bmurmann
commented
8 months ago Also, I am surprised to see the sign of gm flip negative when VDS=0. I know it's a small value, but I've never seen a MOS model that does this. The model expression should give exactly zero in this condition.
PSPNQS103VA: A simulator independent device loaded with OSDI device n.xm1.nsg13_lv_nmos model xm1:sg13g2_lv_nmos_ps _mfactor 1 m 1 l 1.3e-07 w 5e-06 sa 0 sb 0 sd 0 sca 0 scb 0 scc 0 sc 0 nf 1 ngcon 2 xgw 1e-07 nrs 0 nrd 0 jw 1e-06 delvto 0 factuo 1 delvtoedge 0 factuoedge 1 absource 1e-12 lssource 1e-06 lgsource 1e-06 abdrain 1e-12 lsdrain 1e-06 lgdrain 1e-06 as 1.7e-12 ps 1.068e-05 ad 1.7e-12 pd 1.068e-05 mult 1 dta 0 ctype 1 sdint 1 ise -1.41647e-12 ige 2.83294e-12 ide -1.41647e-12 ibe -3.38614e-22 ids 0 idb 0 isb 0 igs 1.41647e-12 igd 1.41647e-12 igb 3.38614e-22 idedge 0 igcs 1.19916e-12 igcd 1.19916e-12 iavl 0 igisl 0 igidl 0 ijs 0 ijsbot 0 ijsgat 0 ijssti 0 ijd 0 ijdbot 0 ijdgat 0 ijdsti 0 vds 0 vgs 0.6 vsb 0 vto 0.436045 vts 0.436045 vth 0.436045 vgt 0.163955 vdss 0.193087 vsat -0.193087 gm -1.38899e-11 gmb -3.15928e-13 gds 0.00127697
The simulated gm/gds versus gate bias (for various L) for an lv_nmos looks quite odd for short channels (see plot below). Does the measured data support this? I found only plots of gds versus drain bias in the documentation. The netlist to duplicate this plot is below.
sch_path: /foss/designs/gds_n.sch .subckt gds_n Xm1 d g GND b sg13_lv_nmos W={Wx} L={Lx} ng=1 m=1 Vgs g GND 0.6 Vds d GND 1.2 Vsb GND b 0 **** begin user architecture code .lib /foss/pdks/ihp-sg13g2/libs.tech/ngspice/models/cornerMOSlv.lib mos_tt
.param temp=27 .param Wx=5u .param Lx=0.13u .dc Vgs 0 1.2 0.01 .save @n.xm1.nsg13_lv_nmos[gm] .save @n.xm1.nsg13_lv_nmos[gds]
.control *pre_osdi ./psp103_nqs.osdi
foreach L_val 0.13u 0.14u 0.15u 0.16u 0.17u 0.18u 0.19u 0.2u 0.3u 0.4u 0.5u 1u 2u 3u alterparam Lx = $L_val reset run let av=@n.xm1.nsg13_lv_nmos[gm]/@n.xm1.nsg13_lv_nmos[gds] end
plot all.av
.endc
** end user architecture code .ends .GLOBAL GND .end