11/11/2014 Overview ECE 553: TESTING AND • Definition • Ad-hoc methods TESTABLE DESIGN OF • Scan design DIGITAL SYSTEMS DIGITAL SYSTEMS – Design rules D i l – Scan register – Scan flip-flops – Scan test sequences – Overhead Design for Testability (DFT) - 1 – Scan design system • Summary 11/11/2014 2 Definition Ad-Hoc DFT Methods • Good design practices learned through • Design for testability (DFT) refers to those design experience are used as guidelines: techniques that make test generation and test – Don’t-s and Do-s application cost-effective. • Avoid asynchronous (unclocked) feedback. • DFT methods for digital circuits: • DFT methods for digital circuits: • Avoid delay dependant logic. – Ad-hoc methods • Avoid parallel drivers. – Structured methods: • Avoid monostables and self-resetting logic. • Scan • Avoid gated clocks. • Partial Scan • Built-in self-test (BIST) • Avoid redundant gates. • Boundary scan • Avoid high fanin fanout combinations. 11/11/2014 3 11/11/2014 4 Ad-Hoc DFT Methods Ad-Hoc DFT Methods • Good design practices learnt through experience • Design reviews are used as guidelines: – Manual analysis – Don’t-s and Do-s (contd.) • Conducted by experts. – Programmed analysis g y • Make flip-flops initializable. • Separate digital and analog circuits. • Using design auditing tools • Provide test control for difficult-to-control signals. – Programmed enforcement • Buses can be useful and make life easier. • Must use certain design practices and cell types. • Limit gate fanin and fanout. • Objective: Adherence to design guidelines and • Consider ATE requirements (tristates, etc.) testability improvement techniques. 11/11/2014 5 11/11/2014 6 1
11/11/2014 Ad-Hoc DFT Methods Scan Design • Disadvantages of ad-hoc DFT methods: – Objectives • Experts and tools not always available. • Simple read/write access to all or subset of storage • Test generation is often manual with no guarantee of elements in a design. high fault coverage. • Direct control of storage elements to an arbitrary value • Direct control of storage elements to an arbitrary value • Design iterations may be necessary. (0 or 1). • Direct observation of the state of storage elements and hence the internal state of the circuit. Key is – Enhanced controllability and observability. 11/11/2014 7 11/11/2014 8 Scan Design Scan Design Rules – Circuit is designed using pre-specified design rules. – Test structure (hardware) is added to the verified design: • Use only clocked D-type flip-flops for all state • Add one (or more) test control (TC) primary input. variables. • Replace flip-flops by scan flip-flops and connect to form one or more shift registers in the test mode. registers in the test mode. • At least one PI pin must be available for test; more • At least one PI pin must be available for test; more • Make input/output of each scan shift register controllable/observable from pins, if available, can be used. PI/PO. • All clocks must be controlled from PIs. – Use combinational ATPG to obtain tests for all testable faults in the combinational logic. • Clocks must not feed data inputs of flip-flops. – Add shift register tests and convert ATPG tests into scan sequences for use in manufacturing test. 11/11/2014 9 11/11/2014 10 Scan Flip-Flop (master-slave) Correcting a Rule Violation Master latch Slave latch D • All clocks must be controlled from PIs. TC Q Logic Comb. overhead logic D1 Q MUX Q SD Comb. FF D2 logic logic CK CK D flip-flop Comb. Master open Slave open CK logic t Q D1 Comb. FF Normal mode, D selected Scan mode, SD selected D2 TC logic t CK 11/11/2014 11 11/11/2014 12 2
11/11/2014 Level-Sensitive Scan-Design Latch Adding Scan Structure (LSSD) Master latch Slave latch PI PO D Q SFF SCANOUT Combinational MCK Q logic logic SFF SFF SCK D flip-flop SFF SD Normal MCK mode Logic TCK overhead MCK TCK Scan mode Not show n: CK or TC or TCK TCK MCK/SCK feed all SCANIN SFFs (scan Flip- SCK t flops). 11/11/2014 13 11/11/2014 14 Comb. Test Vectors Comb. Test Vectors Don’t care or random PI I1 I2 bits SCANIN S1 S2 PI I1 I2 O2 PO O1 TC 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 Combinational SCANIN SCANOUT TC logic PO O1 O2 Next Presen S2 N1 N2 S1 state t SCANOUT N1 N2 state Sequence length = ( n sff + 1) n comb + n sff clock periods n comb = number of combinational vectors n sff = number of scan flip-flops 11/11/2014 15 11/11/2014 16 Multiple Scan Registers Testing Scan Register • Scan flip-flops can be distributed among any • Scan register must be tested prior to application of scan test number of shift registers, each having a separate sequences. scanin and scanout pin. • A shift sequence 00110011 . . . of length n sff +4 in scan • Test sequence length is determined by the longest mode (TC=0) produces 00, 01, 11 and 10 transitions in all flip-flops and observes the result at SCANOUT output. scan shift register. g • Total scan test length: • Just one test control (TC) pin is essential. PI/SCANIN PO/ Combinational (( n sff + 1) n comb + n sff ) + ( n sff + 4) clock periods . SCANOUT M logic U ( n comb + 2) n sff + n comb + 4 clock periods . SFF X • Example: 2,000 scan flip-flops, 500 comb. vectors, total SFF scan test length ~ 10 6 clocks. SFF • Multiple scan registers reduce test length. TC CK 11/11/2014 17 11/11/2014 18 3
11/11/2014 Hierarchical Scan Scan Overhead • Scan flip-flops are chained within subnetworks • IO pins: One pin necessary. before chaining subnetworks. • Area overhead: • Advantages: – Gate overhead = [4 n sff /( n g +10 n ff )] x 100%, • Automatic scan insertion in netlist where n g = comb. gates ; n ff = flip-flops ; • Circuit hierarchy preserved – helps in debugging and design • Example – n g = 100k gates, n ff = 2k flip-flops , overhead = 6.7%. p g , f p f p , changes g g ff – More accurate estimate must consider scan wiring and layout • Disadvantage: Non-optimum chip layout. area. Scanin Scanout • Performance overhead: SFF1 SFF4 SFF1 SFF3 – Multiplexer delay added in combinational path; approx. two Scanin gate-delays. Scanout SFF2 SFF3 – Flip-flop output loading due to one additional fanout; approx. SFF4 SFF2 5-6%. Hierarchical netlist Flat layout 11/11/2014 19 11/11/2014 20 Optimum Scan Layout Scan Area Overhead X’ Linear dimensions of active area: X X = (C + S) / r SFF IO y = track dimension, w ire X’ = (C + S + α S) / r cell pad w idth+separation Y’ = Y + ry = Y + Y(1-- β ) / T C = total comb. cell w idth S = total non-scan FF cell SCANIN Flip- Area overhead w idth flop s = fractional FF cell area X Y X’Y’--XY XY cell ll = S/(C+S) = -------------- x 100% Y Y’ α = SFF cell w idth fractional XY increase 1-- β r = number of cell row s TC SCAN or routing channels = [ (1+ α s) ( 1+ ------- ) – 1 ] x 100% OUT β = routing fraction in active T area T = cell height in track dimension y 1-- β Routing channels = ( α s + ------- ) x 100% Active areas: XY and X’Y’ Interconnects T 11/11/2014 21 11/11/2014 22 Example: Scan Layout ATPG Example: S5378 • 2,000-gate CMOS chip • Fractional area under flip-flop cells, s = 0.478 Original Full-scan • Scan flip-flop (SFF) cell width increase, α α = 0.25 • Routing area fraction, β β = 0.471 Number of combinational gates 2,781 2,781 Number of non-scan flip-flops (10 gates each) 179 0 • Cell height in routing tracks, T = 10 Number of scan flip-flops (14 gates each) 0 179 Gate overhead 0.0% 15.66% • Calculated overhead = 17 24% • Calculated overhead = 17.24% Number of faults 4,603 4,603 • Actual measured data: PI/PO for ATPG 35/49 214/228 70.0% 99.1% Fault coverage 70.9% 100.0% Fault efficiency Scan implementation Area overhead Normalized clock rate ______________________________________________________________________ Number of ATPG vectors 414 585 Scan sequence length 414 105,662 None 0.0 1.00 Hierarchical 16.93% 0.87 Optimum layout 11.90% 0.91 11/11/2014 23 11/11/2014 24 4
Recommend
More recommend