1

At-Speed Transition Fault Testing With Low Speed

Scan Enable
Nisar Ahmed, C. P. Ravikumar Mohammad Tehranipoor, Jim Plusquellic
ASIC Product Development Center Dept. of Computer Science and Electrical Eng.
Texas Instruments India University of Maryland Baltimore County
Bangalore - 560 093 Baltimore, MD 21250
n-ahmed2,ravikumar @ti.com

tehrani,plusquel @umbc.edu

IC LC CC
A BSTRACT
CLK
With today’s design size in millions of gates and working fre-
quency in gigahertz range, at-speed test is crucial. The launch- SEN

off-shift method has several advantages over the launch-off-

capture but imposes strict requirements on transition fault test- Scan−in pattern i Scan−in pattern i+1
Scan−out response i−1 Scan−out response i
ing due to at-speed scan enable signal. A novel scan-based at- (a)
speed test is proposed which generates multiple local fast scan
enable signals. The scan enable control information is encap-
IC LC CC
sulated in the test data and transferred during the scan opera-
tion. A new scan cell, referred to as last transition generator CLK

(LTG), is inserted in the scan chains to generate the fast local SEN
scan enable signal. The proposed technique is robust, practice-
oriented and suitable for use in an industrial flow.
Scan−in pattern i Scan−in pattern i+1
Scan−out response i−1 Scan−out response i
(b)
I. I NTRODUCTION
Figure 1. Transition Delay Fault Pattern Generation Methods: (a) Launch-off-
The semiconductor industry is adopting new fabrication pro- shift and (b) Launch-off-capture.
cesses to meet the area, power and performance requirements.
As a result, modern ICs are growing more complex in terms of response of the CUT to the pattern V 2 must be captured at func-
gate count and operating frequency [1]. The deep-submicron tional speed (rated clock period). The whole operation can
(DSM) effects are becoming more prominent with shrinking be divided into 3 cycles: 1) Initialization Cycle (IC), where
technology, thereby increasing the probability of timing-related the CUT is initialized to a particular state (V 1 is applied), 2)
defects [2] [3]. For DSM designs, the stuck-at fault test alone Launch Cycle (LC), where a transition is launched at the tar-
cannot ensure high quality level of chips. In the past, functional get gate terminal (V 2 is applied) and 3) Capture Cycle (CC),
patterns were used for at-speed test. However, functional test- where the transition is propagated and captured at an observa-
ing is not a viable solution because of the difficulty and time tion point.
to generate these tests for complex designs with very high gate Depending on how the transition is launched and captured,
density. Therefore, more robust at-speed techniques are nec- there are two transition fault pattern generation methods. In the
essary as the number of timing-related defects is growing and first method, referred to as launch-off-shift (LOS), the transition
effectiveness of functional and IDDQ testing is reducing [4] [5]. at the gate output is launched in the last shift cycle during the
The transition fault and path delay fault testing together pro- shift operation. Figure 1(a) shows the launch-off-shift method
vide a relatively good coverage for delay-induced defects [6] waveform for a multiplexed-DFF design; similar approach can
[7]. Path delay model targets the cumulative delay through the be applied to an LSSD. The LC is a part of the shift operation
entire list of gates in a pre-defined path while the transition fault and is immediately followed by a fast capture pulse. The scan
model targets each gate output in the design for a slow-to-rise enable (SEN) is high during the last shift and must go low to
and slow-to-fall delay fault [8]. Scan-based structural tests gen- enable response capture at the CC clock edge. The time pe-
erated by an automatic test pattern generator (ATPG) are in- riod for SEN to make this 1 0 transition corresponds to the


creasingly used as a cost-effective alternative to the at-speed functional frequency. Hence, LOS requires the SEN signal to
functional pattern approach [5]. It also provides high controlla- be timing critical. Skewing the clock (CLK) creates a higher
bility and observability. launch-to-capture clock frequency than standard shift clock fre-
To perform a transition fault test, a pattern pair V 1 V 2 is
  
quency. Saxena et al. [9] list more launch and capture wave-
applied to the circuit-under-test (CUT). Pattern V 1 is termed forms used by launch-off-shift approaches.
as the initialization pattern and V 2 as the launch pattern. The Figure 1(b) shows the waveforms of the second approach, re-
 2

From CUT To CUT

ferred to as launch-off-capture (LOC) method. In this method,
the launch cycle is separated from the shift operation. At the TEST_SI / D
TEST_SO /
last scan cell Q next scan cell
end of scan-in (shift mode), pattern V 1 is applied and CUT SD

is set to an initialized state. A pair of at-speed clock pulses SEN_port

SE

are applied to launch and capture the transition at the target D Q

SEN_pipeline
FF
gate terminal. This relaxes the at-speed constraint on the SEN
CLK
signal and dead cycles are added after the last shift to provide
enough time for the SEN signal to settle low. The launch-off- (a)

shift method is more preferable based on ATPG complexity and IC LC CC

pattern count compared to launch-off-capture method. In case
of LOC, a high fault coverage cannot be guaranteed due to the CLK

correlation between the two patterns, V 1 and V 2. SEN_port

As the design size increases, the SEN fanout exceeds any SEN_pipeline
other net. One possible solution is to design the scan enable as a
clock tree network but this is rarely followed due to high design Scan−in pattern i
Scan−out response i−1
Scan−in pattern i+1
Scan−out response i
cost [10]. Multiple SEN ports can be used to reduce the fanout (b)

but results in increasing the number of pins [9]. This may be

Figure 2. Pipeline Scan Enable.
a limitation for designs to be tested using very low cost testers
(VLCT). In [11], a hybrid architecture is proposed which con-
[12]. The scan enable port (SEN port) is de-asserted asyn-
trols a small subset of selected scan cells by launch-off-shift and
chronously in the initialization cycle and the pipeline scan en-
the rest are controlled by launch-off-capture approach. A fast
able (SEN pipeline) goes low synchronously at the active edge
scan enable signal generator is designed which drives the LOS
of launch clock cycle. The SEN port is asserted high after the
controlled scan flip-flops. The ATPG method used is complex
capture cycle to re-enter the shift mode. The pipeline scan en-
and current commercial tools do not support such a technique.
able architecture provides two significant advantages. Firstly,
Moreover, the selection criteria of the LOS controlled scan flip-
the SEN pipeline signal is generated synchronously and has a
flops determines the effectiveness of the method. In some cases
full clock cycle available for transition, assuming the pipeline
the number of patterns generated by the hybrid method exceeds
scan enable signal is generated separately for positive and nega-
the launch-off-capture patterns.
tive edge flip-flops. Secondly, the tester skew at the scan enable
A very widely used method is to pipeline the scan enable
port is decoupled from the internal scan enable net.
signal [12]. In a multi-stage pipeline scan enable, the designer
The pipeline scan enable method requires routing awareness
must carefully select the group of scan cells controlled by the
before insertion. The number of pipelined flip-flops added in
respective scan enable signal. In order to meet timing closure of
the scan enable path depends on the worst case delay. Mul-
the pipeline scan enable signals, this selection criteria manifests
tiple pipeline scan enable stages can be added depending on
into design iterations and additional constraints for the place
the distance of the last stage driving the scan enable network.
and route (PNR) tool.
However, multiple stage pipelines have increased susceptibility
In this paper, we propose a scheme where the scan enable
to noise glitches from tester or due to crosstalk. An assertion
control information for the launch and capture cycle is embed-
glitch on the first pipeline stage would force scan enable active
ded in the test data itself. A new scan cell, called the last tran-
for multiple clock cycles. Note that, the pipeline scan enable
sition generator (LTG), generates the local fast scan enable sig-
scheme is an industry practiced technique and the proposed
nals. The LTG cell has the flexibility to be inserted anywhere
technique discussed in the next section (Section III) provides
in the scan chain and the hardware area overhead is compara-
more robustness and can be practiced along with the pipeline
ble to the pipeline scan enable approach. The proposed method
scan enable scheme.
poses no additional constraints for the place and route tool and
provides more flexibility to re-order the scan cells to meet the III. L OCAL S CAN E NABLE S IGNAL G ENERATION
timing closure of the local scan enable signals.
The rest of the paper is organized as follows. The pipeline A. Motivation
scan enable method is reviewed in Section II. Section III de- The pipeline scan enable methodology provides a good
scribes the local scan enable generation and the architecture mechanism to divide the fanout of the scan enable signal with-
and operation of LTG cell. Section IV explains the test archi- out using multiple external pins and eliminates the external scan
tecture and launch-off-shift clock timing waveforms. The DFT enable port tester skew. It is independent of the floor planning
insertion and test protocol file changes for ATPG along with the step and does not provide much flexibility to the PNR tool in
experimental results are discussed in Section V. Finally, con- terms of selecting the scan cell group driven by the pipeline
cluding remarks are in Section VI. scan enable, which in some cases may result in design itera-
tions. In order to provide more flexibility with all the advan-
II. P IPELINE S CAN E NABLE
tages of the pipeline scan enable, the local scan enable gener-
The launch-off-shift method requires the scan enable sig- ator cells are inserted within the scan chains. Therefore, the
nal to be at-speed. Figure 2 shows the basic implementation control information is to be passed as part of the test data. The
and operation of a pipelined scan enable for launch-off-shift new architecture provides the PNR tool with more flexibility to
 3

10001110
A
0
1 0 D Q D Q Q
SD 1
2 1 0
FF1 FF2
3 1 1 0
4 1 1 1 0
CLK
5 0 1 1 1 0
6 0 0 1 1 1 0 SEN (GSEN)
in
7 0 0 0 1 1 1 0
8 1 0 0 0 1 1 1 0

IC LC CC

CLK
SEN (LSEN)
out
GSEN
Pattern (1 0 0 0 1 1 1 0)
A Figure 4. Last transition generator (LTG) cell.

LSEN=(GSEN+A) LSEN must remain in this state as long as it is asynchronously

set to 1 by GSEN. This requires additional atpg constraints for
Figure 3. Generation of local scan enable signal. a conventional scan flop architecture to control the functional
reorder the scan chain cells including the scan enable generator path (D) to logic 0. This might lead to fault coverage reduction.
cells during timing closure step. In order to avoid loss of coverage without significant change
Figure 3 shows a small example of generating the local scan in the architecture, a new scan cell called last transition
enable signal from the test pattern data for LOS. The external generator (LTG) is designed such that when the scan cell is
scan enable signal from the tester is referred to as the global loaded with logic 0 during shift and the scan enable control
scan enable (GSEN). The internally generated scan enable sig- is made logic 0, i.e. capture state, the cell will remain in
nal is termed as local scan enable (LSEN). The main objective that state until it is shift controlled to 1. This cell is inserted
is to de-assert GSEN in the initialization cycle (not at-speed) into scan chain to generate the 1 0 transition at the launch


and then generate the LSEN signal during the launch and cap- edge. Figure 4 shows the LTG cell architecture. It consists of
ture cycle synchronously from the test data. There are eight two flip-flops which are used to load the control information
scan flip-flops in the scan chain and the test pattern shifted is required for the launch and capture cycles. The port definition
(10001110). The values of the scan flip-flops during the vari- is similar to a scan cell and the output of FF1 is fed back to the
ous shift cycles are shown under each flop. GSEN is de-asserted functional input port of the scan cell. It consists of a scan-in
during the (n 1)th shift (IC) cycle, where n=8. (SENin ) pin which takes GSEN signal as input. An additional
For proper shift operation, the LSEN signal should be logic scan-out (SENout ) pin (GSEN+Q) represents the LSEN signal.
1 in the (n 1)th cycle of the shift operation (IC) and logic 0 The LTG cell can be inserted anywhere in the scan chain and it
in the last shift cycle (LC) to enable capture in the next clock is not connected to the CUT. Therefore, any atpg constraint on
cycle. In other words, the LSEN signal must make a 1 0 
the LTG cell does not affect the CUT fault coverage.
transition at the launch edge. For this particular example, the
pattern during the shift operation generates the required 1 0 
Theorem: The local scan enable signal generated by the LTG
transition at the output of scan flop A. The output of scan flop cell switches at-speed during the capture cycle.
A is ORed with GSEN to generate the LSEN signal. Therefore, Proof: SENout refers to the local scan enable signal in the LTG
the final value of scan flop (A) and its following scan flop at the cell of Figure 4. The clock input to the LTG cell for launch-
end of shift operation must be 0 and 1, respectively, so that A off-shift transition delay ATPG is of the form shown in Figure
is loaded with logic 1 in IC and logic 0 in LC. A full at-speed 3. It is assumed that the clock tree synthesis tool is capable of
cycle is available for LSEN to make the transition. After the routing the clock signal so that the local clock signal at the input
capture cycle, the LSEN signal is aysnchronously set to 1 by of the LTG cell switches at functional speed during the LC and
GSEN for scanning out the response. CC cycles. During the scan shift cycle (IC), a ”1” is scanned
into the LTG cell at low frequency. During the last cycle of
B. Last Transition Generator (LTG) shift, denoted by LC, the clock switches at functional speed
As explained earlier, during launch-off-shift pattern genera- and the output of FF1 also switches to 0 state at the functional
tion, to generate the scan enable transition 1 0 at the launch


speed, since the number of flip-flops driven by SENout is an

edge, the scan flop A should be 0 and the following scan flop be order of magnitude smaller than the total number of flip-flops
1. This is very unlikely to occur in every pattern generated dur- in the design, thereby reducing the capacitive load on the local
ing atpg. It can also be seen in Figure 3 that there is an unknown scan enable signal. The global scan enable signal switches to 0
value loaded in A during the capture edge, which can cause a during the beginning of the LC cycle. Let A refer to the output
problem if the method is to be used for LOC pattern generation. of FF1. SENout is the logical OR of the signal A and the global
For LOC patterns, the LSEN signal must be de-asserted and two scan enable signal; therefore, SENout also switches at the speed
system clocks are applied. Hence, A must be atpg constrained of signal A, except for the small delay in the OR gate.
to 0 during LOC pattern generation. This constraint is neces-
sary but not enough for proper operation as the value loaded in C. Operation of LTG cell
A from the functional path after applying the first system clock Figure 5(a) shows the previous example with the LTG cell
is not known. Therefore, after going to capture control state (0), inserted in the scan chain. Note that, the LTG cell can be
 4

Pattern: 1000[XX]1110 LTG x x

A
x x x x
x x x x
x x x x
x x x x
(a) x x
x x
x x
x x x x
LOS: 1000[01]1110 x x
x x x x x x

x x x x
IC LC CC x
x xx x
x xx
x x
x x x x x x x x
x x
x x
CLK x x
x x x x
x x x x x x
SENin (GSEN)

Q (FF1)
(b)
(a)
SENout (LSEN)

Scan−in pattern i Scan−in pattern i+1

Figure 6. (a) Distribution of flip-flops in the chip layout. (b) Partitioning the
Scan−out response i−1 Scan−out response i set of flip-flops for predictable closure on scan enable timing.
(b)
LOC: 1000[0X]1110
LTG cells one can insert. One can look at the distribution of
IC LC CC
flip-flops in the layout of the chip (see Figure 6).
CLK The difficulty of timing closure for the scan enable signal can
SENin (GSEN) be estimated by some function f of the area of smallest bounding
Q (FF1)
rectangle that covers the majority of the flip-flops (dotted rect-
angle in Figure 6(a)). The rectangle then needs to be partitioned
SENout (LSEN)
such that we have nearly equal number of flip-flops in each of
Scan−in pattern i Scan−in pattern i+1 the partitions, and each bounding sub-rectangle is small enough
Scan−out response i−1 Scan−out response i
to limit the function f to an acceptable value as shown in Figure
(c)
6(b). The process of partitioning can be performed through the
Figure 5. Operation of LTG cell, (b) LOS and (c) LOC. repeated use of a bi-partitioning algorithm such as Kernighan-
placed anywhere in the scan chain and it is not connected to Lin [13]. Note that the assignment of flip-flops to scan chains
the CUT. The values of the two flip-flops of the LTG cell in ordering of the flip-flops within the scan chains is not directly
the test pattern are shown as X (1000[XX]1110). These flip- relevant to the partitioning problem; therefore, this step can be
flops are constrained during atpg to assign specific values to the performed after the physical design flow has reordered the flip-
Xs. Figure 5(b) shows the pattern and the timing waveform for flops in the scan chain and optimized the placement and routing
LOS. During the shift operation, at the last shift the scan enable for functional timing closure. The location of the LTG cell in
must make a 1 0 transition. Thus, FF1 of LTG cell should

a scan chain can be selected such that the cell is nearly in the
be loaded with 1 in IC followed by 0 in the next cycle (LC). center of the leaf-level rectangle.
The SENin (GSEN) signal is asynchronously deasserted in IC. In general, there can be multiple scan chains in a design to
The SENout (LSEN) signal is generated by the boolean equation reduce the test application time. The test application time is
SENout Q FF1 SENin. After the capture cycle, the LSEN
 

directly proportional to the longest scan chain length in the de-

is asserted to 1 asynchronously by GSEN. sign. Figure 7 shows a multiple scan chain architecture with n
For LOC, the GSEN signal is high during the entire shift op- scan chains. Each scan chain i, where 1 i n, consists of  

eration. At the end of the shift operation, the GSEN signal is a LTG cell which generates the fast scan enable signal LSENi .
asynchronously deasserted and the LSEN signal must be logic Note that, if scan enable timing is not met then multiple LTG
0. The value of FF2 does not affect the operation. Hence, only cells can be inserted to generate multiple LSEN signals to con-
FF1 of LTG cell must be constrained to 0 during atpg. Figure trol different segments of the same scan chain. The fanout load
5(c) shows the pattern and the timing waveform for LOC. The on the global scan enable (GSEN) signal is reduced and the
LSEN signal is asynchronously de-asserted and asserted back fanout load driven by a local scan enable signal is used as a
by the GSEN signal. It can be noticed that these transitions are constraint to find the number of LTG cells inserted. For exam-
not at-speed. ple, for m total number of flip-flops in a design and n being the
maximum number of flip-flops that can be allowed for the lo-
IV. T EST A RCHITECTURE cal scan enable to be timing closed for a particular operating
The LTG-enabled solution presented in this paper consider- frequency, the number of LTG cells are estimated by m n. 

ably eases the problem of routing the scan enable signal by tak- The methodology is not effected by multiple clock domains.
ing in a global scan enable signal that need not switch at func- Since all scan chains are shifted at the same speed, the launch
tional speed and generating the local scan enable signals inter- edge of all the clock domains occur at the same time. The
nally. The number of local scan enable signals can be specified timing constraint for the LSEN signal is that it must be timing
by the user. The overhead of generating the local scan enable closed for the fastest clock domain in the scan chain. The tran-
signal is the addition of an LTG cell in the scan chain. The area sition fault patterns for LOS are generated per clock domain. If
overhead of an LTG cell is a few extra gates, which is negli- the pattern is shifted at a slow speed followed by a fast capture,
gible in modern designs. Each local scan enable signal drives the time from the launch edge of LC to the capture edge of CC
a fraction of the total number of flip-flops in a clock domain. is not really at-speed. Figure 8(a) shows the limitation of the
The question naturally arises on what is the largest number of clock timing waveform. The functional operating frequency is
 5
TABLE I
D ESIGN C HARACTERISTICS
LSEN1

Chain 1
Transition Delay Faults 4363144
LTG cell Scan Flops 69836
Non-scan Flops 97
GSEN
Scan Chains 16
LSEN2 Chain 2
Clock Domains 2
TABLE II
LSENn
E XPERIMENTAL R ESULTS
Chain n

Launch-off-Capture (LOC) Launch-off-shift (LOS)

Figure 7. Test Architecture. CLK1 CLK2 CLK1 CLK2
IC LC CC
100
Detected faults 466687 3733728 466707 3893150
45 55 2 6
Test Coverage 10.7 85.75 10.7 89.44
CLK Pattern Count 3 9401 3 14332
CPU Time [sec] 235792 172031
SEN
Memory Usage [MB] 1031 1194

Scan−in pattern i
Scan−out response i−1
Scan−in pattern i+1 used for ATPG. We see that LOS methodology gives about 4%
Scan−out response i
(a) higher coverage than the LOC methodology. In separate experi-
ments performed at our organization, it has been independently
IC DC LC CC confirmed that LOS technique gives up to 10% higher coverage
45 55 2 6 2 6
on most designs. For the design under consideration, we found
CLK
that using 1312 patterns, the LOS method gave the same cov-
SEN erage as the LOC method. This represents a reduction of about
88% as compared to LOC. The main barrier to the practice of
Scan−in pattern i Scan−in pattern i+1 LOS is the difficulty in closing the timing on the scan enable
Scan−out response i−1 Scan−out response i
signal, as a result, LOS is not used on designs where the turn-
(b)
around-time is critical. The price to be paid for this is two fold
Figure 8. LOS clock timing waveform. - (a) increase in test cost, since the pattern volume for LOC is
higher, and (b) reduction in coverage, which impacts the defec-
125MHz. The launch edge in the last shift occurs at 45ns and
tive parts per million (DPPM) metric.
the capture edge occurs at 2ns in the capture cycle of 8ns time
period. The time from the launch edge to the capture edge is A. DFT Insertion
(55+2)=57ns. Figure 8(b) shows the modified at-speed clock Synopsys DFTCompiler [15] is used for scan chain insertion
timing waveform used for LOS. The last shift is done at-speed in a design. Figure 9 shows the list of additional commands in
corresponding to the clock domain being tested and the capture the tcl script. Here, we assume that one LTG cell is inserted
clock is applied only for that clock domain. A dead cycle (DC) per scan chain. To insert the LTG cells, additional commands
is added after the initialization cycle for the scan chain to settle are required during the scan chain configuration. The synthesis
down. tool must recognize the LTG cell as a scan cell in order to stitch
V. E XPERIMENTAL R ESULTS it into the scan chain. This requires it to be defined as a new
In this paper, we have argued in favor of the ‘launch off shift’ library cell with the scan cell attributes. A workaround is to
transition delay ATPG methodology and presented a technique design the LTG cell as a module and declare it as a scan segment
that can ease the implementation of this technique. We exper- using the set scan segment command (line 04). The tool then
imented with an industrial-strength design that had the follow- identifies LTG cell as a scan segment of length 2. The GSEN
ing characteristics (Table I). The design has 16 scan chians and signal is connected to all the LTG cells SENin input pin. To
almost 70K scan cells. There are 97 non-scan cells and two in- make the insert scan command insert the LTG cells in the scan
ternal clock domains, 125MHz and 250MHz, respectively. The chain, set scan path command must be used to declare the scan
test strategy is to get the highest possible test coverage for the path (line 09). Only the LTG cell is specified in the scan path,
stuck-at faults. When generating test patterns for the transition as the tool will stitch the rest of the cells including the LTG
faults, we target only the faults in the same clock domain. Dur- cell and balances the scan chain depending on the longest scan
ing pattern generation, only one clock is made active during the chain length parameter defined in the set scan con f iguration
capture cycle. Hence, only faults in that particular clock domain command. The set scan signal command (line 10) is used to
are tested. All PIs remain unchanged and all POs are unobserv- hookup each LTG cells SENout port in a particular chain to all
able while generating the test patterns for the transition faults. the scan enable input port of the scan flip-flops in the respective
This is because the tester is not fast enough to provide the PI chain.
values and strobe POs at speed. B. ATPG
The results for LOS and LOC transition-delay ATPG on this There is no fundamental difference in the ATPG methodol-
design are shown in the Table II. TetraMAX [14] tool was ogy when we use the LTG-based solution. The scan enable sig-
 6

01: for set i 0
 $i  no chains
 incr i
method can be practiced along with other techniques such as
02: pipelined scan.
03: create cell LTG$i LTG VI. C ONCLUSION
04: set scan segment scan segment$i
-access test scan in LTG$i/SD, test scan out LTG$i/Q

05:
In this paper, a new method has been proposed to enable
06: -contains LTG$i/FF1, LTG$i/FF2
the design teams to practice launch-off-shift (LOS) transition
07: connect net GSEN find( pin, LTG$i/SEN IN) delay testing. LOS testing is known to provide better quality
08: connect net CK find( pin, LTG$i/CLK)
09: set scan path c$i LTG$i -dedicated scan out true -clock CK results, both in terms of pattern count and fault coverage, but
10: set scan signal test scan enable -port GSEN design teams may not use launch-off-shift due to the challenge
11: -hookup find( pin, LTG$i/SEN OUT) -chain c$i of routing the scan enable signal. Our solution is to generate
12:

local scan-enable signals that can switch at functional speeds;
for this purpose, we rely on embedding some control informa-
Figure 9. DFTCompiler Tcl Script commands. tion in the patterns. We use a special cell called the LTG cell
01: ”load unload” for the generation of the local scan enable signal. This cell is
02: W ” slow WFT ”; simple to design and layout, and its area overhead is compara-
03: V ”CLK1”=0; ”CLK2”=0;”GSEN”=1;
ble to that of a scan flop. The number of LTG cells inserted in
04: Shift
05: W ” slow WFT ”; the design will be small, thereby making the area overhead due
06: V ”CLK1”=P; ”CLK2”=P; ”GSEN”=1; ” so”=#; ” si”=#;
to our technique negligible. The LTG-based solution provides
07:
greater flexibility and fewer contraints to the backend flow dur-
//ADDING DEAD CLOCK CYCLE
08: V ”CLK1”=0; ”CLK2”=0; ”GSEN”=0;
ing place and route step. The DFT insertion and ATPG can be
09: W ” fast WFT ”;//Nth SHIFT CYCLE easily performed using the commercial ATPG tools; therefore
10: V ”CLK1”=P; ”CLK2”=P; ”GSEN”=0; ” so”=#; ” si”=#;

11:
our solution is easy to practice.
R EFERENCES
Figure 10. TetraMAX ATPG protocol file. [1] International Technology Roadmap for Semiconductors 2001
(http://public.itrs.net).
nal for the flip-flops now comes from an internally generated [2] S. Natarajan, M.A. Breuer, S.K. Gupta, “Process Variations and
Their Impact on Circuit Operation,” in Proc. IEEE Int. Symp. on
signal. An ATPG tool must be able to place the scan enable Defect and Fault Tolerance in VLSI Systems, pp. 73-81, 1998.
signal in the active mode during scan shift operation. In or- [3] R. Wilson, “Delay-Fault Testing Mandatory, Author Claims,” EE
der to get around the problem of the ”design rule check” phase Design, Dec. 2002.
of the ATPG tool failing due to the internal nature of the scan [4] G. Aldrich and B. Cory, “Improving Test Quality and Reducing
enable signals of the flip-flops, we set the global scan enable Escapes,” in Proc. Fabless Forum, Fabless Semiconductor Assoc.,
pp. 34-35, 2003.
GSEN signal to 1 during test setup. Notice that the OR gate [5] X. Lin, R. Press, J. Rajski, P. Reuter, T. Rinderknecht, B. Swan-
in the LTG cell generates the local scan enable signal through son and N. Tamarapalli, “High-Frequency, At-Speed Scan Test-
a logical OR of the global scan enable and the Q output of the ing,” IEEE Design & Test of Computers, pp. 17-25, Sep-Oct 2003.
flop FF1 (see Figure 4). As a result, the DRC phase of the [6] K. Cheng, “Transition Fault Testing for Sequential Circuits,”
ATPG passes without any violations. In a commercial ATPG IEEE Transactions on Computer-Aided Design of Integrated Cir-
cuits and Systems, vol. 12, no. 12, pp. 1971-1983, Dec. 1993.
tool, the loss in coverage due to undetectability of stuck-at-1 [7] T. M. Mak, A. Krstic, K. Cheng, L. Wang, “New challenges in
faults on the scan enable signals can be recovered by declaring delay testing of nanometer, multigigahertz designs,” IEEE Design
the scan enable signal as a clock signal. Figure 10 shows the & Test of Computers, pp. 241-248, May-Jun 2004.
load unload procedure of the test protocol file. Each vector (V) [8] M. Bushnell, V. Agrawal, Essentials of Electronics Testing,
statement is a tester clock cycle. The waveform table (W) state- Kluwer Publishers, 2000.
[9] J. Saxena, K. M. Butler, J. Gatt, R. Raghuraman, S. P. Kumar, S.
ment determines the clock period of the tester clock cycle de- Basu, D. J. Campbell, J. Berech, “Scan-Based Transition Fault
fined in the timing waveform procedure of the test protocol file. Testing - Implementation and Low Cost Test Challenges,” in
The waveform table slow W FT is the slow shift clock cycle Proc. International Test Conference (ITC’02), pp. 1120 - 1129,
and f ast W FT is the at-speed clock cycle. The GSEN signal Oct. 2002.
is high till the (n-1)th shift and is made low in the dead cycle [10] M. Chen Chi; S. Huang; “A reliable clock tree design method-
ology for ASIC designs,” in Proc. International Symposium on
before the last shift which provides enough time for GSEN to Quality Electronic Design (ISQED’00), pp. 269-274, 2000.
go low. The last shift is done at-speed and it can be seen that [11] S. Wang, X. Liu, S.T. Chakradhar, “Hybrid Delay Scan: A Low
the waveform timing is changed to f ast W FT before the last Hardware Overhead Scan-Based Delay Test Technique for High
V statement. Fault Coverage and Compact Test Sets,” in Proc. Design, Automa-
We believe that the LTG-based solution has the following ad- tion and Test in Europe (DATE’03), pp. 1296-1301, 2004.
[12] Synopsys Application Note, “Tutorial on Pipelining Scan En-
vantages. The technique can be practiced using existing com- ables”
mercial tools for DFT insertion and ATPG. The technique is ap- [13] B.W. Kernighan and S. Lin, “An efficient heuristic procedure for
plicable to all scan-based DFT techniques, including Determin- partitioning graphs,” Bell System Tech. Journal, vol. 49, pp. 291-
istic BIST techniques that are based on scan [16]. The method 307, 1970.
is least intrusive and can be incorporated in existing physical [14] Tetramax ATPG Users Guide, “SYNOPSYS Toolset Version
2004.06,” Synopsys, Inc., 2004.
design flows. In our experiments, we demonstrated the use of [15] Synopsys DFT Compiler, “User Manual for SYNOPSYS Toolset
the technique using DFT Compiler and TetraMAX tools from Version 2004.06,” Synopsys, Inc., 2004.
Synopsys. The area overhead and impact on the functional tim- [16] Synopsys SoCBIST DBIST Users Guide, “User Manual for
ing due to inclusion of LTG cells is negligible. Finally, the SYNOPSYS Toolset Version 2004.06,” Synopsys, Inc., 2004.