luke
/
ngspice
mirror of https://git.code.sf.net/p/ngspice/ngspice
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Implementing mobility mudulation (theta).Analytical derivatives for weak-inversion branch.

This commit is contained in:
dwarning 2018-11-28 20:26:50 +01:00 committed by Holger Vogt
parent 8b424dd3d3
commit 71524f6060
1 changed files with 67 additions and 71 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

138
src/spicelib/devices/vdmos/vdmosload.c
View File

@ -15,9 +15,7 @@ VDMOS: 2018 Holger Vogt
#include "ngspice/suffix.h"
static double
cweakinv(double sl, double shift, double vgst, double vds, double lambda, double beta, double vt, double mtr);
static double
cweakinv2(double sl, double shift, double vgst, double vds, double lambda, double beta, double vt, double mtr);
cweakinv2(double sl, double shift, double vgst, double vds, double lambda, double beta, double vt, double mtr, double theta);
int
@ -282,59 +280,83 @@ VDMOSload(GENmodel *inModel, CKTcircuit *ckt)
*/
double betap;
double vgst;
double onfg, fgate, Betam, dfgdvg;
von = (model->VDMOSvt0*model->VDMOStype);
vgst = (here->VDMOSmode == 1 ? vgs : vgd) - von;
vdsat = MAX(vgst, 0);
onfg = 1.0+model->VDMOStheta*vgst;
fgate = 1.0/onfg;
Betam = Beta * fgate;
dfgdvg = -model->VDMOStheta*fgate*fgate;
/* drain current including subthreshold current
* numerical differentiation for gd and gm with a delta of 2 mV */
if (model->VDMOSksubthresGiven && (here->VDMOSmode == 1)) {
double delta = 0.001;
cdrain = cweakinv(model->VDMOSksubthres, model->VDMOSsubshift, vgst, vds, model->VDMOSlambda,
Betam, vt, model->VDMOSmtr);
/* gd */
double vds1 = vds + delta;
double cdrp = cweakinv(model->VDMOSksubthres, model->VDMOSsubshift, vgst, vds1, model->VDMOSlambda,
Betam, vt, model->VDMOSmtr);
vds1 = vds - delta;
double cdrm = cweakinv(model->VDMOSksubthres, model->VDMOSsubshift, vgst, vds1, model->VDMOSlambda,
Betam, vt, model->VDMOSmtr);
here->VDMOSgds = (cdrp - cdrm) / (2. * delta);
/* gm */
double vgst1 = vgst + delta;
cdrp = cweakinv(model->VDMOSksubthres, model->VDMOSsubshift, vgst1, vds, model->VDMOSlambda,
Betam, vt, model->VDMOSmtr);
vgst1 = vgst - delta;
cdrm = cweakinv(model->VDMOSksubthres, model->VDMOSsubshift, vgst1, vds, model->VDMOSlambda,
Betam, vt, model->VDMOSmtr);
here->VDMOSgm = (cdrp - cdrm) / (2. * delta) * fgate + dfgdvg * cdrain;
if (model->VDMOSksubthresGiven) {
/* Alternative simple weak inversion model, according to https://www.anasoft.co.uk/MOS1Model.htm
* Scale the voltage overdrive vgst logarithmically in weak inversion.
* Best fits LTSPICE curves with shift=0
* Drain current including subthreshold current */
double t0, t1, t2, t3, t4, t5, t6;
double slope = model->VDMOSksubthres;
double lambda = model->VDMOSlambda;
double theta = model->VDMOStheta;
double shift = model->VDMOSsubshift;
double mtr = model->VDMOSmtr;
/* scale vds with mtr (except with lambda) */
double vdss = vds*mtr*here->VDMOSmode;
t0 = 1 + lambda*vds*here->VDMOSmode;
t1 = 1 + theta*vgs;
betap = Beta*t0/t1;
vgst = slope * log(1 + exp((vgst - shift) / slope));
if (vgst <= vdss) {
/* saturation region */
cdrain = betap*vgst*vgst*0.5;
t2 = exp((vgst-shift)/slope);
t3 = Beta*slope*t2*log(t2+1)*t0/(t1*(t2+1));
t4 = Beta*slope*slope*theta*log((t2+1)*(t2+1))*t0/(2*t1*t1);
here->VDMOSgm = t3-t4;
t3 = Beta*slope*slope*log((t2+1)*(t2+1))*lambda;
t4 = 2*t1;
here->VDMOSgds = t3/t4;
}
else {
/* linear region */
cdrain = betap * vdss * (vgst - 0.5 * vdss);
t3 = exp((vgst-shift)/slope);
t4 = Beta*vdss*t3*t0/(t1*(t3+1));
t5 = Beta*vdss*theta*(slope*log(t3+1)-0.5*vdss)*t0/(t1*t1);
here->VDMOSgm = t4-t5;
t4 = Beta*mtr*(slope*log(t3+1)-0.5*vdss)*t0/t1;
t5 = Beta*0.5*mtr*vdss*t0/t1;
t6 = Beta*vdss*(slope*log(t3+1)-0.5*vdss)*lambda/t1;
here->VDMOSgds = t4-t5+t6;
}
}
else if (model->VDMOSsubslGiven && (here->VDMOSmode == 1)) {
else if (model->VDMOSsubslGiven) {
/* numerical differentiation for gd and gm with a delta of 2 mV */
double vdsm = vds * here->VDMOSmode;
double delta = 0.001;
cdrain = cweakinv2(model->VDMOSsubsl, model->VDMOSsubshift, vgst, vds, model->VDMOSlambda,
Betam, vt, model->VDMOSmtr);
cdrain = cweakinv2(model->VDMOSsubsl, model->VDMOSsubshift, vgst, vdsm, model->VDMOSlambda,
Beta, vt, model->VDMOSmtr, model->VDMOStheta);
/* gd */
double vds1 = vds + delta;
double vds1 = vdsm + delta;
double cdrp = cweakinv2(model->VDMOSsubsl, model->VDMOSsubshift, vgst, vds1, model->VDMOSlambda,
Betam, vt, model->VDMOSmtr);
vds1 = vds - delta;
Beta, vt, model->VDMOSmtr, model->VDMOStheta);
vds1 = vdsm - delta;
double cdrm = cweakinv2(model->VDMOSsubsl, model->VDMOSsubshift, vgst, vds1, model->VDMOSlambda,
Betam, vt, model->VDMOSmtr);
Beta, vt, model->VDMOSmtr, model->VDMOStheta);
here->VDMOSgds = (cdrp - cdrm) / (2. * delta);
/* gm */
double vgst1 = vgst + delta;
cdrp = cweakinv2(model->VDMOSsubsl, model->VDMOSsubshift, vgst1, vds, model->VDMOSlambda,
Betam, vt, model->VDMOSmtr);
cdrp = cweakinv2(model->VDMOSsubsl, model->VDMOSsubshift, vgst1, vdsm, model->VDMOSlambda,
Beta, vt, model->VDMOSmtr, model->VDMOStheta);
vgst1 = vgst - delta;
cdrm = cweakinv2(model->VDMOSsubsl, model->VDMOSsubshift, vgst1, vds, model->VDMOSlambda,
Betam, vt, model->VDMOSmtr);
here->VDMOSgm = (cdrp - cdrm) / (2. * delta) * fgate + dfgdvg * cdrain;
cdrm = cweakinv2(model->VDMOSsubsl, model->VDMOSsubshift, vgst1, vdsm, model->VDMOSlambda,
Beta, vt, model->VDMOSmtr, model->VDMOStheta);
here->VDMOSgm = (cdrp - cdrm) / (2. * delta);
} else {
double onfg, fgate, Betam, dfgdvg;
onfg = 1.0+model->VDMOStheta*vgst;
fgate = 1.0/onfg;
Betam = Beta * fgate;
dfgdvg = -model->VDMOStheta*fgate*fgate;
if (vgst <= 0) {
/*
* cutoff region
@ -805,43 +827,17 @@ scalef(double nf2, double vgst)
*/
static double
cweakinv2(double slope, double shift, double vgst, double vds, double lambda, double beta, double vt, double mtr)
cweakinv2(double slope, double shift, double vgst, double vds, double lambda, double beta, double vt, double mtr, double theta)
{
double betam = beta / (1.0+theta*vgst);
vgst += shift * (1 - scalef(0.5, vgst));
double n = slope / 2.3 / 0.0256; /* Tsividis, p. 208 */
double n1 = n + (1 - n) * scalef(0.7, vgst); /* n < n1 < 1 */
double first = log(1 + exp(vgst / (2 * n1 * vt)));
double second = log(1 + exp((vgst - vds * mtr * n1) / (2 * n1 * vt)));
double cds =
beta * n1 * 2 * vt * vt * (1 + scalef(1, vgst) * lambda * vds) *
betam * n1 * 2 * vt * vt * (1 + scalef(1, vgst) * lambda * vds) *
(first * first - second * second);
return cds;
}
/* Alternative simple weak inversion model, according to https://www.anasoft.co.uk/MOS1Model.htm
* Scale the voltage overdrive vgst logarithmically in weak inversion.
* Best fits LTSPICE curves with shift=0
*/
static double
cweakinv(double slope, double shift, double vgst, double vds, double lambda, double beta, double vt, double mtr)
{
NG_IGNORE(vt);
double cdrain, betap;
vgst = slope * log(1 + exp((vgst - shift) / slope));
betap = beta*(1 + lambda*vds);
/* scale vds with mtr (except with lambda) */
if (vgst <= vds * mtr) {
/* saturation region */
cdrain = betap*vgst*vgst*.5;
}
else {
/* linear region */
cdrain = betap * vds * mtr *
(vgst - .5 * vds * mtr);
}
return cdrain;
}