Low power illinois scan architecture-1
( Chandra et al.)
●
Low power illinois scan architecture-2
(Ling et al.)
●
Some proposals for further improvement
The problem of Power Dissipation
in testing
●
●
●
●
Scan circuit generally causes excessive switching activity
compared to normal circuit operation. The higher switching
activity in turn causes higher peak power supply current which
results into supply voltage droop and eventually yield loss.
Power consumption during testing of scan circuit is an
important issue to address for today’s very complex circuits
when chips are being designed with small feature size and
higher frequency.
Excessive average power results into burn out of chip.
Excessive peak power results into power droop problem which
can falsely classify a good chip as a faulty chip.
Introduction to illinois scan
architecture
Ref-Prof.J.Patel's slides,”illinois Scan slides 6 per page”.
Illinois Scan with multiple pins
The scan chains which are compatible are grouped together
in one group.
Two Test Modes of Illinois Scan
Beyond Illinois Scan
Illinois Scan has achieved phenomenal success in
reducing Test data volume,Pin overhead & Test
application time.
It does not cater to the need of low power because
because like basic scan, all the scan flops and the
CUT are subjected to continuous switching activity
during shift and capture cycles.
Some architectures have been presented for achieving
low power along with reduced test data volume,pin
overhead & test application time.
Low power illinois scan architecture-1
Ref--“Low Power Illinois Scan Architecture for Simultaneous Power and Test Data
Volume Reduction “Anshuman Chandra, Felix Ng, Rohit Kapur, In Pro. DATE, IEEE,
2008. 462-467
Procedure for one scan-in & scan-out cycle
●
●
●
●
Set LPWR sel = 1
Scan-in data into the Reference chain and the shared chains
for (n-1)b shifts while scanning out values from all the chains.
Set LPWR sel = 0
Scan-in data in the scan blocks of the shared chains from the
previous scan block in the Reference chain for the last b
shifts. For the first scan block, scan in data into Reference
chain and shared chains from the SI.
●
Set the primary input values and the scan enable
●
Apply the capture clock to capture the response of the CUT
Response Captured in scan flops
In this scheme,scan block is of size 'b' and number of scan blocks
per chain is 'n', thus the shift cycle length is equal to 'nb' .
State of scan chains after '(n-1)b' shifts
State of scan chains after scan in
shift for 'nb' shifts
Discussion
●
●
While the reference chain is updated, in this
configuration the shared chains receive
constant data from the pseudo scan input that
is connected to the ground and the first n-1
scan blocks in the shared chains are filled with
value 0.
The additional multiplexers added to the design
increase the load on the last flop of each scan
block of shared chains.
Low power illinois scan architecture-2
●
●
It is an extension of the previous scheme with
the added feature of scan cell reordering.
It divides scan chains into several scan chain
segments first and then putting the similar scan
cells in adjacent positions to reduce dynamic
power dissipation during test data shift in
operation.
Ref--“A low power broadcast scan scheme”,Zhang Ling, Yan Bowu and Li
Tonghan,Journal of Chemical and Pharmaceutical Research, 2014,
6(7):2158-2162
Identifying which Scan cell columns to be grouped
together?
COMPARATIVE STUDY OF 3 IMPLEMENTATIONS OF
ILLINOIS SCAN
●
Original Implementation of Illinois scan
COMPARATIVE STUDY OF 3 IMPLEMENTATIONS OF
ILLINOIS SCAN
●
Implementation using low power illinois scan
architecture-1
COMPARATIVE STUDY OF 3 IMPLEMENTATIONS OF
ILLINOIS SCAN
●
Implementation using scan cell reordering of low power illinois
scan architecture-1
Discussion
●
●
Compared with conventional broadcast scan this
scheme enormously reduces scan-in shift power.
The low power scan with scan cell reordering may
result in a slight longer routing length. However, this
technique only makes modification for the adjacent
two segments, so the routing length would not be
very long.
Some proposals for further
improvement
●
●
In the first architecture,experiment can be
carried out for a different value of number of
scan blocks per chain(n) and scan block
size(b).
Also in this architecture, instead of supplying all
0's as pseudo scan input,all 1's may be applied
as pseudo scan input.
Some proposals for further
improvement
●
●
In the second architecture instead of reordering
over adjacent blocks, reordering can be done
over the entire row (single scan chain).
There is no theoretical basis for this proposed
reordering.A systematic formulation based on
graph theory may help achieve higher
efficiency.