drive, and
telecommunications industries. One of Micro Linear's innovations
is the user-specified integrated circuit (USIC), which is based
on "tile arrays" of uncommitted transistors and passive
elements. By completing the fabrication process using two levels
of metal interconnections, Micro Linear enables customers to
design their own USICs with very rapid turn-around and very low
risk of a design error. Micro Linear's CA122 BiCMOS Tile Array
chip was ideally suited for the kinds of experimental circuits
needed for the new EE 105 laboratory. Dr. McCreary authorized
Micro Linear's commitment of engineering and production resources
to support the experimental project in April 1992.
During the summer of 1992, the EECS Department
at Berkeley supported Chris Rudell, a graduate student in the IC
group, in the development, design, simulation, and layout of the
first generation of seven EE 105 lab chips. The Fall 1993 version
of EE 105 was the first course using the new laboratory. After
two semesters of experience with the first and second generations
of the chip set, the laboratory manual was completely overhauled
during the summer of 1994. George Chien, a Berkeley IC group
graduate student working at Micro Linear during that summer,
designed the final set of 6 Lab Chips.
Many people have made significant contributions
to the development of the Lab Chips. Micro Linear's corporate
support for this project has been based on a commitment to
improve undergraduate electrical engineering education. In
addition to Jim McCreary, Dave Schwan provided major help with
the layout of the initial version, Henry Young saw the second
version through production and assembly, and Ken Fields was the
contact and project advocate during the final revision. Carlos
Laber supported the development and release of simplified SPICE
models.
At Berkeley, Chris Rudell spent countless hours
--far beyond those paid by the Department -- on refining the
concepts and implementing the designs for the first version of
the chips. George Chien developed the final layout while spending
a summer at Micro Linear.
Over the past three years, several graduate and
undergraduate students have helped to develop the experiments.
The first lab manual was developed in 1993 by graduate students
Chris Rudell and Srenik Mehta, along with undergraduates Albert
Lo, Young Shin, Jim Young, Dennis Yee, and Wayne Yeung. During
the summer of 1994, Wayne Yeung overhauled the lab manual to
reflect the introduction of changes in the second version of the
Lab Chips. The current lab manual was contributed to by Frank
Cheung, Rafael Cortina, Charles Hsiung, Ming Yang, Chi Yeung, and
Wing Yeung. Frank Cheung also developed the simplified SPICE
models for the active devices.
Circuit Schematics for Lab Chips 1-6
II. UC Berkeley EE 105 Laboratory Experiments
Exp. 1 Introduction to Electronic Test
Equipment
Exp. 4 MOS Device Characterization
Exp. 5 Inverter Characteristics
Exp. 7 Bipolar Junction Transistor
Characteristics
Exp. 8 Single Stage Amplifiers with Passive
Loads -- BJT
Exp. 9 Single Stage Amplifiers with Passive
Loads -- MOS
Exp. 11 Current Sources and Voltage Sources
Exp. 13 Differential Amplifiers
Exp. 14 Frequency Response of Differential
Amplifiers
Micro Linear's 1.5
m, 5 Volt BiCMOS process is an advanced,
state-of-the-art process technology optimized for exceptional
analog and mixed signal performance. This technology is a 16 mask
process with 62 total processing steps. It is currently a single
polysilicon, double metal technology but has the capability for
double polysilicon and triple metal interconnects.
This BiCMOS process is robust and reliable. It
has a p+ buried layer to eliminate latch-up problems
common to CMOS, a DDD n+ source/drain to prevent hot
electron formation, and 1st and 2nd metal
planarization for high yield and reliability.
Precision passive components can be implemented
easily in this process. Stable, low temperature-coefficient, low
parasitic-capacitance resistors are made using a polysilicon
deposition. Stable, low temperature-coefficient capacitors are
made from a polysilicon gate/gate oxide/sinker collector
sandwich. The npn transistors have low collector saturation
resistance with the addition of a sinker and n+ buried
layer. The MOS transistors exhibit good and repeatable
performance due to the self-aligned polysilicon gate.
In order to minimize noise and leakage
currents, the process has special enhancements over a
conventional BiCMOS technology. It is capabile of implementing
large-scale mixed-signal circuits in a space- and cost-efficient
manner. The combination of high-performance bipolar and MOS
devices allows for realization of a wide variety of analog and
digital circuits. The npn bipolar transistors have a transition
frequency fT = 4 GHz. In combination
with the L = 1.5
m MOSFETs,
mixed-signal circuits can be implemented with 300-500 MHz
bandwidths. Switching speeds are less than 3 ns for CMOS logic
and less than 500 ps for bipolar emitter-coupled logic.
![]()
SPICE simulation is a very useful component of
pre-lab exercises. The MOSFETs in the Micro Linear 1.5
m BiCMOS technology exhibit
short-channel effects that require higher-level SPICE models such
as BSIM for accurate simulation. Reasonably accurate results can
be obtained from the Level 1 SPICE model, if it is optimized for
the operating conditions of the particular circuit. In the
following list, each MOSFET has an "analog" and a
"digital" model. Note that there are large
discrepancies between the two for the NMOS transistor.
NMOS Level 1 Analog SPICE Model
(W/L) =
(46.5
m/1.5
m)
VTO = 0.77
GAMMA = 0.80
KP = 86u
PHI=0.74
LAMBDA = 0.080
RD = 0
RS = 98
CGDO = 4.1E-10
CGSO = 4.1E-10
CJ = 5.7E-4
CJSW = 1.5E-10
MJ = 0.5
MJSW = 0.33
PB = 0.79
NMOS Level 1 Digital SPICE Model
(W/L) =
(46.5
m/1.5
m)
VTO = 3.2
GAMMA = 0.22
KP = 230u
PHI=0.62
LAMBDA = 0.037
RD = 7.5
RS = 7.5
CGDO = 4.1E-10
CGSO = 4.1E-10
CJ = 5.7E-4
CJSW = 1.5E-10
MJ = 0.5
MJSW = 0.33
PB = 0.79
PMOS Level 1 Analog SPICE Model
(W/L) =
(46.5
m/1.5
m)
VTO = -0.66
GAMMA = 0.42
KP = 26u
PHI=0.77
LAMBDA = 0.13
RD = 49
RS = 100
CGDO = 5.6E-10
CGSO = 5.6E-10
CJ = 3.9E-4
CJSW = 1.2E-10
MJ = 0.5
MJSW = 0.33
PB = 0.87
PMOS Level 1 Digital SPICE Model
(W/L) =
(46.5
m/1.5
m)
VTO = -1.2
GAMMA = 0.53
KP = 24u
PHI=0.5
LAMBDA = 0.12
RD = 0
RS = 26
CGDO = 5.6E-10
CGSO = 5.6E-10
CJ = 3.9E-4
CJSW = 1.2E-10
MJ = 0.5
MJSW = 0.33
PB = 0.87
3505S (similar to 3500, 3501) vertical npn
.model 3505S NPN IS=5.65E-17 RB=465 RE=0.495
RC=41.5 CJE=0.555p
+CJC=0.155p CJS=0.205p BF=94.5 VAF=50.0 BR=0.720 VAR=4.40
VJE=0.960
+MJE=0.330 VJC=0.730 MJC=0.520 VJS=0.450 MJS=0.330 TF=39.0p
TR=1.00n
.model 3505AS NPN IS=3.10E=17 RB=64.0 RE=0.100
RC=27.4 CJE=2.65p
+CJC=0.445p CJS=0..360p BF=98.0 VAF=50.0 BR=7.22 VAR=4.40
VJE=0.960
+MJE=0.330 VJC=0.730 MJC=0.520 VJS=0.450 MJS=0.330 TF=39.0p
TR=1.00n
.model 3516S PNP IS=1.54E-17 BF=12.5 VAF=29.5
BR=50.0 VAR=29.5
+RB=220 RE=4.00 RC=800 CJE=0.133p VJE=0.900 MJE=0.200 CJC=0.043p
+VJC=0.65 MJC=0.5 TF=0.5n
![]()
100 ![]()
500 ![]()
1 k
(2)
2 k
(2)
2.5 k
(2)
3 k![]()
(2)
5 k![]()
10 k
(2)
16 k
20 k
(2)
50 k
(2)
90 k
100 k![]()
(2)
112.5 k![]()
(6)
250 k
(2)
1 M
(2)
5 M![]()
1 pF
(5)
3.6 pF
1 nF
10 nF
100 nF
1
F
10
F
(2)