Académique Documents
Professionnel Documents
Culture Documents
DESIGN
MIXDES 2010, 17th International Conference "Mixed Design of Integrated Circuits and Systems", June 24-26, 2010, Wrocaw, Poland
Achim Graupner
LOP: G
ID
VGS,0
I. I NTRODUCTION
Automation of analog circuit design gets more and more
important to increase design efciency and to shorten time for
migration of analog circuit blocks from one semiconductor
technology to another one. Besides the desired performance
specications, the fulllment of constraints (e.g. saturation
of transistors) is important to ensure functionality, yield and
robustness of integrated circuits [1]. Optimization based sizing
techniques have been reported, that take into account constraints of devices or device groups [2]. These methods are
based on spice-level circuit simulations or macromodels of
the circuit components [3]. Furthermore, there exist rule based
design migration strategies, which allow resizing of analog
circuits to a new technology node [4], but require additional
checking of the feasibility of a given circuit topology with
respect to reduced supply voltages, which plays an important
role in modern deep submicron CMOS technologies.
In [5] and [6] a fast evaluation method of analog circuit
feasibility (i.e. the fulllment of constraints) by means of
linear programming is presented. The constraints are linearized
around an initial operating point by calculation of constraint
sensitivities with respect to circuit parameters (e.g. device
sizes) and operating parameters (e.g. supply voltage). For
this purpose multiple circuit simulations are required, where
the number of simulations increases with the number of
parameters.
In contrast to that, this work presents a fast voltage range
analysis method which requires a single DC operating point
simulation only. It employs a linearized operating point (LOP)
G
VGS
B
S
VSB
VDS
ID
kGS VSB
B
VSB
nullator
(1)
(if S=B)
VG VGS,0
VG = VS (1 + kGS ) + VGS,0 (ID ) kGS VB
(3)
j38
TABLE I
E XAMPLES OF MOS
technology
supply VDD
TRANSISTOR
LOP
MODEL ACCURACY VS .
0.600m
5.0V
0.60
0.350m
3.3V
0.35
0.180m
1.8V
0.18
0.065m
1.0V
0.07
[1, 100]
[100, 1]
[1, 100]
[100, 1]
[1, 200]
[200, 1]
[1, 200]
[200, 1]
nmos
pmos
nmos
pmos
nmos
pmos
nmos
pmos
(4)
(5)
(6)
VGS [V]
1.20
1.00
0.80
simulation
linear model
Vsat [V]
0.60
0.15
0.14
0.5
1.0
1.5
VSB [V]
2.0
2.5
3.0
j3e
[0.0, 2.5]
[2.5, 0.0]
[0.0, 1.15]
[1.15, 0.0]
[0.0, 0.9]
[0.9, 0.0]
[0.0, 0.5]
[0.5, 0.0]
[0.5, 0.5]
[0.5, 0.5]
[0.5, 0.5]
[0.5, 0.5]
[0.3, 0.5]
[0.5, 0.3]
[0.3, 0.5]
[0.5, 0.3]
51
21
34
16
11
14
24
24
(1.0%
(0.4%
(1.0%
(0.5%
(0.6%
(0.8%
(2.4%
(2.4%
VDD )
VDD )
VDD )
VDD )
VDD )
VDD )
VDD )
VDD )
simulation
linear model
0.13
0.12
0.0
(7)
(8)
save
ID
D
G
VGS
B
VDS
W
L S
VGS,0
kGS
Vsat,0
ksat
t VGS
Vsat
Fig. 3.
VDG
ID
D
B
G W
L
VSB
VSB
save
DCOP simulation
XML
XML
LUT data
Classify nodes
Net dependencies
1
Optimization goal
Solve LP
Fig. 4.
where MA , MB and MD are n n matrices that include the
dependency coefcients from Eq. 2 and Eq. 3 respectively and
v = [v1 , . . . , vn ]T are (n1) node voltage vectors. Dependent
nodes depend on other dependent nodes if MD = 0. If all
dependencies are evaluated, i.e. there are no more unevaluated
devices, but there are uncategorized nodes remaining, one of
them is dened as free node. The according entries in matrix
MD are moved to MB . Then the node propagation algorithm
is started again.
Constraints
Setup LP
Results
conguration
Iteration loop
4
5
6
Analysis ow chart
7
8
9
(9)
10
11
12
13
14
15
When the numbers of xed (nxed ), free (nfree ) and dependent (ndep ) nodes have been determined the matrix dimensions
can be reduced. We introduce the matrices MA (ndep nxed ),
MB (ndep nfree ), MD (ndep ndep ) and the vector v0 (ndep 1)
and calculate their values by removing those entries from the
original matrices MA to MD respectively, which correspond
to nodes that are not of the according category. The next step
is to resolve the dependencies between dependent nodes by
solving Eq. 10 with the new matrix denitions for vdep .
vdep = (MD E)1 (MA vxed
MB vfree v0 )
(11)
j3d
TABLE II
N ODE PROPAGATION DECISION TABLE ( F : NODE FIXED , NF : NODE NOT
FIXED , C : NODE CATEGORY KNOWN , NC : NODE CATEGORY NOT KNOWN )
G
f
f
f
f
nf
nf
action performed
c
nc
c
nc
(12)
Q = (MD E)1 MB
(13)
r = (MD E)
v0
(14)
Q
r
vdep
P
vfree = 0 vxed + E vfree + 0
(15)
E
0
0
vxed
v
r
P
Q
Each node voltage is a linear combination of free and xed
node voltages.
4
1 (VDD )
0 (Vgnd )
M1
M0
5 (Vout )
(17)
(18)
M2
(19)
0
0
M5
Fig. 6.
3 (Vin )
0
M3
xed node
dependent node
j33
(16)
2 (Vin )
M4
free node
TABLE III
S IMPLE OTA
1
2
3
4
LOP
VS .
S PECTRE SIMULATION
Conditions
Objective
LOP model
Spectre
Error
VDD,min
Vin,min
VDD,min
Vin,max
1.73V
0.86V
1.28V
2.32V
1.69V
0.84V
1.24V
2.33V
40mV
20mV
40mV
10mV
No. of Simulations
300
300 300 300
300 300
300 300
(2.3%)
(2.4%)
(3.2%)
(0.4%)
case numbers for an accuracy of 10mV in a range from 0V to 3V for linear search of all free dimensions
4.0
3.5
M0
3.0
Vin [V]
a Worst
2.5
DCOP
2.0
1.5
M2
1.0
M4
0.5
0.0
1.0
1.5
2.0
2.5
VDD [V]
3.0
3.5
4.0
TABLE IV
3.5
3.0
type
VGS,0 [V]
kGS
Vsat,0 [V]
ksat
M0
M1
M2
M3
M4
M5
pmos
pmos
nmos
nmos
nmos
nmos
-1.04
-1.04
0.69
0.69
0.76
0.76
-0.10
-0.10
0.22
0.22
0.22
0.22
-0.18
-0.18
0.10
0.10
0.14
0.14
0.005
0.005
0.003
0.003
0.004
0.004
2.5
Vin [V]
device
M2
2.0
1.5
M3
1.0
M4
M0
0.5
0.0
0.0
0.5
1.0
1.5
2.0
2.5
ID [uA]
Vout [V]
25.0
20.0
15.0
10.0
5.0
0.0
1.0
Fig. 7.
LIN
ID(M4)
SAT
2.0
2.5
VDD [V]
3.0
DCOP simulation
3.5
4.0
j3N
TABLE V
O P - AMP LOP ANALYSIS RESULTS
1
2
3
4
5
6
7
VDD
Conditions
Objective
Result
VDD
VDD
VDD
VDD
VDD
VDD
VDD,min
VOUT,min
VOUT,max
VIP,min
VIP,max
VIP,max
VBIAS2,max
2.58V
0.13V
2.40V
0V (GND)
0.93V
1.16V
1.41V
= 2.58V
= 2.58V
= 2.58V
= 2.58V
= 2.58V , BIAS2 free, no Md
= 2.58V , BIAS2 free, no Md
TABLE VI
1
2
3
4
Md
VDD
IP
VDD
VDD VDD
IN
OUT
IREF
Fig. 10. Op-amp schematic with modied topology for low voltage operation
VDD,min
limiting device
disabled devices
2.58V
2.58V
2.07V
2.06V
Mc1
Mc2
Ms1
Ms2
Mc1
Mc1, Mc2
Mc1, Mc2, Ms1
VDD
VDD
Mc1
VDD
VDD VDD
Mt
BIAS2
IP
Md
IN
Ms1
Ms2
OUT
IREF
Mc2
Fig. 9.
Op-amp schematic
jNy
[1] T. Massier, H. Graeb, and U. Schlichtmann, The sizing rules method for
CMOS and bipolar analog integrated circuit synthesis, Computer-Aided
Design of Integrated Circuits and Systems, IEEE Transactions on, vol. 27,
no. 12, pp. 22092222, Dec. 2008.
[2] G. Stehr, M. Pronath, F. Schenkel, H. Graeb, and K. Antreich, Initial
sizing of analog integrated circuits by centering within topology-given implicit specication, in ICCAD 03: Proceedings of the 2003 IEEE/ACM
international conference on Computer-aided design. Washington, DC,
USA: IEEE Computer Society, 2003, p. 241.
[3] R. Rutenbar, G. Gielen, and J. Roychowdhury, Hierarchical modeling,
optimization, and synthesis for system-level analog and RF designs,
Proceedings of the IEEE, vol. 95, no. 3, pp. 640669, March 2007.
[4] T. Levi, N. Lewis, J. Tomas, and P. Fouillat, Scaling rules for MOS
analog design reuse, in Proceedings of Mixed Design of Integrated
Circuits and Systems MIXDES06, Gdynia, Poland, 2006, pp. 378382.
[5] D. Mueller, G. Stehr, H. Graeb, and U. Schlichtmann, Fast evaluation
of analog circuit structures by polytopal approximations, in Circuits
and Systems, 2006. ISCAS 2006. Proceedings. 2006 IEEE International
Symposium on, 2006.
[6] G. Stehr, H. Graeb, and K. Antreich, Analog performance space
exploration by normal-boundary intersection and by Fourier-Motzkin
elimination, Computer-Aided Design of Integrated Circuits and Systems,
IEEE Transactions on, vol. 26, no. 10, pp. 1733 1748, oct. 2007.
[7] R. J. Baker, CMOS Circuit Design, Layout and Simulation, Second
Edition. IEEE, WILEY-INTERSCIENCE, 2005.
[8] S. Hoppner, J. Gorner, S. Henker, R. Schuffny, and A. Graupner, A
lookup table ow for analog design automation, in Proceedings of
GMM/ITG-Fachtagung Analog 2010. VDE Verlag, Mar 2010.
[9] G. B. Dantzig and M. N. Thapa, Linear Programming, 1: Introduction.
SPRINGER, 1997.