You are on page 1of 24

University of Padova

Information Engineering Dept. - Microelectronics lab

Circuiti Integrati per lElaborazione dei Segnali

Simulazione di sistemi tempo varianti


Andrea Gerosa
University of Padova
Information Engineering Dept. - Microelectronics lab

Simulation of time-varying Networks


Basic concepts
single out a periodic steady-state (PSS) solution
linearize the circuit around the PSS point
time-varying equivalent small signal circuit
on the basis of PSS analysis frequency-domain response
can be evaluated
account for frequency translations

Andrea Gerosa C.I.E.S.: Spectre RF 2


University of Padova
Information Engineering Dept. - Microelectronics lab

Periodic steady-state (PSS) analysis


Assumptions
all stimuli are periodic (e.g. local oscillator, clock, etc.)
the analyzed circuit has a periodic response
analysis cannot be applied to circuits with a non-
periodic response to periodic stimuli (e.g.
modulators)
stimulus can be large
circuit does not have to respond linearly to the stimuli

Andrea Gerosa C.I.E.S.: Spectre RF 3


University of Padova
Information Engineering Dept. - Microelectronics lab

Periodic steady-state (PSS) analysis


How the analysis is performed
the circuit is simulated in the time domain
all non-idealities and non-linearities typical of time-
domain simulations are accounted for
the circuit is assumed to be periodic of T seconds (user-
defined)
the simulator looks for an initial condition which leads
to a periodic steady state solution of the network

Andrea Gerosa C.I.E.S.: Spectre RF 4


University of Padova
Information Engineering Dept. - Microelectronics lab

Looking for the periodic solution


Estimate an initial condition
for the circuit Shooting
Time domain simulation for method
T seconds
NO
Initial and final con-
ditions are identical?
YES
The simulated behaviuor
is the PSS solution
Andrea Gerosa C.I.E.S.: Spectre RF 5
University of Padova
Information Engineering Dept. - Microelectronics lab

Periodic small signal analysis


evaluate the circuit response to a large periodic stimulus
single out the PSS solution
evaluate the circuit response to a small signal, in the
presence of the large periodic stimulus
the circuit is linearized around the PSS solution
the circuit is assumed to respond linearly to the small
signal

Andrea Gerosa C.I.E.S.: Spectre RF 6


University of Padova
Information Engineering Dept. - Microelectronics lab

Periodic small signal analysis


the PSS solution is singled out
for each time instant an equivalent small-signal circuit is
calculated
effect of bias point variations during time is taken into
account
the equivalent circuit is simulated
using Fourier integral based transformation (not DFT) the
frequency-domain response is evaluated

Andrea Gerosa C.I.E.S.: Spectre RF 7


University of Padova
Information Engineering Dept. - Microelectronics lab

Classical vs PSS small signal analysis


Classical PSS
DC operating point is PSS solution is evaluated
evaluated circuit is linerized around the
circuit is linearized around the time-variant periodic solution
time-invariant operating point circuit is solved in the time
circuit is solved using domain, then applying
Laplace-domain analysis integral Fourier analysis

Andrea Gerosa C.I.E.S.: Spectre RF 8


University of Padova
Information Engineering Dept. - Microelectronics lab

Frequency translation
a periodically time-varying
circuit responds with
sinusoid at several
frequencies fund
the PAC evaluates the
transfer function for
each frequency
components of the
output spectrum k=0 k=1

f k
out = f in + k fund pss
Andrea Gerosa C.I.E.S.: Spectre RF 9
University of Padova
Information Engineering Dept. - Microelectronics lab

SC biquad
filter
The circuit

Andrea Gerosa C.I.E.S.: Spectre RF 10


University of Padova
Information Engineering Dept. - Microelectronics lab

Example: frequency response


of the SC biquad
PSS analysis set-up
need to define the fundamental
periodicity of the circuit
initial guess for the PSS
solution
how many harmonics do we want
to account for at the system output
allow some time to stabilize

Andrea Gerosa C.I.E.S.: Spectre RF 11


University of Padova
Information Engineering Dept. - Microelectronics lab

Example: frequency response of the


SC biquad
PAC analysis set-up
define frequency
sweep
define number of
sidebands to be
considered at the
system output

Andrea Gerosa C.I.E.S.: Spectre RF 12


University of Padova
Information Engineering Dept. - Microelectronics lab

Frequency response of the SC biquad

Andrea Gerosa C.I.E.S.: Spectre RF 13


University of Padova
Information Engineering Dept. - Microelectronics lab

Distortion simulation for


the CMOS mixer
Setting up PSS for distortion
LO @ 50MHz and Vin @ 40MHz
choose common periodicity
force to show the output
harmonics of interest

Andrea Gerosa C.I.E.S.: Spectre RF 14


University of Padova
Information Engineering Dept. - Microelectronics lab

Distortion results for


the CMOS mixer
Output spectrum
harmonics due to non-
linear response to the
large tone
2nd order
3rd order

Andrea Gerosa C.I.E.S.: Spectre RF 15


University of Padova
Information Engineering Dept. - Microelectronics lab

Gain compression
3rd order distortion produces an output
components which subtracts from the output
fundamentals
1st order gain is reduced
this effect is measured by the 1-dB compression
point
input value for which the gain is 1-dB lower
than the ideal behaviour
to quantify gain compression large signal gain
can be evaluate for different input amplitudes
sweep PSS analysis
Andrea Gerosa C.I.E.S.: Spectre RF 16
University of Padova
Information Engineering Dept. - Microelectronics lab

Swept PSS setup


Repetition of the PSS
analysis
define the fundamental
parameters of PSS

the amplitude of the input


tone varies at each
simulation
define sweep range and
type
Andrea Gerosa C.I.E.S.: Spectre RF 17
University of Padova
Information Engineering Dept. - Microelectronics lab

Swept PSS results

Results browser
choose output parameters to
plot
main figures of merit
automatically computed

Andrea Gerosa C.I.E.S.: Spectre RF 18


University of Padova
Information Engineering Dept. - Microelectronics lab

First-order gain of the mixer

1-db compression point


1st order harmonic is
plotted versus input signal
amplitude (swept
parameter)
ideal linear characteristic
is plotted too
1-dB compression point is
calculated

Andrea Gerosa C.I.E.S.: Spectre RF 19


University of Padova
Information Engineering Dept. - Microelectronics lab

First-order gain of the SC


biquad

same procedure previously


discussed is applied to the SC
biquad filter

Andrea Gerosa C.I.E.S.: Spectre RF 20


University of Padova
Information Engineering Dept. - Microelectronics lab

Third order distortion component


for increasing input amplitude the 3rd order distortion
component grows faster than the fundamental
3rd order components intercepts the fundamental
component
IP3 point is usually computed

Andrea Gerosa C.I.E.S.: Spectre RF 21


University of Padova
Information Engineering Dept. - Microelectronics lab

IP3 point simulation

perform swept-PSS
all parameters for the IP3 analysis
are automatically computed
select frequencies for 1st and
3rd harmonic component

Andrea Gerosa C.I.E.S.: Spectre RF 22


University of Padova
Information Engineering Dept. - Microelectronics lab

IP3 point for the


mixer
1st and 3rs harmonics
are both plotted as a
function of input
amplitude
the ideal characteristic
is added
the IP3 point is
extrapolated from the
linear behaviour

Andrea Gerosa C.I.E.S.: Spectre RF 23


University of Padova
Information Engineering Dept. - Microelectronics lab

IP3 point for the SC biquad

same procedure
previously discussed is
applied to the SC biquad
filter

Andrea Gerosa C.I.E.S.: Spectre RF 24

You might also like