Top Down Modeling and Test Bench Development: Verification Case Study: Pipeline ADC

Top Down Modeling and Test Bench 10/8/2002
Development: A Verificaton Case

Study of a Pipeline ADC
Top Down Modeling and

Test Bench Development
Verification Case Study: Pipeline ADC
2002 IEEE International Workshop on

Behavioral Modeling and Simulation
October 8, 2002
Jonathan David – Mixed Signal Methodology – Cadence

TOP DOWN DESIGN

Still the future of Mixed Signal Design?
• Theoretical Approach
– Actually in use, Matlab/SPW -> Spec -> Designers
• Bottom Up Design is still Powerful
– Circuit Knowledge + Creativity = New Approaches
BUT
• Practical Mixed Signal Simulation + Design Reuse +
• Decent Verification Environments =
• Top Down Mixed Signal System Verification
– Starting EARLY in the design process
– Gives Team Higher Visibility into Design Status


Behavioral Modeling and Simulation 1
A Methodology for Top Down Verification
• Specification -> Test Plan • ADC Specs

– Quick Ramp Test – Input Range
– INL & DNL via Binning – Supplies and Biasing
– ENOB via Nyquist Rate Sine Test – Clocking & Timing
• Pin Accurate Model – Output Drive
– Define all known I/O – Linearity
– Add behavior later – Noise & Distortion
• Start Test Development &
Architecture
– Calc DNL INL from Bins
– Calc ENOB from FFT of Sine

6b 1.5b/stage 4 stage Pipeline ADC.
• 3v supplies
• Differential Signal path .25 – 2.25v
• Vcm = 1.25
• Digital Error Correction


Trap to Avoid
Early Simulation Results
aren’t the same thing as a
Top-Down Verification
program!

First Order Model

• Delays
– clk –> samp & clk -> out always @(posedge clk)
• ENOB – Noise #dT sclk = !sclk;
• INL-DNL - Random Offsets
always @(posedge sclk) begin

// add the noise
vn = vnoise * ($dist_normal(seedn,0,1000)%6000) / 1000.0;
vnr = vnref * ($dist_normal(seedr,0,1000)%6000) / 1000.0;
// get the input and reference
vin = V(inp,inn) + vn;
vref = V(refmax,refmin) + vnr;
end

First Order Model

• Real to Bits
– Store sequence to model latency
always @(negedge sclk) begin

#td code = code1;
// set the overflow bits
overflow = of1;
underflow = uf1;
code1 = code2; of1 = of2; uf1 = uf2;
code2 = code3; of2 = of3; uf2 = uf3;
of3 = vin > vref;
uf3 = vin < -vref;
codeval = (vin/vref/2.0)*(fullscale) + fullscale/2 -0.5;
// internal storage in 2's complement
code3 = ( codeval>=(fullscale-1) ? fullscale-1 :
( codeval<=0 ? 0 : codeval ));
end
Functional Test Details
• Delays
– Compare outputs at clock edge and after MAX Tp spec
– Mismatch at other times indicates wrong prop delays in models
• Input Range
– Simple Low to High Sweep – to hit all Codes
– Mismatch in output codes indicates wrong input model/ sampling delay


Test bench Schematic

Compare Block Details

• Instantaneous Comparison (SimVision Compare Function is Better)
assign overCMP = overDUT^overGLD ;
assign undrCMP = undrDUT^undrGLD ;
assign dataCMP = dataDUT-dataGLD ; // difference better for the data
• Open a Log File (include date & time in name – Thx to Jon Brenner)
initial begin
$system("rm now.txt; date +%Y%m%d_%H%M > now.txt "); // update datestring
datefile = $fopen("now.txt","r"); // open the date for read
rstat = $fscanf(datefile,"%s",datestring );
$fclose(datefile);
filestring = {filename, datestring,".dat"};
fileid = $ fopen(filestring );
$fstrobe(fileid,
"ADC Output Data Comparison File : Testblock %M : date %s.%s.%s:%s",
datestring[8*13:8*9+1],datestring[8*9:8*7+1],datestring [8*7:8*5+1],
datestring[8*4:1]);
$fstrobe(fileid, "Time Delta : O ## U (DUT) sb= (GLD) O ## U");


Record Delta Warnings & Failures to File & Log

always @(clk) begin
#Td GLD = dataGLD+overGLD-undrGLD ;
DUT = dataDUT+overDUT-undrDUT ;
CMP = GLD - DUT;
if (enable) begin
$fstrobe(fileid, "%t %d : %b %d %b (%d) sb= (%d) %b %d %b ",
$realtime , CMP, overDUT , dataDUT, undrDUT, DUT,
GLD, overGLD, dataGLD , undrGLD);
if ( ( dataDUT !== dataGLD )
|| ( overDUT !== overGLD )
|| ( undrDUT !== undrGLD ) ) begin
if ( CMP >= -TolLsb && CMP <= TolLsb ) begin // specwarn
$strobe("SPECWARN: %t ADC DUT(%d ) != ADC GLD(%d )",
$realtime, DUT, GLD);
$fstrobe(fileid, "SPECWARN: %t ADC DUT(%d) != ADC GLD(%d )",
$realtime, DUT, GLD);
end else begin // specfail will also fail on any X values
$strobe("SPECFAIL: %t ADC code Delta=%d > spec=%d",
$realtime, CMP, TolLsb );
$fstrobe(fileid, "SPECFAIL: %t ADC code Delta=%d > spec=%d",
$realtime, CMP, TolLsb );
Functional Results – Ideal vs Wreal model


Ramp Test Details
• INL & DNL Determination

– Classic method : Determine transition points exactly
– AutoTest method: Take many Samples, Use Histogram for DNL
– Need to correct counts based on input wave type unless Ramps are used
• Second method with Ramp Source is present Solution

– Could be adjusted for Sinusoidal input fairly easily
– Warn User if endpoints / out-of-range values hit!
• Use endpoint bins?
– Only if over-range and under-range indicator
– To separate Out-of-Range values from valid measurements

Test Bench for INL DNL tests


DNL test – Initialize the bins
always @(posedge strt_fnsh) begin

for (i = 0; i<= maxcode; i = i+1) begin
bins[i] = 0; dnl[i] = 0; inl[i] = 0;
end
dnlmax = 0;
inlmax = 0;
totcount = 0;
counting = 1;
sum = 0;
$fdisplay(datafile,
" counts DNL INL ");
end

Count Each Code
always @(posedge clk) begin

if (counting && ((^data) !== 1'bx) && ((^data) !== 1'bz)
&& !under && !over ) begin
bins[data] = bins[data] +1;
// debug !!!
bintest = bins[data];
// debug !!!
totcount = totcount+1;
sum = sum + data;
end
end


Do the Math!
always @(negedge strt_fnsh ) begin
counting = 0;
if (totcount > 0) begin
idealbin = totcount/numcodes; // numcodes is real
for (j = 0; j<= maxcode; j = j+1) begin
dnl[j] = (bins[j]/idealbin) - 1.0;
if (j>0) inl[j ] = inl[j-1] +dnl[j ];
else inl[j ] = dnl[j];
$fdisplay(inlfile, "%10.3g # bin %d",inl[j], j ); // for SPW plotting
$fdisplay(dnlfile, "%10.3g # bin %d",dnl[j], j ); // for SPW plotting
$fdisplay(datafile, "%d %d %10.3g %10.3g ", j, bins[j], dnl[j ], inl[j] );
if (dnl[j] > dnlmax) dnlmax = dnl[j ];
else if (-dnl[j] > dnlmax ) dnlmax = -dnl[j];
if (inl[j] > inlmax) inlmax = inl[j];
else if (-inl[j] > inlmax ) inlmax = -inl[j];
end
$fstrobe( datafile, "\n max abs dnl: %10.3g inl: %10.3g ", dnlmax, inlmax);
$strobe( "\n %s max abs dnl: %10.3g inl: %10.3g ", filename, dnlmax, inlmax);
end
end

DNL & INL Plots


Sine Test Details

• Sample input ˜ Nyquist Rate
• Integer # Cycles in 2 N samples
• NOT Subharmonic of Sample Rate
• -> Need Prime Number ˜ 2N-1
• Input cannot oversaturate codes
– (no “over” or “under” allowed)
• No Harmonics (can’t measure THD)
• SNR = sample(dB) / RSS noise(dB)
• FFT Methods
– package FFTW routine for VPI
– Matlab
– SPW <- Cadence tool

Test Bench for Sine Test


SPW FFT Plot For 2nd Order Model

S2=fft_transform(S1,
fftlen=4096,
num_frames=1, SNR = 20log10(Fund/noise) = 37.222 dB
overlap=0, ENOB = (SNR-1.76)/6.02 = 5.89
window=rectangular,
rm_mean,
half_spectrum,
normalize
S3= abs(
conjugate(S2)*S2)
S4=sqrt(S3)
Fund = Peak(S4)=
sqrt(S3[2039])
Noise =
Sqrt(Sum(S3[0:2038,
2040:2049))

Signal Sink – Formatted for SPW

initial begin
//datestring function was here
filestring = {libpath,"/",SignalName, datestring,".", viewname};
fileid = $fopen(filestring);
$fstrobe(fileid, "$SIGNAL_FILE 9" );
$fstrobe(fileid, "$USER_COMMENT" );
$fstrobe(fileid, "Output Data File for ADC %M");
$fstrobe(fileid, "$COMMON_INFO");
$fstrobe(fileid, "SPW Version = 4.81");
$fstrobe(fileid, "System Type = solaris2");
$fstrobe(fileid, "Sampling Frequency = %d",SampRate);
$fstrobe(fileid, "Starting Time = 0");
$fstrobe(fileid, "$DATA_INFO");
$fstrobe(fileid, "Number of points = %d", NumPoints);
$fstrobe(fileid, "Signal Type = Integer");
$fstrobe(fileid, "$DATA ASCII");
$timeformat(-9,," ns",20);
count = 0;
end


Signal Sink – Formatted for SPW

always @(posedge clk) begin
if (((^data) !== 1'bx) && ((^data) !== 1'bz)) decimal_value = data;
if (enable&& (count<NumPoints)) begin
$fstrobe(fileid, "%d # %t ", data, $realtime);
count = count +1;
end
end
//--------------------------------------------------------------
analog begin
@(initial_step) begin
outfile = $fopen("SampleInfo%I.%M.%T.dat");
$fstrobe(outfile, "# Output Data File for ADC %M");
$fstrobe(outfile, "# Time Sample");
$fstrobe(outfile, "%20.15e %d",$abstime, decimal_value);
end
@(posedge clk) begin
if (enable) $fstrobe(outfile, "%20.15e %d",
$abstime, decimal_value);
end
@(final_step) $fclose(outfile);
end
Second Order Model
• Offsets/GainErrors that match Arch.

• Can expand First Order Model
• Or use first Order Models of SubBlocks
• Sub-Block Models:
– SAH
– 1.5b Pipeline Stage
– 2 bit Flash
– Digital correction Block


SAH

SampHold.vams
real posval, negval ;
// Analog Values in Discrete time events
wreal res_p = posval; wreal res_m = negval;
// initialize the variables in Initial Block
always @(negedge clk ) begin
sampval = V(ain_p , ain_n);
posval = V(cmref) + gain*(sampval+vos+vn)/2;
if (posval > vhi) posval = vhi;
else if (posval < vlo ) posval = vlo;
negval = V(cmref) - gain*(sampval+vos+vn)/2;
if (negval > vhi) negval = vhi;
else if (negval < vlo ) negval = vlo;
end
always @(posedge clk ) begin
sampval = 0;
posval = V(cmref);
negval = V(cmref);
end
• Models Fixed Gain and Offset errors + Gaussian Noise @ Sample rate


1.5b Pipeline Stage

Transfer Function 10 01
00 10
100 100
Residue
Vin
00 01 10


PLADC_1r5b_stage.vams
real res_pl , res_mi; wreal res_p , res_m; wreal ain_p , ain_n ;
assign res_p = res_pl; assign res_m = res_mi;
always @(posedge clk ) begin // sample the input
#(td/1n) code = 2'bx; // set to unknown until other edge of clock
res_pl = V(cmref);
res_mi = V(cmref);
end
always @(negedge clk ) begin // evaluate and drive the outputs
valin = ain_p - ain_n;
refin = V(adcrefp , adcrefn);
#(td/1n) code = 1+((valin+vospcomp)>refin)-((valin+vosncomp)<-refin);
#(td/1n) resout = (valin+vosamp+vn )*Cgain + (1.0-
code)*V(dacrefp,dacrefn );
res_pl = V(cmref)+0.5*resout;
res_mi = V(cmref) -0.5*resout;
end

Flash Stage


PLADC_flash.vams
wreal ain_p, ain_n;

logic clk;
output overflow, underflow; reg overflow, underflow;
output [1:0] code; reg [1:0] code;
always @(posedge clk) begin // sample the input
//valin = ain_p - ain_n;
//refin = V(adcrefp, adcrefn);
#(td/1n) code = 2'bx; // set to unknown
end
always @(negedge clk) begin // evaluate and drive the outputs
valin = ain_p - ain_n;
refin = V(adcrefp, adcrefn);
#(td/1n) code = 1+(valin>0)+(valin>refin)-(valin<-refin);
overflow = valin>V(dacrefp,dacrefn);
underflow = valin<-V(dacrefp,dacrefn);
end

Digital Correction Block


PLADC_DigCorr4.vams
always @(posedge PHIodd) begin
#Td // just the even bits
St2a = St2; St4a = St4; code = Sum;
overflow = Over || (OF &&
St1c[1] && St2c[1] && St3b[1] && St4b[1] && St5a[1] &&
St5a[0]);
underflow = UF && !St1c && !St2c && !St3b && !St4b && !St5a ;
end
always @(posedge PHIeven) begin
#Td // need to do this in order,
// or use non-blocking with the same delay?
St1c = St1b; St1b = St1a; St1a = St1; //1
St2c = St2b; St2b = St2a; //2
St3b = St3a; St3a = St3; //3
St4b = St4a; St5a = St5; //4 & 5
Sum = St5a + (St4b<<1) + (St3b<<2) + (St2c<<3) + (St1c<<4);
// these are clocked and EVALUATED on other Edge
Over = Sum[6]; OF = OFlast; UF = UFlast;
end

Third Order Models – Behavioral Models of

Analog Building Blocks
• Switch • 2nd order models allow allocation of
• Opamp gain error, offset and noise specs
• Comparator • Simulation Is FASTER than electrical
but slower than Matlab. WHY repeat
• Clock Generation this?
• Confirm Matlab conclusions in a
TEST BENCH compatible with
Extracted simulation
• Other Specs (loading driving, Non
ideal Opamp vs Cap Mismatch)
cannot be separated out at 2 nd order
level.


Sw_no.vams
// log Cubic Spline Transition
analog function real lcubefn ;
input x,K; real x,K;
lcubefn = (x<=0)?1:(x>=1)?K: pow(K,(3 -2*x)*x*x);
endfunction
initial Control = 0;
always @( posedge control) Control = 1;
always @( negedge control) Control = 0;
analog begin
if (Control == 1) swres = 0.0; // on means R is minimum
else swres = 1.0; // off means R is maximum
end
@(posedge Control) swres = 0.0; // on means R is minimum
@(negedge Control) swres = 1.0; // off means R is maximum
// RoutExponent calculated from a transistion function
rsmooth = transition(swres, tdelay,trise,tfall);
rout = ron*lcubefn(rsmooth,roff/ron );
V(vin,vout) <+ I(vin,vout )*rout;
end

DiffOpamp.va – Start with ModelWriter,

Add vcm, outn
analog begin
@(initial_step) begin // by default ALL analyses included (446+)
r1 = gain; gm_nom = 1.0;
c1 = 1/(`M_TWO_PI * pole_freq * gain); r_rout = rout;
end
vin_val= V(vin_p, vin_n) + vin_offset;
// ------ Vref is at Virtual Ground
V(vref, vspply_n) <+ 0.5*V(vspply_p,vspply_n);
// ------ Input Stage
I(vin_p, vin_n) <+ vin_val / rin;
I(vref, vin_p) <+ ibias; I(vref, vin_n) <+ ibias;
// ------ GM stage
I(vref, coutp) <+ gm_nom*vin_val ; I(vref, coutn) <+ -gm_nom*vin_val ;
// ------ Dominant Pole.
I(coutp, vref) <+ 2*c1*ddt(V(coutp, vref)); I(coutp, vref) <+ 2*V(coutp, vref)/r1;
I(coutn, vref) <+ 2*c1*ddt(V(coutn, vref)); I(coutn, vref) <+ 2*V(coutn, vref)/r1;
// ------ Output Stage.
I(vref, vout_p) <+ 2*V(coutp, vref)/r_rout;
I(vout_p, vref) <+ 2*V(vout_p, vref)/r_rout;
I(vref, vout_n) <+ 2*V(coutn, vref)/r_rout;
I(vout_n, vref) <+ 2*V(vout_n, vref)/r_rout;
// ------ Soft Output Limiting.
if (V(vout_p) > (V(vspply_p) - vsoft)) I(coutp, vref) <+ gm_nom*(V(vout_p, vspply_p)+vsoft);
else if (V(vout_p) < (V(vspply_n) + vsoft)) I(coutp, vref) <+ gm_nom*(V(vout_p, vspply_n)-vsoft);
// ------ Soft Output Limiting.
if (V(vout_n) > (V(vspply_p) - vsoft)) I(coutn, vref) <+ gm_nom*(V(vout_n, vspply_p)+vsoft);
else if (V(vout_n) < (V(vspply_n) + vsoft)) I(coutn, vref) <+ gm_nom*(V(vout_n, vspply_n)-vsoft);
end


CompDiffLatched.vams
reg d, D;
assign D_ = !D;
analog begin
halfhys = hys/2.0;
Tplh = (td + trise/2)/1n; Tphl = (td + tfall/2)/1n;
end
vin =V(vin_p,vin_n ) - V(ref_p,ref_n ) + p_off + n_off ;
end
initial begin
TPlh = 1; TPhl = 1; // value will be corrected soon
#0.1 TPlh = Tplh ; TPhl = Tphl; // until analog initial_step
d = vin>0; // initialize the register
end
always @(cross(vin - halfhys , +1 )) if (enable) d = 1;
always @(cross(vin + halfhys , -1 )) if (enable) d = 0;
always @(posedge enable) begin
if ((vin < -halfhys)&&(d)) d = 0;
else if ((vin > halfhys)&&(!d)) d = 1;
end
always @(posedge d) # TPlh D = d;
CompLatched.vams is easier!
always @(negedge d) # TPhl D = d;

PLADC_StgClkGen.vams
initial begin
Y2 = 1;
Y1 = 0;
Y1a = 0;
Y1b = 0;
end
always @(posedge A) begin
#Tdhl2 Y2 = !A;
#Tdlh1 Y1 = A;
#Tdlh1a Y1a = A;
#Tdlh1b Y1b = A;
end
always @(negedge A) begin
#Tdhl1b Y1b = A;
#Tdhl1a Y1a = A;
#Tdhl1 Y1 = A;
#Tdlh2 Y2 = !A;
end


Characterizing The SubBlocks
• Gain Errors • A Design Reuse Method will aid this.

• Offsets • Scripts need to create datafiles
• Noise accessed by higher level models
• New file access functions allow
easier use of data between various
simulations

Variables are key to TB Flexibility


Summary
• Top Down Verification Methodology

• 1st order, Pin Accurate model defined
• Specifications and models to support those were defined
• 2nd Order models were developed, and assembled for more detail.
Retested and compared to original model
• 3rd Order model was developed from 1st order model of analog
building blocks. – Additional Specifications need to be Allocated to
make that part work again.
Questions?

References
• J. Doenberg, HS Lee, DA Hodges, “ Full-Speed Testing of A/D Converters” IEEE Journal of Solid-State
Circuits(1984), Vol. 19, No. 6, Dec. 1984, 820-827
• TE Linnenbrink, SJ Tilden, MT Miller, “ ADC Testing with IEEE-Std 1241-2000” Proceedings IEEE Instrumentation
and Measurement Conference 2001, 1986-1991
• G. Chiorboli, C. Morandi , “ ADC Modeling and Testing” Proceedings IEEE Instrumentation and Measurement
Conference 2001, 1992-1999
• Nyquist data converter testing and yield analysis using behavioral simulation
Liu, E.W.Y.; Sangiovanni -Vincentelli, A.L.
Computer-Aided Design, 1993. ICCAD -93. Digest of Technical Papers., 1993 IEEE/ACM International Conference on
, 1993 Page(s): 341 -348
• Influence of the architecture on ADC error modeling
Arpaia, P.; Daponte, P.; Michaeli, L.
Instrumentation and Measurement, IEEE Transactions on , Vol.48, Iss.5, 1999
Pages: 956- 966
• Metrological characterisation of analog-to-digital converters-a state of the art
Arpaia, P.; Cennamo, F.; Daponte , P.
Advanced A/D and D/A Conversion Techniques and Their Applications, 1999. Third International Conference on
(Conf. Publ. No. 466) , 1999
Page(s): 134 -144
• Some thoughts on sine wave ADC testing
Sugawara, H.; Kobayashi, H.; Arpaia, P.
Instrumentation and Measurement Technology Conference, 2000. IMTC 2000. Proceedings of the 17th IEEE , 2000
Page(s): 125 -130 vol.1
• J. David, “ Functional Verification Of A Differential Operational Amplifier” International Cadence User Group,
2001


More References
• Behavioral simulation of a 3-bit flash ADC
Mantooth, H.A.; Allen, P.E.
Circuits and Systems, 1990., IEEE International Symposium on , 1990
Page(s): 1356 -1359 vol.2
• Behavioral model of pipeline ADC by using SIMULINK(R)
Bilhan, E.; Estrada-Gutierrez, P.C.; Valero-Lopez, A.Y.; Maloberti, F.
Mixed-Signal Design, 2001. SSMSD. 2001 Southwest Symposium on , 2001 Page(s): 147 -151
• Behavioral modeling and simulation of data converters
Liu, E.; Gielen, G.; Chang, H.; Sangiovanni-Vincentelli, A.L.; Gray, P.R.
Circuits and Systems, 1992. ISCAS '92. Proceedings., 1992 IEEE International Symposium on ,
Volume: 5 , 1992
Page(s): 2144 -2147 vol.5
• IEEE Std 1241: the benefits and risks of ADC histogram testing
Max, S.
Instrumentation and Measurement Technology Conference, 2001. IMTC 2001. Proceedings of the
18th IEEE , Volume: 1 , 2001
Page(s): 704 -709 vol.1
• Fast accurate and complete ADC testing
Max, S.
Test Conference, 1989. Proceedings. Meeting the Tests of Time., International , 1989
Page(s): 111 -117
• FFT Testing of ADCs
Odom, Bill
Conference on Analog and Mixed-Signal Applications, 1998 Proceedings
Page(s): 201-203
• New High Speed Technique for Pipeline ADC Design
Sarraj, Maher
Conference on Analog and Mixed-Signal Applications, 1998 Proceedings
Page(s): 205-208
Bibliography
Don Lewis “Testing Operational Amplifiers” Electronics Test ( Benwill Publishing) January 1979
Don Lewis “Compensation of Linear IC Test Loops” Electronics Test ( Benwill Publishing) May 1979
David Johns, Ken Martin,”Analog Integrated Circuit Design”, Wiley & Sons, New York 1997, esp chapter 6.
Jacob Millman, “MicroElectronics: Digital and Analog Circuits and Systems” McGraw-Hill 1979
Paul Gray and Robert Meyer “Analysis and Design of Analog Integrated Circuits: 2nd Ed” Wiley & Sons 1984
C.F. Wojslaw & E.A. Moustakas “Operational Amplifiers” Wiley & Sons, NY, 1986
Dan Fitzpatrickj , Ira Miller, “Analog Behavioral Modeling with the Verilog -A Language” Kluwer, Boston, 1998
Samir Palnitkar , “ Verilog HDL” SunSoft, Mountain View, CA 1996
Ken Kundert, “The Designer’s Guide to Spice & Spectre”, Kluwer, Boston, 1995
J.E. Solomon. "The monolithic op amp: a tutorial study." IEEE Jo urnal of Solid-State Circuits(1974) SC -9.6 (Dec. 1974
(Special Issue on Analog Circuits)): 314-332. (Also published as Application Note AN -A from National
Semiconductor)
G.Ferri and W. Sansen “A Rail-toRail Constant-g m Low-Voltage CMOS Operational Transconductance Amplifier.” IEEE
Journal of Solid-State Circuits(1997), vol 32, October 1997 1563-1567.
M. Yamatke, “A Simplified Test-Set for Op Amp Characterization” National Semiconductor Application Note 24, April
1986



Top Down Modeling and Test Bench Development: Verification Case Study: Pipeline ADC

Uploaded by

Copyright:

Available Formats

Top Down Modeling and Test Bench Development: Verification Case Study: Pipeline ADC

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Top Down Modeling and Test Bench Development: Verification Case Study: Pipeline ADC

Uploaded by

Copyright:

Available Formats

Top Down Modeling and Test Bench 10/8/2002

Development: A Verificaton Case

Top Down Modeling and

2002 IEEE International Workshop on

2002 IEEE International Workshop on

TOP DOWN DESIGN

2002 IEEE International Workshop on

2002 IEEE International Workshop on

A Methodology for Top Down Verification

• Specification -> Test Plan • ADC Specs

2002 IEEE International Workshop on

6b 1.5b/stage 4 stage Pipeline ADC.

2002 IEEE International Workshop on

2002 IEEE International Workshop on

2002 IEEE International Workshop on

First Order Model

always @(posedge sclk) begin

2002 IEEE International Workshop on

First Order Model

always @(negedge sclk) begin

Functional Test Details

2002 IEEE International Workshop on

2002 IEEE International Workshop on

Test bench Schematic

2002 IEEE International Workshop on

Compare Block Details

2002 IEEE International Workshop on

2002 IEEE International Workshop on

Record Delta Warnings & Failures to File & Log

Functional Results – Ideal vs Wreal model

2002 IEEE International Workshop on

2002 IEEE International Workshop on

Ramp Test Details

• INL & DNL Determination

• Second method with Ramp Source is present Solution

2002 IEEE International Workshop on

Test Bench for INL DNL tests

2002 IEEE International Workshop on

2002 IEEE International Workshop on

DNL test – Initialize the bins

always @(posedge strt_fnsh) begin

2002 IEEE International Workshop on

Count Each Code

always @(posedge clk) begin

2002 IEEE International Workshop on

2002 IEEE International Workshop on

2002 IEEE International Workshop on

DNL & INL Plots

2002 IEEE International Workshop on

2002 IEEE International Workshop on

Sine Test Details

2002 IEEE International Workshop on

Test Bench for Sine Test

2002 IEEE International Workshop on

2002 IEEE International Workshop on

SPW FFT Plot For 2nd Order Model

2002 IEEE International Workshop on

Signal Sink – Formatted for SPW

2002 IEEE International Workshop on