measuring ENOB of a sigma delta ADC

[Ayman Essam]

hi
I would like to measure the ENOB of a sigma-delta ADC on Cadence Virtuoso.
what are the steps to be done ?
do i need MATLAB ?

 

[pancho_hideboo]

I would like to measure the ENOB of a sigma-delta ADC on Cadence Virtuoso.
what are the steps to be done ?
do i need MATLAB ?

No, you don't need MATLAB at all.

Simply use Skill Language in Cadence dfII.

ENOB=(SQNR-1.76)/6.02

Excerption from My Skill Code

Tstep = 1/fs

Tstart = round(5*(1/fIF)/Tstep)*Tstep + Tstep/4
Tstop = Tstart + Nfft*Tstep
fprintf(stdout, "Tstart=%.2fmsec\n", Tstart/1m)
fprintf(stdout, "Tstop=%.2fmsec\n", Tstop/1m)

delta_f = 1/(Tstop-Tstart)

fprintf(stdout, "delta_f=%.2fkHz\n", delta_f/1k)

tone_bin = int( pv("top-level", "tone_bin", ?result strcat(boke, "designParamVals-info")) )

fin = pv( "top-level", "fin", ?result strcat(boke, "designParamVals-info") )

/*
tone_bin = int(fIF / delta_f)
if( mod(tone_bin, 2) == 0 then
   if( tone_bin >= 0 then ++tone_bin else --tone_bin )
)
fin = tone_bin * delta_f
*/

tone_bin0 = mod(tone_bin, Nfft)
if( tone_bin0 < 0 then tone_bin0 = -tone_bin0 )
if( tone_bin0 >= Nfft / 2 then tone_bin0 = Nfft - tone_bin0 )
fin0 = tone_bin0 * delta_f

fprintf(stdout, "f0=%.2fkHz, fIF=%.2fkHz, tone_bin=%d, fin=%.7fkHz, fin0=%.7fkHz\n",
	f0/1k, float(fIF)/1k, tone_bin, fin/1k, fin0/1k)

NBW = 1.5 * (fs/Nfft)
fprintf(stdout, "NBW=%.3fkHz\n", NBW/1k)

V = getData( strcat(aho_path, "V") )

time_vec = drGetWaveformXVec( V )
tstop = drGetElem( time_vec, drVectorLength(time_vec)-1 )
fprintf(stdout, "tstop=%gmsec\n", tstop/1m)

fprintf(stderr, "length(V)=%d\n", drVectorLength(time_vec))

Tstop = int( (tstop-Tstep/4)/Tstep ) * Tstep + Tstep/4
Tstart = Tstop - Nfft*Tstep
fprintf(stdout, "Tstart=%.2fmsec\n", Tstart/1m)
fprintf(stdout, "Tstop=%.2fmsec\n", Tstop/1m)

V  = 2 * dft0( V, Tstart, Tstop, Nfft, "Hanning" ) / VFS
;;; Factor "2" of head is for compensation of "Hanning" windowing.

V = 2 * V ; Single-Sided

x_vec = drGetWaveformXVec(V)
Npts_V = drVectorLength( x_vec )
fprintf(stderr, "length(spec0)=%d=2^%d\n", Npts_V, int(log(Npts_V)/log(2)))

Smoothed_V = circ_smooth( mag(V)**2, 16 )
;Smoothed_V = moving_average( mag(V)**2, 3 )

V->plotStyle = 'joined ; could be set to nil too

Raw_Spectrum = db20(V)
Smoothed_Spectrum = db10(Smoothed_V)

 

[Ayman Essam]

No, you don't need MATLAB at all.

Simply use Skill Language in Cadence dfII.

ENOB=(SQNR-1.76)/6.02

Excerption from My Skill Code

Tstep = 1/fs

Tstart = round(5*(1/fIF)/Tstep)*Tstep + Tstep/4
Tstop = Tstart + Nfft*Tstep
fprintf(stdout, "Tstart=%.2fmsec\n", Tstart/1m)
fprintf(stdout, "Tstop=%.2fmsec\n", Tstop/1m)

delta_f = 1/(Tstop-Tstart)

fprintf(stdout, "delta_f=%.2fkHz\n", delta_f/1k)

tone_bin = int( pv("top-level", "tone_bin", ?result strcat(boke, "designParamVals-info")) )

fin = pv( "top-level", "fin", ?result strcat(boke, "designParamVals-info") )

/*
tone_bin = int(fIF / delta_f)
if( mod(tone_bin, 2) == 0 then
   if( tone_bin >= 0 then ++tone_bin else --tone_bin )
)
fin = tone_bin * delta_f
*/

tone_bin0 = mod(tone_bin, Nfft)
if( tone_bin0 < 0 then tone_bin0 = -tone_bin0 )
if( tone_bin0 >= Nfft / 2 then tone_bin0 = Nfft - tone_bin0 )
fin0 = tone_bin0 * delta_f

fprintf(stdout, "f0=%.2fkHz, fIF=%.2fkHz, tone_bin=%d, fin=%.7fkHz, fin0=%.7fkHz\n",
	f0/1k, float(fIF)/1k, tone_bin, fin/1k, fin0/1k)

NBW = 1.5 * (fs/Nfft)
fprintf(stdout, "NBW=%.3fkHz\n", NBW/1k)

V = getData( strcat(aho_path, "V") )

time_vec = drGetWaveformXVec( V )
tstop = drGetElem( time_vec, drVectorLength(time_vec)-1 )
fprintf(stdout, "tstop=%gmsec\n", tstop/1m)

fprintf(stderr, "length(V)=%d\n", drVectorLength(time_vec))

Tstop = int( (tstop-Tstep/4)/Tstep ) * Tstep + Tstep/4
Tstart = Tstop - Nfft*Tstep
fprintf(stdout, "Tstart=%.2fmsec\n", Tstart/1m)
fprintf(stdout, "Tstop=%.2fmsec\n", Tstop/1m)

V  = 2 * dft0( V, Tstart, Tstop, Nfft, "Hanning" ) / VFS
;;; Factor "2" of head is for compensation of "Hanning" windowing.

V = 2 * V ; Single-Sided

x_vec = drGetWaveformXVec(V)
Npts_V = drVectorLength( x_vec )
fprintf(stderr, "length(spec0)=%d=2^%d\n", Npts_V, int(log(Npts_V)/log(2)))

Smoothed_V = circ_smooth( mag(V)**2, 16 )
;Smoothed_V = moving_average( mag(V)**2, 3 )

V->plotStyle = 'joined ; could be set to nil too

Raw_Spectrum = db20(V)
Smoothed_Spectrum = db10(Smoothed_V)

ok thanks ... any good reference for Skill language on cadence ?

 

[pancho_hideboo]

ok thanks ... any good reference for Skill language on cadence ?

All documents are available in install directory of Cadence dfII.

**broken link removed**