Академический Документы
Профессиональный Документы
Культура Документы
Introduction
r IEE, 2005
IEE Proceedings online no. 20041113
doi:10.1049/ip-cds:20041113
Paper rst received 27th October 2003 and in revised form 23rd August 2004
M. Faisal and A.B.M.H. Rashid are with the Department of Electrical and
Electronic Engineering, Bangladesh University of Engineering and Technology,
Dhaka-1000, Bangladesh
A. Hasib is with the Institute of Information and Communication Technology,
Bangladesh University of Engineering and Technology, Dhaka-1000,
Bangladesh
E-mail: abdulhasib@iict.buet.ac.bd
IEE Proc.-Circuits Devices Syst., Vol. 152, No. 3, June 2005
M1
M4
M3
Q
NAND
M1
M3
A
M4
M2
Q
AND
B
M2
M4
B
M2
2.1
M1
Fig. 1
M3
Q
OR
Q
EXOR
NOR
EXNOR
Struck-on fault
VIH
cannot be detected by logic monitoring. However, steadystate current is very large due to the low resistance path
between VIH and ground. Steady-state current is given by
I VIH =Rf Ron
2
Hence, the fault can be detected by current monitoring, i.e.
IDDQ testing.
SPICE simulation has been used to analyse the effect of
stuck-on fault on all transistors of all the basic CPL gates.
Figure 3 shows the variation of output voltage Vout and
signal current IDDQ as a function of Rf for a fault on the
transistor M1 of the basic CPL AND gate with test vector
[A 1, B 0]. This is in agreement with the analysis made
above. The simulation revealed that the current under fault
condition, varied from 3 mA to 0.24 mA, whereas the
normal operating current was only 5 pA. Thus, this fault
can be detected by current monitoring. Similarly it has been
found that all single stuck-on faults in all CPL basic circuits
can be detected by current monitoring by applying
appropriate test vectors, but no logic monitoring is possible.
The result is summarised in Table 1.
2.2
Stuck-at fault
A
0
5
M1
output voltage, V
VIH
Rf
M2
1
3
current (mA)
2
output voltage
current IDDQ, mA
3
0
0
10
15
20
Stuck-on
MOS
XOR/XNOR
gate
M1
(00)
no
no
no
(01)
no
yes
yes
(10)
yes
no
no
(11)
no
no
yes
(00)
no
no
yes
(01)
yes
no
no
(10)
no
yes
yes
(11)
no
no
no
(00)
no
no
yes
(01)
yes
no
no
(10)
no
yes
yes
(11)
no
no
no
(00)
no
no
no
(01)
no
yes
yes
(10)
yes
no
no
(11)
no
no
yes
M3
M4
4
1
output voltage, V
2
current (mA)
output voltage
3
0
0
VIH
current IDDQ, mA
M2
10
15
20
fault strength Rf , k
VIH
VIH
M1
2.3
M2
B
Rf
Vout
Stuck-open fault
Physical defects or electromigration in aluminum conductors may cause a MOS transistor to become permanently
open and insensitive to its input signal. To model a stuckopen fault a large resistance is inserted between the source/
drain terminal and the circuit node to which the terminal
would otherwise be connected.
A single stuck-open fault can be detected by applying a
two-pattern test; the rst vector to be applied is called the
initialisation vector and the second vector is called the test
vector [13, 14]. Two vectors are applied to the faulty circuit
sequentially. These two vectors are chosen so that under
fault-free conditions, the outputs are complements to each
other. The rst one initialises the relevant output node to a
denite logic state. The second one sensitises the fault; it
causes both nMOS devices connected to the same output
node to be OFF. As a result the output node becomes
217
XOR/XNOR
gate
M1
(00)
yes
VIH
VIH
VIH
M1
M2
M3
M4
no
yes
(01)
yes
yes
yes
(10)
no
no
no
(11)
no
yes
no
(00)
yes
no
no
(01)
no
no
no
(10)
yes
yes
yes
(11)
yes
no
yes
(00)
no
yes
yes
(01)
yes
yes
yes
(10)
no
no
no
(11)
no
yes
no
(00)
yes
no
no
(01)
no
no
no
(10)
yes
yes
yes
(11)
yes
no
yes
M2
M3
M4
XOR/XNOR
gate
yes
o004
no
yes
o014
yes
no
no
o104
yes
no
no
o114
no
yes
yes
o004
yes
no
yes
o014
yes
no
no
o104
yes
no
no
o114
yes
no
yes
o004
no
yes
no
o014
no
yes
yes
o104
no
yes
yes
o114
no
yes
no
B
Rf
Cout
Vout
2.4
Bridging fault
o004
yes
yes
yes
Stuck-open
AND/NAND
OR/NOR
XOR/XNOR
o014
no
no
no
MOS
o104
no
no
no
o114
yes
yes
yes
Successful
2-pattern
test
oAB,AB4
Successful
2-pattern
test
oAB,AB4
Successful
2-pattern
test
oAB,AB4
M1
M2
M2
M3
M4
o00, 114
o01,004
o01,004
o10,114
o11,004
o10,004
o11,004
o00,014
o00,014
o11,104
o00,114
o11,014
o11,004
o00,014
o00,014
o11,104
o00,114
o11,014
o00,114
o01,004
o01,004
o10,114
o11,004
o10,004
M4
M1
3.1
M2
M3
Q
OR
Rf
Q
NOR
output voltage, V
3
1.0
2
0.5
Q
Q
IDDQ
1.5
Stuck-on fault
M2
A
M1
A
0
10
fault strength, k
15
M3
0
20
M4
M6
Fig. 8 Variation of Vout and IDDQ with fault strength Rf for output
bridging of AND/NAND module
M5
M8
M7
a
M1
M2
M6
M5
B
B
M7
M3
M4
M8
M10
M9
A
M11
C
M12
b
Fig. 9
a SUM
b CARRY
219
VIH
3.2
VIH
A
VIH
M3
M4
VIH
3.3
Stuck-open fault
Stuck-at fault
M5
M7
Rf
Fig. 10 Equivalent circuit for stuck-on fault on M5 of CPL fulladder SUM circuit for test vector [A 0, B 0, C 1]
Table 6: Simulation results for stuck-on faults in CPL fulladder CARRY circuit
Fault
LM
CM
M1
(001),(100)
no
yes
M3
(011),(110)
no
yes
M5
(011),(110)
no
yes
M7
(001),(100)
no
yes
M9
(001),(011)
no
yes
M11
(100),(110)
no
yes
Table 7: Simulation results for stuck-at faults in CPL fulladder SUM circuit
Fault
LM
CM
M1
no
yes
M2
no
yes
M3
no
yes
M4
no
yes
M5
no
yes
M6
no
yes
M7
no
yes
M8
no
yes
Table 5: Simulation results for stuck-on faults in CPL fulladder SUM circuit
Fault
LM
CM
M1
no
yes
M2
no
yes
M3
no
M4
no
M5
M6
Table 8: Simulation results for stuck-at faults in CPL fulladder CARRY logic circuit
Fault
LM
CM
yes
M1
no
yes
yes
M3
(000),(100) (110),(111)
no
yes
no
yes
M5
(000),(001) (011),(111)
no
yes
no
yes
M7
(010),(100) (101),(110)
no
yes
M7
no
yes
M9
(011), (100)
no
yes
M8
no
yes
M11
no
yes
Table 9: Simulation results for stuck-open fault in CPL fulladder SUM circuit
Fault
Successful
two-pattern
vectors
O/P logic
level
un-faulted
O/P logic
level
faulted
LM
CM
Fault
Successful
two-pattern
vectors
oABC4
M1
(000,100)
01
00
yes
no
(011,100)
01
00
yes
no
M1
(000,111)
0,1
0,0
yes
no
(100,110)
0,1
0,0
yes
no
M2
(101,100)
01
00
yes
no
(110,100)
01
00
yes
no
M7
M8
(011,100)
1,0
1,1
yes
no
1,0
1,1
yes
no
10
11
yes
no
10
11
yes
no
(110,100)
1,0
1,1
yes
no
no
(111,100)
1,0
1,1
yes
no
no
(100,101)
0,1
0,0
yes
no
(101,000)
1,0
1,1
yes
no
10
10
11
11
yes
yes
M5
(101,111)
01
00
yes
no
(110,111)
01
00
yes
no
(111,000)
1,0
1,1
yes
no
no
(101,001)
1,0
1,1
yes
no
no
(110,001)
1,0
1,1
yes
no
(111,001)
1,0
1,1
yes
no
(001,000)
10
10
11
11
yes
yes
(011,000)
10
11
yes
no
(100,000)
10
11
yes
no
(111,000)
10
11
yes
no
(000,010)
01
00
yes
no
M7
M9
(100,011)
0,1
0,0
yes
no
(110,010)
0,1
0,0
yes
no
(011,100)
1,0
1,1
yes
no
0,1
0,0
yes
no
no
(111,101)
10
11
yes
no
(000,101)
(000,001)
01
00
yes
no
(001,101)
0,1
0,0
yes
no
(010,100)
0,1
0,0
yes
no
(101,010)
1,0
1,1
yes
no
(011,001)
M6
CM
(001,110)
(010,000)
M5
LM
(001,101)
(111,101)
M4
O/P logic
level
faulted
(101,100)
(010,101)
M3
M3
O/P logic
level
un-faulted
01
00
yes
M11
(101,001)
01
00
yes
no
(001,011)
10
11
yes
no
(110,010)
1,0
1,1
yes
no
(100,011)
10
11
yes
no
(111,010)
1,0
1,1
yes
no
(111,011)
10
11
yes
no
(001,000)
10
11
yes
no
(111,000)
10
11
yes
no
(000,110)
01
00
yes
no
(110,010)
01
00
yes
no
(001,100)
10
11
yes
no
(101,100)
10
11
yes
no
(001,110)
01
00
yes
no
(111,110)
01
00
yes
no
(000,001)
01
00
yes
no
(110,001)
01
00
yes
no
(010,011)
10
11
yes
no
(100,011)
10
11
yes
no
(000,111)
10
11
yes
no
(110,111)
10
11
yes
no
(010,101)
01
00
yes
no
(100,101)
01
00
yes
no
Conclusions
Theoretical analysis and SPICE simulations of the testability of basic CPL circuits under various single stuck faults
has been presented. It was found that all stuck-on faults on
all CPL basic gates were detected by current monitoring but
no logic monitoring was possible. Similar results have been
obtained for stuck-at faults between gate and source of the
MOS devices of all basic CPL gates. However, for stuck-at
faults between gate and drain, it was found that all stuck-at
faults between gate and drain could be detected by current
monitoring except for the following two MOS devices: (i)
MOS M3 of the basic AND/NAND gate and (ii) MOS M2
of the basic OR/NOR gate for which the gate and drain
terminals have the same input variable. In the case of stuckopen faults, it was found that stuck-open faults on all MOS
transistors of all basic CPL gates were detected with logic
monitoring by applying an appropriate two-pattern test.
Stuck-at and stuck-on are the most common faults on VLSI
circuits and for CPL basic gate circuits IDDQ testing gives
fault coverage of more than 94% for stuck-at, stuck-on and
bridging faults. In the case of a CPL full-adder, we have
found that stuck-on and stuck-at faults on all transistors of
SUM logic and CARRY logic circuits can be detected by
current monitoring, i.e. IDDQ testing provides 100% fault
coverage. Like CPL basic circuits, stuck-open faults on all
transistors of a CPL full-adder are detectable by logic
monitoring applying an appropriate two-pattern test.
Therefore, it can be concluded that signal source current
monitoring (IDDQ testing) is the best method for common
fault detection in CPL circuits and that it gives a very wide
range of fault coverage. Again for detecting stuck-open
faults, logic monitoring with a two-pattern test is the only
available method so far and for CPL basic circuits it gives
fault coverage of 100%. Therefore, other than low power
222
References
1 Yano, K., Yamanaka, T., Nishida, T., Saito, M., Shimohigashi, K.,
and Shimizu, A.: A 3.8 ns CMOS 16 16 -b multiplier using
complementary pass-transistor logic, IEEE J. Solid-State Circuits,
1990, 25, (2), pp. 388395
2 Avci, M., and Yildirim, T.: General design method for complementary pass transistor logic circuits, Electron. Lett., 2003, 39, (1),
pp. 4648
3 Abu-Khater, I.S., Bellaouar, A., and Elmasry, M.I.: Circuit
techniques for CMOS low-power high-performance multipliers, IEEE
J. Solid-State Circuits, 1996, 31, (10), pp. 15351546
4 Abu-Khater, I.S., Bellaouar, A., Elmasry, M.I., and Yan, R.H.:
Circuit/architecture for low-power high-performance 32-bit adder.
Proc. IEEE 5th Great Lakes Symp. on VLSI, Buffalo, NY, USA,
1995, pp. 7477
5 Strallo, A.G.M., and Napoli, E.: A fast and area efcient
complimentary pass-transistor logic carry-skip adder. Proc. 21st Int.
Conf. on Microelectronics, MIEL, 1997, Vol. 2, pp. 701704
6 Wang, L.K., and Chen, H.H.: A low power high speed error
correction code macro using complementary pass-transistor logic.
Proc. 10th Annual IEEE Int. ASIC Conf. and Exhibition, 1997,
pp. 1720
7 Fuse, T., Oowaki, Y., and Terauchi, M.: An ultra low voltage SOI
CMOS pass-gate logic, IEICE Trans. Electron, 1997, E80-C, (3),
pp. 472477
8 IBM S/390 G3 and G4, IBM J. Res. Dev., 1997, 41, special issue,
Nos. 4/5, July/Sept. 1997
9 Williams, T., and Parker, K.: Design for testability - a survey, IEEE
Trans. Comput., 1982, C-31, pp. 215
10 Aziz, S.M., Rashid, A.B.M.H., and Karim, M.: Fault characterization of complementary pass-transistor logic circuits. Proc. IEEE Int.
Conf. on Semiconductor Electronics, Malaysia, 2000, pp. 8084
11 Rashid, A.B.M.H., Karim, M., and Aziz, S.M.: Testing complementary pass-transistor logic circuits. Proc. IEEE Int. Symp. on Circuits
and Systems (ISCAS 2001), Sydney, Australia, 2001, Vol. 4, pp. 58
12 Jha, N.K., and Kundu, S.: Testing and reliable design of CMOS
circuits (Kluwer Academic Publishers, USA, 1990)
13 Wadsack, R.L.: Fault modeling and logic simulation of
CMOS and MOS integrated circuits, Bell Syst. Tech. J., 1978, 57,
(5), pp. 14491474
14 Maly, W., Nag, P.K., and Nigh, P.: Testing oriented analysis of
CMOS ICs with opens. Proc. Int. Conf. on Computer-Aided Design,
Santa Clara, CA, 1995, pp. 344347
15 Mei, K.C.Y.: Bridging and stuck-at faults, IEEE Trans. Comput.,
1974, C-23, (7), pp. 720727
16 Bernardini, S., Portal, J. M., and Masson, P.: A tunneling model for
gate oxide failure in deep sub-micron technology. Proc. Design,
Automation and Test in Europe (DATE) Conf., 1620 Feb 2004,
Vol. 2, pp. 14041405
17 Dragic, S., and Margala, M.: A 1.2 V built-in architecture for high
frequency on-line IDDQ/delta IDDQ test. Proc. IEEE Computer
Society Annual Symp. on VLSI, Pittsburg, USA, 2526 April,
pp. 148153
18 Powell, T.J., Pair, J., St. John, M., and Counce, D.: Delta IDDQ for
testing reliability. Proc. 18th IEEE VLSI Symp., Montreal, Canada,
45 April 2000, pp. 439443