When using two masks per layer (Double Patterning Technology,
DPT) there is an issue of mask alignment where any
mis-alignment will cause layer spacing values to change, therefore
changing the parasitic coupling capacitance values.
Misalignment scale and direction are not deterministic facts:
coupling cap and total cap may be increased or decreased.
Five new corners are added in a DPT flow to account
for RC variations accurately:
sapced-dependent side-wall dielectric constant also affect coupling
cap
cworst_CCworst, cbest_CCbest, rcworst_CCworst, rcbest_CCbest and
typical
The others are for pre-color RC calculation purpose
**_T** stands for "Tighten DPT corner"; these are
less pessimistic 1.5 sigma corners
Below table is caputre of Aragio's TSMC16: LVDS datasheet
BEOL corner
Spacing variation is implicitly defined by \(\Delta W_m\).
We denote the conductor width and thickness of the layer m
by \(W_m\) and \(T_m\), respectively.
Similarly, we denote the thickness of the layer's interlayer
dielectric (i.e., the distance between layer m and layer m +1) by \(H_m\)
C-based means worst and best caps
RC-based means worst and best R in adjustment
with C (RC product)
Based on experience, it was found that C-based
extraction provides worst and best case over RC for internal
timing paths because Capacitance dominates short
wire.
However, for large design, inter-block timing paths were often worst
with RC worst parasitic since R dominates for
long wires.
T. -B. Chan, S. Dobre and A. B. Kahng, "Improved signoff methodology
with tightened BEOL corners," 2014 IEEE 32nd International Conference on
Computer Design (ICCD), Seoul, Korea (South), 2014, pp. 311-316, doi:
10.1109/ICCD.2014.6974699.
Chan, T. (2014). Mitigation of Variability and Reliability Margins in
IC Implementation /. UC San Diego. ProQuest ID:
Chan_ucsd_0033D_14269. Merritt ID: ark:/20775/bb52916761. Retrieved from
https://escholarship.org/uc/item/35r1m001
Clock Gating is defined as: "Clock gating is a
technique/methodology to turn off the clock to certain parts of the
digital design when not needed".
Clock Gating Overview
AND gate-based clock gating
In simplest form a clock gating can be achieved by using an
AND gate as shown in picture below
However, this simplest form of clock gating technique has some
problem of generating glitches in the clock provide to
the FF, which are not desirable.
Glitches in enable/gated clock
Latch based clock gating
These glitches can be removed by introducing a negative edge
triggered FF (assuming downstream FFs are positive edge) or low-level
sensitive latch at the output of the clock enable signal.
This will make sure that any glitch in the clock enable signal will
not be visible to the gated clock output. The Latch output will only be
updated during the negative clock cycle and thus input to AND gate will
be stable high.
It seems that dspf_emir don't contain the
rectangle pin information.
only label is necessary
setup
spectre result
netlist type
dspf option
emir analysis
dspf
/
disable
✓
dspf_emir
/
disable
✗
dspf_emir
=shortPins=”yes”
disable
✓
dspf_emir
=shortPins=”no”
disable
✗
dspf_emir
/
enable
✓
dspf_emir
=shortPins=”yes”
enable
✓
dspf_emir
=shortPins=”no”
enable
✓
shortPins=”yes” is preferred default option for
dspf_emir, which has split pins
DSPF Syntax
::=*|P ?
describes pins in the net. Multiple pin descriptions can be listed in
one line.
::=( {}?)
represents the name of the pin. represents the type
of the pin. It can be any of the following: I (Input), O (Output),
B (Bidirectional), X (don’t care), S (Switch), and J (Jumper).
represents the capacitance value associated with the pin.
is optional. It represents the location of the pin. Multiple pin
locations are allowed
split pins
1 2 3 4 5
*|P (avss_1 O 0 207.7555 59.9170) *|P (avss_10 O 0 181.1610 151.1130) *|P (avss_11 O 0 186.6330 151.1130) *|P (avss_12 O 0 192.1050 151.1130) *|P (avss_13 O 0 197.5770 151.1130)
* DSPF files to use with Corner Definitions * This is an example file showing how to define different dspf files for different corners * using model files for individual components as the * building blocks. simulator lang=spectre library dspf_files_corners
In addition to lowering the required capacitor value, Miller
compensation entails a very important property: it moves the output pole
away from the origin. This effect is called pole
splitting
The 1st stage is replaced with Thevenin equivalent circuit , \(V_i \cong V_i \cdot g_{m1}R_{o1}\)
\[\begin{align}
\frac{V_i-V_{o1}}{R_{o1}} &= V_{o1}\cdot sC_{o1}+(V_{o1}-V_o)\cdot
sC_c \\
V_{o1} &= \frac{V_i+sR_{o1}C_cV_o}{1+sR_{o1}(C_{o1}+C_c)}
\end{align}\]\[
(V_{o1}-V_o)sC_c=g_{m2}V_{o1}+V_o(\frac{1}{R_{o2}+sC_L})
\] substitute \(V_{o1}\), we
get
\(s^3\) terms in denominator \[
H_3 = s^3\cdot(R_{o1}R_{o2}R_c+R_{o1}R_{o2}R_{sw} +R_{o1}R_cR_{sw})\cdot
C_{o1}C_cC_L
\]\(s^2\) terms in denominator
\[\begin{align}
H_2
&=s^2\cdot(R_{o1}R_{o2}C_{o1}C_c+R_{o1}R_{o2}C_{o1}C_L+R_{o2}R_cC_cC_L+R_{o1}R_{o2}C_cC_L+R_{o1}R_cC_{o1}C_c\\
&+R_{o2}R_{sw}C_cC_L+R_{o1}R_{sw}C_cC_L\cdot
g_{m2}R_{o2}+R_{o1}R_{sw}C_{o1}C_L+R_{sw}R_cC_cC_L+R_{o1}R_{sw}C_cC_L)
\end{align}\]
\(s^1\) term in denominator \[
H_1=s(R_{o1}\cdot
g_{m2}R_{o2}C_c+R_{o1}C_{o1}+R_cC_c+R_{o1}C_c+R_{o2}C_c+R_{o2}C_L+R_{sw}C_L)
\]\(s^0\) term in denominator
\[
H_0=1
\] set \(R_c=0\) and \(R_{sw}=0\), the \(H_*\) reduced to \[\begin{align}
H_3 &= 0 \\
H_2 &=s^2R_{o1}R_{o2}(C_{o1}C_c+C_{o1}C_L+C_cC_L) \\
H_1&=s(R_{o1}\cdot
g_{m2}R_{o2}C_c+R_{o1}C_{o1}+R_{o1}C_c+R_{o2}C_c+R_{o2}C_L) \\
H_0&=1
\end{align}\] That is \[
H=s^2R_{o1}R_{o2}(C_{o1}C_c+C_{o1}C_L+C_cC_L)+s(R_{o1}\cdot
g_{m2}R_{o2}C_c+R_{o1}C_{o1}+R_{o1}C_c+R_{o2}C_c+R_{o2}C_L)+1
\]
which is same with our previous analysis of Generic circuit in
textbook
And we know \[
\frac{V_o}{V_{o2}}=\frac{1}{1+sR_{sw}C_L}
\] Finally, we get \(\frac{V_o}{V_i}\)\[\begin{align}
\frac{V_o}{V_i} &= \frac{V_{o2}}{V_i} \cdot \frac{V_o}{V_{o2}} \\
&= -g_{m2}R_{o2}\frac{\left[ sC_c(R_c-1/g_{m2})+1
\right](sR_{sw}C_L+1)}{H_3+H_2+H_1+1} \cdot \frac{1}{1+sR_{sw}C_L} \\
&= -g_{m2}R_{o2}\frac{ sC_c(R_c-1/g_{m2})+1}{H_3+H_2+H_1+1}
\end{align}\]
The loop transfer function is \[
\frac{V_o}{V_i} =-g_{m1}R_{o1}g_{m2}R_{o2}\frac{
sC_c(R_c-1/g_{m2})+1}{H_3+H_2+H_1+1}
\]
The poles can be deduced \[\begin{align}
\omega_1 &= \frac{1}{R_{o1}\cdot g_{m2}R_{o2}C_c} \\
\omega_2 &= \frac{1}{1+g_{m2}R_{sw}}\cdot \frac{g_{m2}}{C_L} \\
&= \frac{1}{(gm_2^{-1}+R_{sw})C_L}
\end{align}\]
The pole \(\omega_2=\frac{1}{gm_2^{-1}C_L}\) is
changed to \(\omega_2=\frac{1}{(gm_2^{-1}+R_{sw})C_L}\)
In order to cancell \(\omega_2\)
with \(\omega_z\), \(R_c\) shall be increased
\[
R_{eq}=g_{m2}^{-1}+R_{sw}
\]
omit \(C_{o1}\) is
same with 2nd system simplification
non-dominant pole in Sansen's
book
Following demonstrate how derive \(f_{nd}\) from Razavi's equation. We copy
\(\omega_2\) here \[
\omega_2 = \frac{R_{o1}C_c\cdot
g_{m2}R_{o2}+R_{o2}(C_c+C_L)+R_{o1}(C_{o1}+C_c)}{R_{o1}R_{o2}(C_cC_{o1}+C_LC_{o1}+C_LC_c)}
\] which can be reduced as below
In short-channel devices, as \(V_{DS}\) increases further, drain-induced
barrier lowering becomes significant, reducing the threshold voltage and
increasing the drain current
Gate induced drain leakage
(GIDL)
Chauhan, Yogesh Singh, et al. FinFET modeling for IC simulation
and design: using the BSIM-CMG standard. Academic Press, 2015.
\[
\frac{g_m}{I_D} = \frac{2}{V_{GS}-V_{TH}}
\] Decrease of gm/Id results from decrease in VT.
GIDL (Gate induced drain leakage) as at weak
inversion may results in a weak lateral electric field causing leakage
current between drain and bulk, which degrade the efficiency of the
transistor (gm/ID).
Resistive degeneration in differential pairs serves as one
major technique for linear amplifier
The linear region for CMOS differential pair would be extended by
\(±I_{SS}R/2\) as all of \(I_{SS}/2\) flows through \(R\). \[\begin{align}
V_{in}^+ -V_{in}^- &= V_{OV} + V_{TH}+\frac{I_{SS}}{2}R - V_{TH} \\
&= \sqrt{\frac{2I_{SS}}{\mu_nC_{OX}\frac{W}{L}}} + \frac{I_{SS}R}{2}
\end{align}\]
The dependence of offset voltage and current mismatches upon the
overdrive voltage is similar to our observations for corresponding
noise quantities
differential pair
In reality, since mismatches are independent statistical
variables
Above shows that the input transistors must be designed for high
gain (\(g_mr_o =
\frac{2}{V_{OV}\lambda}\)), which means they must be designed for
small\(V_{GS}-V_{TH}\).
It is desirable to minimize \(V_{GS}-V_{TH}\) by lowering the tail
current or increasing the transistor widths
Then, we get \[
V_{os}=\frac{V_o'-V_o}{A}+(V_m'-V_m)
\] Due to \(V_o=V_m\) and \(V_o'=V_m'\)\[
V_{os}=(1/A+1)\Delta{V_m}
\] or \[
V_{os}=(1/A+1)\Delta{V_o}
\] if \(A \gg 1\)\[
V_{os}=\Delta{V_o}
\]
we get \[
V_{os}=\frac{V_o'-V_o}{A}+(V_m'-V_m)
\] or \[
V_{os}=\frac{\Delta V_o}{A}+\beta \Delta V_o
\] if \(A \gg 1\)\[
V_{os}=\beta \Delta V_o
\] or \[
V_{os}=\Delta V_m
\]
Lecture 22 Variability and Mismatch of Dr. Hesham A. Omran's
Analog IC Design
There is discrepancy between model operating point and \(V_{ds}/I_{ds}\)
I believe that the equation \(V_{ds}/I_{ds}\) is more appropriate where
mos is used as switch, though \(V_{ds}=0\) is an outlier.
Schmitt Inverter
gm/ID Intuition
small gm/ID for High ro, or high Early voltage \(V_A\)
Transit Frequency \(f_T\)
Defined as the frequency at which the small-signal current
gain of a device is unity
MOSFET ZTC Condition
Analysis
zero temperature coefficient (ZTC)
MOM cap of wo_mx
Monte Carlo model:
\(C_{pa}=C_{pa1}\), \(C_{pb}=C_{pb1}\) for each iteration during
Process Variation
different variation is applied to \(C_{ab}\) and \(C_{a1b1}\) each iteration during
Mismatch Variation, though \(C_{pa}\), \(C_{pb}\), \(C_{pa1}\) and \(C_{pb1}\) remain constant
\(C_\text{eq}\) and \(R_\text{eq}\) are obtained \[\begin{align}
C_\text{eq} &= \frac{1+|A|^2-2A_r}{1-A_r}\cdot C_f \\
R_\text{eq} &= \frac{A_i}{1+|A|^2-2A_r}\cdot \frac{1}{\omega C_f}
\end{align}\]
D/S small signal model
The Drain and Source of MOS are determined
in DC operating point, i.e. large signal.
That is, top of \(M_2\) is
drain and bottom is source, \[\begin{align}
R_\text{eq2} &= \frac{r_\text{o2}+R_L}{1+g_\text{m2}r_\text{o2}} \\
& \simeq \frac{1}{g_\text{m2}}
\end{align}\]
PMOS small signal model
polarity
The small-signal models of NMOS and PMOS transistors are
identical
A negative \(\Delta V_\text{GS}\)
leads to a negative \(\Delta I_D\).
Recall that \(I_D\), in the
direction shown here, is negative because the actual current of holes
flows from the source to the drain.
Conversely, a positive \(\Delta
V_\text{GS}\) produces a positive \(\Delta I_D\), as is the case for an NMOS
device.
Leakage in MOS
reference
W. M. Elgharbawy and M. A. Bayoumi, "Leakage sources and possible
solutions in nanometer CMOS technologies," in IEEE Circuits and Systems
Magazine, vol. 5, no. 4, pp. 6-17, Fourth Quarter 2005, doi:
10.1109/MCAS.2005.1550165.
X. Qi et al., "Efficient subthreshold leakage current optimization -
Leakage current optimization and layout migration for 90- and 65- nm
ASIC libraries," in IEEE Circuits and Devices Magazine, vol. 22, no. 5,
pp. 39-47, Sept.-Oct. 2006, doi: 10.1109/MCD.2006.272999.
P. Monsurró, S. Pennisi, G. Scotti and A. Trifiletti, "Exploiting the
Body of MOS Devices for High Performance Analog Design," in IEEE
Circuits and Systems Magazine, vol. 11, no. 4, pp. 8-23, Fourthquarter
2011, doi: 10.1109/MCAS.2011.942751.
REF: D. A. Yokoyama-Martin et al., "A Multi-Standard Low Power
1.5-3.125 Gb/s Serial Transceiver in 90nm CMOS," IEEE Custom Integrated
Circuits Conference 2006, 2006, pp. 401-404, doi:
10.1109/CICC.2006.320970.
Power/Ground and I/O Pins
Power / Ground Pin
Information
In both digital and analog I/O, power and ground pins appear at the
sub-circuit definiton, allowing user to use the I/O in voltage islands.
They follow certain naming conventions.
digital I/O sub-circuit
VDD: pre-driver core voltage (supplied by PVDD1CDGM)
VSS: pre-driver ground and also global ground (supplied by
PVDD1CDGM)
VDDPST: I/O post-driver voltage, i.e. 1.8V (supplied by PVDD2CDGM or
PVDD2POCM)
VSSPOST: I/O post-driver ground (supplied by PVDD2CDGM or
PVDD2POCM)
POCCTRL: POCCTRL signal (supplied by PVDD2POCM)
analog I/O placed in a core voltage domain, the convention is
TACVDD: analog core voltage (supplied by PVDD3ACM)
TACVSS: analog core ground (supplied by PVDD3ACM)
VSS: global core ground
analog I/O placed in an I/O voltage domain, the convention is:
TAVDD: analog I/O voltage, i.e. 1.8V (supplied by PVDD3AM)
TAVSS: analog I/O ground (supplied by PVDD3AM)
VSS: global core ground
Power/Ground Combo Cells
power/ground combo pad cell
pins to be connected to bump
to core side pin name
PVDD1CDGM
VDD VSS
VDD VSS
PVDD2CDGM PVDD2POCM
VDDPST VSSPST
N/A
PVDD3AM
TAVDD TAVSS
AVDD AVSS
PVDD3ACM
TACVDD TACVSS
AVDD AVSS
Note for the retention mode
At initial state, IRTE must be 0 when VDD is
off.
IRTE must be kept >= 10us after VDD turns on again (from the
retention mode to the normal operation mode).
IRTE can be switched only when both VDD and VDDPST are on.
When the rention function is needed, IRTE signal must come from an
"always-on" core power domain. If you don't need the rention function,
it is required to tie IRTE to ground. In other words, no matter
the rention feature is needed or not, it is required to have PCBRTE in
each domain.
Note: PCBRTE does not need PAD
connection.
Internal Pins
There are 3 internal global pins, i.e. ESD,
POCCTRL, RTE, in all digital domain
cells.
In real application,
ESD pin is an internal signal and
active in ESD event happening
POCCTRL is an internal signal and active in
Power-on-control event.
However, these special events (i.e. ESD event and Power-on-control
event) are not modeled in NLDM kit (.lib), only normal function is
covered, so ESD and POCCTRL pins are
simply defined as ground in NLDM kit (.lib).
These 3 global pins will be connected automatically after
cell-to-cell abutting in physical layout.
Power-Up sequence in
Digital Domain
Power up the I/O power (VDDPST) first, then the core power (VDD)
PVDDD2POCM cell would generate Power-On-Control signal (POCCTRL) to
have the post-driver NMOS and PMOS off, so that the crowbar current
would not occur in the post-driver fingers when the I/O voltage is on
while the core voltage remains off. As such, I/O cell would be in the
Hi-Z state. when POCCTRL is on, the pll-up/down resistor is disabled and
C is 0.
The POCCTRL signal is transmitted to I/O cells through cell
abutment. There is no need to have routing for POCCTTRL nor
give a control signal to the POCCTRL pin any of I/O cells. Note that the
POCCTRL signal would be cut if inserting a power-cut (PRCUT) cell.
Power-Down sequence in
Digital Domain
It's the reverse of power-up sequence.
Use model in Innovus
1 2 3 4 5 6 7 8 9 10 11 12
set init_gnd_net "vss_core vss DUMMY_ESD DUMMY_POCCTRL"
The antenna effect is a common name for the effects
of charge accumulation in isolated nodes of an
integrated circuit during its processing.
This effect is also sometimes called "Plasma Induced Damage",
"Process Induced Damage" (PID) or "charging effect".
antenna ratio
The antenna rule specifies the maximum tolerance for the ratio of a
metal line area to the area of connected gates.
metal jumping
Long metal can be taken to higher metal routing layer. This
is known as metal jumping.
This metal jumping will break the long interconnect and hence the
charge collected on the long interconnect will not discharge through
gate oxide because the higher metal layer is not yet fabricated.
so, if the gate immediately connects to the highest level by jump-up
metals, large amount of charges can not be collected, while the poly
finally connected to the diffusion part by highest level, thus no
antenna violation will normally occure.
Diode Insertion
Diode helps dissipate charges accumulated on metal. Diode should be
placed as near as possible to the gate of device on low level of
metal.
Diode should always be connected in reverse bias, with
cathode connected to gate electrode and anode connected to ground
potential.
During processing, even if the diodes are reversely biased, because
of the elevated wafer temperature (200 o C plus) it will provide a much
conductive path
In the reverse bias region, the reverse saturation current of Si and
Ge diodes doubles for every 10° C rise in temperature
Tuvia Liran, Antenna effect (PID): Do the design rules really protect
us? [link]
When a flip-flop samples an input that is changing during its
aperture, the output Q may momentarily take on a voltage between 0 and
VDD that is in the forbidden zone. This is called a metastable state.
Eventually, the flip-flop will resolve the output to a stable state of
either 0 or 1. However, the resolution time required to reach the stable
state is unbounded
Kinniment, D. J. Synchronization and arbitration in digital systems.
John Wiley & Sons Ltd (2007).
Slewing
In practice, we choose \(I_P \simeq
I_{SS}\)
Avoid zero current in cascodes
left circuit
\(I_b \gt I_a\)
right circuit
\(I_b \gt 2I_a\)
Step Response of higher
order system
Since \(1/sC_1+R_1 \gg R_0\)\[
\frac{V_m}{V_i}(s) \simeq \frac{R_0}{R_0 + 1/sC_0} =
\frac{sR_0C_0}{1+sR_0C_0}
\]step response of \(V_m\)\[
V_m(t) = e^{-t/R_0C_0}
\] where \(\tau = R_0C_0\)
And \(V_o(s)\) can be expressed as
\[\begin{align}
\frac{V_o}{V_i}(s) & \simeq \frac{sR_0C_0}{1+sR_0C_0} \cdot
\frac{sR_1C_1}{1+sR_1C_1} \\
&= \frac{sR_0C_0R_1C_1}{R_0C_0-R_1C_1}\left(\frac{1}{1+sR_1C_1} -
\frac{1}{1+sR_0C_0}\right)
\end{align}\]
M. Tian, V. Visvanathan, J. Hantgan and K. Kundert, "Striving for
small-signal stability," in IEEE Circuits and Devices Magazine, vol. 17,
no. 1, pp. 31-41, Jan. 2001, doi: 10.1109/101.900125.
Ali Sheikholeslami, Circuit Intuitions: Thevenin and Norton
Equivalent Circuits, Part 3 IEEE Solid-State Circuits Magazine, Vol. 10,
Issue 4, pp. 7-8, Fall 2018.
Ali Sheikholeslami, Circuit Intuitions: Thevenin and Norton
Equivalent Circuits, Part 2 IEEE Solid-State Circuits Magazine, Vol. 10,
Issue 3, pp. 7-8, Summer 2018.
Ali Sheikholeslami, Circuit Intuitions: Thevenin and Norton
Equivalent Circuits, Part 1 IEEE Solid-State Circuits Magazine, Vol. 10,
Issue 2, pp. 7-8, Spring 2018.
Ali Sheikholeslami, Circuit Intuitions: Miller's Approximation IEEE
Solid-State Circuits Magazine, Vol. 7, Issue 4, pp. 7-8, Fall 2015.
Ali Sheikholeslami, Circuit Intuitions: Miller's Theorem IEEE
Solid-State Circuits Magazine, Vol. 7, Issue 3, pp. 8-10, Summer
2015.
Shanthi Pavan, "Demystifying Linear Time Varying Circuits"
Mismatch between the pole and zero frequencies leads to the
“doublet problem”. If the pole and the zero do not
exactly coincide, we say that they constitute a
doublet
Problem 10.19 in Razavi 2nd book
Suppose the open-loop transfer function of a two-stage op amp is
expressed as \[
H_{open}(s)=\frac{A_0(1+\frac{s}{\omega_z})}{\left( 1+
\frac{s}{\omega_{p1}}\right)\left( 1+ \frac{s}{\omega_{p2}}\right)}
\] Ideally, \(\omega_z=\omega_2\) and the feedback
circuit exhibits a first-order behavior, i.e., its step response
contains a single time constant and no overshoot.
Then the transfer function of the amplifier in a unity-gain
feedback loop is given by \[\begin{align}
H_{closed}(s)
&=\frac{A_0\left(1+\frac{s}{\omega_z}\right)}{\frac{s^2}{\omega_{p1}\omega_{p2}}+\left(
\frac{1}{\omega_{p1}} +
\frac{1}{\omega_{p2}}+\frac{A_0}{\omega_{z}}\right)s+A_0+1} \\
&=\frac{\frac{A_0}{A_0+1}(1+\frac{s}{\omega_z})}{\frac{s^2}{\omega_{p1}\omega_{p2}(A_0+1)}+\left(
\frac{1}{\omega_{p1}} +
\frac{1}{\omega_{p2}}+\frac{A_0}{\omega_{z}}\right)\frac{s}{A_0+1}+1}
\end{align}\]
The denominator part of \(H_{closed}(s)\) is \[
D(s) = \frac{s^2}{\omega_{p1}\omega_{p2}}+\left( \frac{1}{\omega_{p1}} +
\frac{1}{\omega_{p2}}+\frac{A_0}{\omega_{z}}\right)s+A_0+1
\]
Assuming two poles (\(\omega_{pA}
\ll\omega_{pB}\)) of \(H_{closed}(s)\) are widely spaced, \[\begin{align}
D(s) &= \left( 1+ \frac{s}{\omega_{pA}}\right)\left( 1+
\frac{s}{\omega_{pB}}\right)\\
&\cong \frac{s^2}{\omega_{pA}\omega_{pB}}+\frac{s}{\omega_{pA}} + 1
\end{align}\]
Thus, the two poles of the closed-loop transfer function of system
are \[\begin{align}
\omega_{pA} &= \frac{A_0+1}{\frac{1}{\omega_{p1}} +
\frac{1}{\omega_{p2}}+\frac{A_0}{\omega_{z}}} \\
&= \frac{(A_0+1)\omega_{p1} \omega_{p2}}{\omega_{p1} + \omega_{p2}
+ \frac{A_0}{\omega_z}\omega_{p1} \omega_{p2}} \\
\omega_{pB} &= \omega_{p1} + \omega_{p2} +
\frac{A_0}{\omega_z}\omega_{p1} \omega_{p2}
\end{align}\]
Assuming\(\omega_z \simeq
\omega_{p2}\) and \(\omega_{p2}\ll
(1+A_0)\omega_{p1}\)\[
\omega_{pA} = \omega_{p2}
\] and \[
\omega_{pB} = (1+A_0)\omega_{p1}
\] The closed-loop transfer function is \[
H_{closed}(s) =
\frac{\frac{A_0}{A_0+1}\left(1+\frac{s}{\omega_z}\right)}{\left(1+\frac{s}{(1+A_0)\omega_{p1}}\right)\left(
1+\frac{s}{\omega_{p2}} \right)}
\]
The step response of the closed-loop amplifier
Consider the Laplace transform function of step response, \(X(s)=\frac{1}{s}\)\[
Y(s)=\frac{1}{s}\times H_{closed}(s)
\] Thus, the small-signal step response of the
closed-loop amplifier is \[
y(t)=\frac{A_0}{A_0+1}\left[1-e^{-(A_0+1)\omega_{p1}t}-\left(1-\frac{\omega_{p2}}{\omega_z}\right)e^{-\omega_{p2}t}
\right]u(t)
\] Since, \(\omega_{p2}\ll
(1+A_0)\omega_{p1}\). Therefore, rewrite the \(y(t)\)\[
y(t)\cong
\frac{A_0}{A_0+1}\left[1-\left(1-\frac{\omega_{p2}}{\omega_z}\right)e^{-\omega_{p2}t}
\right]u(t)
\] The step response contains an exponential term of the form
\(\left(1-\frac{\omega_{p2}}{\omega_z}\right)e^{-\omega_{p2}t}\).
This is an important result, indicating that if the zero does not
exactly cancel the pole, the step response exhibits an exponential with
an amplitude proportional to \(\left(1-\frac{\omega_{p2}}{\omega_z}\right)\),
which depends on the mismatch between \(\omega_z\) and \(\omega_{p2}\) and a time constant \(\tau\) of \(\frac{1}{\omega_{p2}}\) or \(\frac{1}{\omega_{z}}\)
B. Y. T. Kamath, R. G. Meyer and P. R. Gray, "Relationship between
frequency response and settling time of operational amplifiers," in IEEE
Journal of Solid-State Circuits, vol. 9, no. 6, pp. 347-352, Dec. 1974,
doi: 10.1109/JSSC.1974.1050527.
B. Y. T. Kamath, R. G. Meyer and P. R. Gray, "Relationship between
frequency response and settling time of operational amplifiers," in IEEE
Journal of Solid-State Circuits, vol. 9, no. 6, pp. 347-352, Dec. 1974,
doi: 10.1109/JSSC.1974.1050527.
P. R. Gray and R. G. Meyer, "MOS operational amplifier design-a
tutorial overview," in IEEE Journal of Solid-State Circuits, vol. 17,
no. 6, pp. 969-982, Dec. 1982, doi: 10.1109/JSSC.1982.1051851.