% Mike Scott % HW2 Spring 02 % Solution Set % MATLAB script. % MOSFET MODELING % Questions 1-4. % Vector of lengths (in um) L = [

Transcription

1 % Mike Scott % HW Spring % Solution Set % MATLAB script % MOSFET MODELING % Questions 1-4 % Vector of lengths (in um) L = [ ]; % Row key for Vth, ucox and lambda matrices are: % NOMINAL % SLOW % FAST % Extrapolated directly from either triode plot % or diode connected plot (see plots) Vth = [ ; ; ;]; % Slopes of triode plot are in fact ucox directly % Measure slopes from Awaves (slopes not shown on plot) ucox = [7e-6 5e-6 e-6 3e-6; 174e-6 177e-6 184e-6 187e-6; 36e-6 33e-6 e-6 1e-6;]; % Lambda can be found from saturation slope along % with the average value of Id in saturation % Slopes are from the derivative plot did/dvds (see plots) slope = [1e-6 35e-6 1e-6 7e-6; 8e-6 5e-6 8e-6 7e-6; 19e-6 5e-6 1e-6 7e-6;]; % Average saturation currents from plot (see plots) Id = [1.67e e e-3.9e-3; 1.1e-3 1.6e e-3 1.6e-3;.44e-3.3e-3.4e-3.7e-3;]; lambda = slope./id; % Results (from above line) lambda = [ ; ; ;]; % Plot the Vth, ucox and lambda data subplot(3,1,1) plot(l,vth(1,:),'x-',l,vth(,:),'o--',l,vth(3,:),'s:'); grid; ylabel('threshold Voltage [V]') title('nmos parameters vs channel length') legend('nominal','slow','fast') subplot(3,1,) plot(l,ucox(1,:),'x-',l,ucox(,:),'o--',l,ucox(3,:),'s:'); grid; ylabel('kprime (ucox) [ua/v^]') subplot(3,1,3) plot(l,lambda(1,:),'x-',l,lambda(,:),'o--',l,lambda(3,:),'s:'); grid; ylabel('lambda [V^-1]') xlabel('channel length [um]') % Calculate Id, gm, ro, Vdsat for W/L=1/1 and Vgs=Vds=1V (nominal) % Question 5 Vgs = 1; Vds = 1; ucox = e-6;

2 Vth =.5; lambda =.55; Id = ucox/*1/1*(vgs-vth)^*(1+lambda*vds) % Result:.5389mA Vdsat = Vgs-Vth % Result:.5V gm = *Id/Vdsat % Result: 1.156mS ro = 1/(lambda*Id) % Result: kohm % 1/ro = go = uS % Questions 6-7 (see following SPICE deck and plots) % TUBE MODELING % Question 8 % From datasheet, we can only realize 1mA of plate current with % -75V on the grid using the data from the 5V screen voltage plot % Bias point from 5V screen voltage plot Vscreen = 5; Iplate =.1; Vgrid = -75; Vplate = 35; % gm is defined as diplate/dvgrid evaluated at constant Vplate % (note that this is just like a BJT, MOS, JFET, MESFET, etc) % Find a point near the bias point (with Vplate=35V) to get the % derivative using the bias point and the new point Iplate =.5; Vgrid = -88; gm = (Iplate-Iplate)/(Vgrid-Vgrid) % Result: 3.8mS % ro is defined as (dvplate/diplate) evaluated at constant Vgrid % (also just like a BJT, MOS, JFET, MESFET, etc) % Find a point near bias point (with Vgrid=-75V) to get derivative Vplate = 75; Iplate =.5; ro = (Vplate-Vplate)/(Iplate-Iplate) %Result: 5kohm % Maximum single stage gain of this tube is achieved by assuming % that the tube only sees its own output impedance % In this case the gain of the stage is just Av = gm*ro % (amazing...just like a BJT, MOS, JFET, MESFET, etc) magav = gm*ro % Result: 19V/V

3 Threshold Voltage [V] NMOS parameters vs channel length Nominal Slow Fast x1-4.4 kprime (ucox) [ua/v ] lambda [V - 1] Channel length [um]

4 * Mike Scott * HW Spring * Solution Set * SPICE deck and results * First set up the devices to test * Normal devices, use them in triode region M1 Vds Vtest NMOS W=35u L=.35u M Vds Vtest NMOS W=5u L=.5u M3 Vds Vtest NMOS W=1u L=1u M4 Vds Vtest NMOS W=u L=u * Diode connected devices, use them in saturation M1d Vtest Vtest NMOS W=35u L=.35u Md Vtest Vtest NMOS W=5u L=.5u M3d Vtest Vtest NMOS W=1u L=1u M4d Vtest Vtest NMOS W=u L=u * Normal devices, use them in both regions M1n Vtest Vgs NMOS W=35u L=.35u Mn Vtest Vgs NMOS W=5u L=.5u M3n Vtest Vgs NMOS W=1u L=1u M4n Vtest Vgs NMOS W=u L=u * Now set up the bias sources Vtest Vtest DC 1V Vds Vds DC.1V Vgs Vgs DC 1V * Level device ML Vtest Vgs nmosl W=1u l=1u VIdcalc VIdcalc DC.55e-3 * Level NMOS model for W/L = 1/1.model nmosl nmos vto=.5 kp=u lambda=.55.op.dc Vtest 5.1 * Probe the relevant currents for finding Vth, ucox lambda.probe I1(M1) I1(M) I1(M3) I1(M4) rti1=par('sqrt(i1(m1d))') rti=par ('sqrt(i1(md))') rti3=par('sqrt(i1(m3d))') rti4=par('sqrt(i1(m4d))') I1 (M1n) I1(Mn) I1(M3n) I1(M4n) I1(ML).options dccap accurate brief post= * Include the Level 49 models.lib 'lib/tsmc35.txt' fast.end ************************************************ ************************************************ * Resulting operating point for level device * subckt * element :ml * model :nmosl * id.5389m * Calculated.5389mA * ibs. * ibd -1.f * vgs 1. * vds 1.

5 * vbs. * vth 5.m * vdsat 5.m * Calculated 5mV * beta.3111m * gam eff 57.65m * gm m * Calculated 1.156mS * gds u * Calculated uS * gmb m * cdtot 4.64e-17 * cgtot 3.489f * cstot 3.11f * cbtot 3.417e-16 * cgs 3.11f * cgd 4.64e-17

6 u 1m.5m m 1.5m 1m 5u Triode: Id vs Vds Circle - 35/.35u Diamond - 5/.5u Triangle - 1/1u X - /u 5m Saturation (diode Triangle connected): - 1/1u root(id) vs Vds 5m Full IV: Id vs Vds (Vgs=1.V) 5m Params (lin)

7 m 18u 16u 14u 1u 1u 8u 6u 4u u Full IV: Id vs Vds (Vgs=1.V) 5m Derivative of Id w.r.t. Vds (Vgs=1.V) Result (lin)

8 u 1m 3m.5m m 1.5m 1m 5u Triode: Id vs Vds FAST models 5m Saturation (diode connected): root(id) vs Vds FAST models 5m Full IV: Id vs Vds (Vgs=1.V) FAST models 5m Params (lin)

9 .8m.6m.4m.m m 1.8m 1.6m 1.4m 1.m 1m 8u 6u 4u u 4u u u 18u 16u 14u 1u 1u 8u 6u 4u u Full IV: Id vs Vds (Vgs=1.V) FAST models 5m Derivative of Id w.r.t.vds (Vgs=1.V) FAST models Result (lin)

10 u 1m m Triode: Id vs Vds SLOW models 5m Saturation (diode connected): root(id) vs Vds SLOW models 5m Full IV: Id vs Vds (Vgs=1.V) SLOW models 5m Params (lin)

11 m 1u 1u 8u 6u 4u u Full IV: Id vs Vds (Vgs=1.V) SLOW models 5m Derivative of Id w.r.t.vds (Vgs=1.V) SLOW models Result (lin)

12 4m m 1k 1k 1k 1 1 Level vs Level 49: Id vs Vds (Vgs=1.V) Breakdown Calculated Vdsat =.5V Level Calculated Id =.53mA Level 49 5m Level vs Level 49: ro vs Vds (Vgs=1.V) Level Level 49 5m Result (log)