ECE4902 Lecture 12
Improving Performance: Cascode Op-amp *
Tradeoff: + Higher node impedance
- Reduced signal swing
2-Stage Op-Amp
CL, Phase Margin Issue
1-stage op-amp (Johns & Martin 6.2)
Improving Performance: Cascode
* Bandgap Voltage Reference
( * Optional Lab Circuit )
Hand In:
HW 5
Handout:
Op-Amp Slides
Cascode Op-Amp
Bandgap Voltage Reference
Principle
Circuit
ECE4902 B2015 Lec 12
1
Review: Small signal (ac) model of 2-stage op-amp
ECE4902 B2015 Lec 12
2
Complete transfer function, stability plot
Increasing node impedance at 1st stage output does not affect phase margin! Separate gain, stability problems
ECE4902 B2015 Lec 12
3
CASCODE OP-AMP
Start with your previous op-amp, compensated. Hook it up for an inverting gain of 100.
Use a 10mV input sine wave at 100Hz to get an output swing of around 1V.
The nice thing about doing it this way is you can look directly at the signal at the - input,
Cascode Op-Amp (Optional
10)to ground, and view it up close on the scope: since there's no DC
whichLab
is close
component, expanding to the max scale will not shoot the signal off the screen.
Without the cascode, your open loop gain of about 1000 should give a 1mV sine wave at
the - input. With the scope probe on 1X and the scope on 2mV/division, you should eb
able to (barely) see a signal at the - input. Record the signal amplitude as best you can,
then do the cascode as indicated below:
The nice thing about doing it this way is you can look directly at the signal at the - input,
which is close to ground, and view it up close on the scope: since there's no DC
component, expanding to the max scale will not shoot the signal off the screen.
Without the cascode, your open loop gain of about 1000 should give a 1mV sine wave at
the - input. With the scope probe on 1X and the scope on 2mV/division, you should eb
able to (barely) see a signal at the - input. Record the signal amplitude as best you can,
then do the cascode as indicated below:
CASCODE
GOES HERE
CASCODE
GOES HERE
ECE4902 B2015 Lec 12
4
ECE4902 B2015 Lec 12
5
2-Stage Op-Amp Phase Margin vs. CLOAD
•CLOAD = 10pF, 100pF, 1000pF
•Phase margin !M degrades as CLOAD increases
#$#%&'(")('*+",-."*("
!"
1-Stage (“gm-C” “transconductor”) Op-Amp
•Only one high impedance node: CLOAD compensates
#$#%&'(")('*+",-."*("
/"
1-Stage (“gm-C” “transconductor”) Op-Amp
•CLOAD = 1000pF, 10000pF: !M same as CLOAD !
•Unity gain frequency fT worse, but always stable
#$#%&'(")('*+",-."*("
0"
1-Stage Problem
•Lousy DC gain ! 90 (39 dB)
•Solution: Add cascode to high impedance node
#$#%&'(")('*+",-."*("
&"
1-Stage with Cascodes
#$#%&'(")('*+",-."*("
*'"
1-Stage with Cascodes
•CLOAD = 1000pF, 10000pF: !M same as CLOAD !
•Unity gain frequency fT worse, but always stable
#$#%&'(")('*+",-."*("
**"
Bandgap Voltage Reference (Optional Lab 11)
Motivation: DC Bias
ECE4902 B2015 Lec 12
12
Bandgap Principle
VBE
T = 0° K
(ABSOLUTE ZERO)
ECE4902 B2015 Lec 12
13
Bandgap CircuitBANDGAP
VOLTAGE
REFERENCE
BANDGAP
BANDGAP
VOLTAGE
VOLTAGE
REFERENCE
REFERENCE
Here's
bandgap
circuit
Here's the
the
Here's
bandgap
the bandgap
circuit covered
covered
circuit in
covered
in Friday's
Friday's
in lecture:
Friday's
lecture: lecture:
R
R22
R
R11
R2
R1
V
VBE2
BE2
VBE2
V
VBE1
BE1
VBE1
The
is
array.
data
is
on
The CA3046
CA3046
The CA3046
is aa 55 NPN
NPN
is atransistor
transistor
5 NPN transistor
array. Its
Itsarray.
data sheet
sheet
Its data
is available
available
sheet is available
on the
the course
course
on thewebsite.
website.
course website.
You
probably
need
the
240!
resistor
to
achieve
bandgap
voltage
You will
willYou
probably
will probably
need to
to adjust
adjust
need to
the
adjust
240!the
resistor
240!value
value
resistor
to value
achieve
to the
the
achieve
bandgap
the bandgap
voltage voltage
of
=
1.205V.
Once
that's
done,
check
the
various
voltages
throughout
the
circuit
of V
VBG
=
of
1.205V.
V
=
1.205V.
Once
that's
Once
done,
that's
check
done,
the
check
various
the
voltages
various
voltages
throughout
throughout
the
circuit
the(is
(iscircuit (is
BG
BG
the
"V
across
the
51!
what
you
expected
based
on
the
analysis
in
lecture?).
Note
the
the "VBE
across
"VBEthe
across
51! the
what
51!
youwhat
expected
you expected
based onbased
the analysis
on the analysis
in lecture?).
in lecture?).
Note theNote the
BEthe
room
values
V
,
V
,
"V
,
and
V
.
room temperature
temperature
room temperature
values of
ofvalues
VBE1
,
of
V
V
,
"V
,
V
,
and
,
"V
V
,
.
and
V
.
BE2
BE
BG
BE1
BE2 BE1 BE
BE2
BE
BG
BG
If
access
aa heat
gun
(or
or
some
other
means
heating
the
If you
you have
have
If you
access
haveto
toaccess
heatto
gun
a heat
(or hair
hair
gun dryer,
dryer,
(or hair
ordryer,
some or
other
some
means
otherof
ofmeans
heating
ofup
up
heating
the up the
CA3046
array),
ahead
and
blast
some
hot
air
the
chip.
the
CA3046CA3046
array), go
go
array),
aheadgo
and
ahead
blastand
some
blast
hotsome
air on
on
hot
theairCA3046
CA3046
on the CA3046
chip. Measure
Measure
chip. Measure
the new
new the new
values
of
V
,
V
,
"V
,
and
V
.
Do
the
changes
correspond
to
what
you
expected
values ofvalues
VBE1
VBE2
V,BE1
"V
,V
, and
, "VVBE
, .and
DoVthe
Do the changes
correspond
correspond
to what you
to what
expected
you expected
BE
BG
BE1,of
BE2
BE
BE2
BG
BG. changes
based
the
in
Note
the
may
be
good
based on
onbased
the analysis
analysis
on the analysis
in lecture?
lecture?
in lecture?
Note that
thatNote
the behavior
behavior
that the behavior
may not
not may
be as
asnot
good
be as
as expected:
good
expected:
as expected:
the
depends
on
between
resistors
R
tracking
over
the lecture
lecture
theanalysis
analysis
lecture analysis
depends depends
on the
the ratio
ratio
on the
between
ratio between
resistorsresistors
R11 and
and R
RR
tracking
and
R
tracking
over
over
22 1
2
temperature.
This
happen
both
resistors
are
but
temperature.
temperature.
This will
willThis
happen
willwhen
when
happen
both
when
resistors
both14resistors
are on
on the
the
aresilicon
silicon
on thedie,
die,
silicon
but with
with
die, but with
ECE4902 B2015 Lec 12
discrete
in
will
good.
discrete resistors
discrete
resistorsresistors
in the
the lab
labinthe
the
thetracking
tracking
lab the tracking
will not
not be
will
be as
asnot
good.
be as good.