SUBJECT: mainTITLE: Table of ContentsTEXT: HTEXT: HTEXT: HTEXT: HTEXT: HTEXT: HTEXT: HSUBTOPIC: SPICE:INTRODUCTIONSUBTOPIC: SPICE:CIRCUIT DESCRIPTIONSUBTOPIC: SPICE:CIRCUIT ELEMENTS AND MODELSSUBTOPIC: SPICE:ANALYSES AND OUTPUT CONTROLSUBTOPIC: SPICE:INTERACTIVE INTERPRETERSUBTOPIC: SPICE:BIBLIOGRAPHYSUBTOPIC: SPICE:APPENDIX ASUBTOPIC: SPICE:APPENDIX BSUBJECT: INTRODUCTIONTITLE: INTRODUCTIONTEXT: HTEXT: H _1.  _I_N_T_R_O_D_U_C_T_I_O_NTEXT: HTEXT: HTEXT: H      SPICE is a general-purpose circuit  simulation  programTEXT: H for  nonlinear  dc,  nonlinear transient, and linear ac ana-TEXT: H lyses.  Circuits may contain resistors,  capacitors,  induc-TEXT: H tors,  mutual  inductors,  independent  voltage  and currentTEXT: H sources, four types of dependent sources, lossless and lossyTEXT: H transmission lines (two separate implementations), switches,TEXT: H uniform distributed RC lines, and the five most common  sem-TEXT: H iconductor  devices:  diodes, BJTs, JFETs, MESFETs, and MOS-TEXT: H FETs.TEXT: HTEXT: H      The SPICE3 version is based  directly  on  SPICE  2G.6.TEXT: H While  SPICE3 is being developed to include new features, itTEXT: H continues to support those  capabilities  and  models  whichTEXT: H remain in extensive use in the SPICE2 program.TEXT: HTEXT: H      SPICE has built-in models for  the  semiconductor  dev-TEXT: H ices,  and  the  user  need specify only the pertinent modelTEXT: H parameter values.  The model for the BJT  is  based  on  theTEXT: H integral-charge  model  of Gummel and Poon;  however, if theTEXT: H Gummel- Poon parameters are not specified, the model reducesTEXT: H to  the  simpler  Ebers-Moll model.  In either case, charge-TEXT: H storage effects, ohmic resistances, and a  current-dependentTEXT: H output  conductance may be included.  The diode model can beTEXT: H used for either junction diodes or Schottky barrier  diodes.TEXT: H The  JFET  model  is  based on the FET model of Shichman andTEXT: H Hodges.   Six  MOSFET  models  are  implemented:   MOS1   isTEXT: H described by a square-law I-V characteristic, MOS2 [1] is anTEXT: H analytical model, while MOS3 [1] is a semi-empirical  model;TEXT: H MOS6  [2]  is a simple analytic model accurate in the short-TEXT: H channel region; MOS4 [3,  4]  and  MOS5  [5]  are  the  BSIMTEXT: H (Berkeley Short-channel IGFET Model) and BSIM2.  MOS2, MOS3,TEXT: H and MOS4 include second-order effects such as channel-lengthTEXT: H modulation,   subthreshold   conduction,  scattering-limitedTEXT: H velocity  saturation,  small-size   effects,   and   charge-TEXT: H controlled capacitances.SUBTOPIC: SPICE:TYPES OF ANALYSISSUBTOPIC: SPICE:ANALYSIS AT DIFFERENT TEMPERATURESSUBTOPIC: SPICE:CONVERGENCESUBJECT: TYPES OF ANALYSISTITLE: TYPES OF ANALYSISTEXT: HTEXT: H _1._1.  _T_Y_P_E_S _O_F _A_N_A_L_Y_S_I_STEXT: HSUBTOPIC: SPICE:DC AnalysisSUBTOPIC: SPICE:AC SmallSignal AnalysisSUBTOPIC: SPICE:Transient AnalysisSUBTOPIC: SPICE:PoleZero AnalysisSUBTOPIC: SPICE:SmallSignal Distortion AnalysisSUBTOPIC: SPICE:Sensitivity AnalysisSUBTOPIC: SPICE:Noise AnalysisSUBJECT: DC AnalysisTITLE: DC AnalysisTEXT: HTEXT: H _1._1._1.  _D_C _A_n_a_l_y_s_i_sTEXT: HTEXT: HTEXT: H      The dc analysis portion  of  SPICE  determines  the  dcTEXT: H operating  point  of  the circuit with inductors shorted andTEXT: H capacitors opened.  The dc analysis options are specified onTEXT: H the  .DC,  .TF,  and  .OP  control  lines.  A dc analysis isTEXT: H automatically performed prior to  a  transient  analysis  toTEXT: H determine  the transient initial conditions, and prior to anTEXT: H ac  small-signal  analysis  to  determine  the   linearized,TEXT: H small-signal  models  for  nonlinear devices.  If requested,TEXT: H the dc small-signal value of a transfer function  (ratio  ofTEXT: H output variable to input source), input resistance, and out-TEXT: H put resistance is also computed as a part of  the  dc  solu-TEXT: H tion.   The  dc  analysis  can  also  be used to generate dcTEXT: H transfer curves:  a specified independent voltage or currentTEXT: H source  is  stepped  over  a user-specified range and the dcTEXT: H output variables  are  stored  for  each  sequential  sourceTEXT: H value.TEXT: HSUBJECT: AC SmallSignal AnalysisTITLE: AC Small-Signal AnalysisTEXT: HTEXT: H _1._1._2.  _A_C _S_m_a_l_l-_S_i_g_n_a_l _A_n_a_l_y_s_i_sTEXT: HTEXT: HTEXT: H      The ac small-signal portion of SPICE  computes  the  acTEXT: H output  variables  as  a function of frequency.  The programTEXT: H first computes the dc operating point  of  the  circuit  andTEXT: H determines  linearized,  small-signal  models for all of theTEXT: H nonlinear devices in the circuit.  The resultant linear cir-TEXT: H cuit  is  then  analyzed over a user-specified range of fre-TEXT: H quencies.   The  desired  output  of  an  ac  small-  signalTEXT: H analysis is usually a transfer function (voltage gain, tran-TEXT: H simpedance, etc).  If the circuit has only one ac input,  itTEXT: H is  convenient to set that input to unity and zero phase, soTEXT: H that output variables have the same value  as  the  transferTEXT: H function of the output variable with respect to the input.TEXT: HSUBJECT: Transient AnalysisTITLE: Transient AnalysisTEXT: HTEXT: H _1._1._3.  _T_r_a_n_s_i_e_n_t _A_n_a_l_y_s_i_sTEXT: HTEXT: H      The transient analysis portion of  SPICE  computes  theTEXT: H transient  output  variables  as  a  function of time over aTEXT: H user-specified time interval.  The  initial  conditions  areTEXT: H automatically  determined  by  a  dc  analysis.  All sourcesTEXT: H which are not time dependent (for example,  power  supplies)TEXT: H are  set  to their dc value.  The transient time interval isTEXT: H specified on a .TRAN control line.TEXT: HSUBJECT: PoleZero AnalysisTITLE: Pole-Zero AnalysisTEXT: HTEXT: H _1._1._4.  _P_o_l_e-_Z_e_r_o _A_n_a_l_y_s_i_sTEXT: HTEXT: HTEXT: H      The pole-zero analysis portion of  SPICE  computes  theTEXT: H poles and/or zeros in the small-signal ac transfer function.TEXT: H The program first computes the dc operating point  and  thenTEXT: H determines  the  linearized, small-signal models for all theTEXT: H nonlinear devices in the circuit.  This circuit is then usedTEXT: H to find the poles and zeros of the transfer function.TEXT: HTEXT: H      Two types of transfer functions are allowed  :  one  ofTEXT: H the  form  (output voltage)/(input voltage) and the other ofTEXT: H the form (output voltage)/(input current).  These two  typesTEXT: H of  transfer  functions cover all the cases and one can findTEXT: H the poles/zeros of functions like input/output impedance andTEXT: H voltage  gain.   The input and output ports are specified asTEXT: H two pairs of nodes.TEXT: HTEXT: H      The pole-zero analysis works  with  resistors,  capaci-TEXT: H tors,   inductors,  linear-controlled  sources,  independentTEXT: H sources, BJTs,  MOSFETs,  JFETs  and  diodes.   TransmissionTEXT: H lines are not supported.TEXT: HTEXT: H      The method used in the analysis is a sub-optimal numer-TEXT: H ical  search.  For large circuits it may take a considerableTEXT: H time or fail to find all poles and  zeros.   For  some  cir-TEXT: H cuits,  the  method  becomes  "lost"  and finds an excessiveTEXT: H number of poles or zeros.TEXT: HSUBJECT: SmallSignal Distortion AnalysisTITLE: Small-Signal Distortion AnalysisTEXT: HTEXT: H _1._1._5.  _S_m_a_l_l-_S_i_g_n_a_l _D_i_s_t_o_r_t_i_o_n _A_n_a_l_y_s_i_sTEXT: HTEXT: HTEXT: H      The  distortion  analysis  portion  of  SPICE  computesTEXT: H steady-state harmonic and intermodulation products for smallTEXT: H input signal magnitudes.  If signals of a  single  frequencyTEXT: H are  specified  as  the  input  to  the circuit, the complexTEXT: H values of the second and third harmonics are  determined  atTEXT: H every  point  in  the  circuit.  If there are signals of twoTEXT: H frequencies input to the circuit, the analysis finds out theTEXT: H complex  values  of  the  circuit  variables  at the sum andTEXT: H difference of the input frequencies, and at  the  differenceTEXT: H of  the  smaller  frequency  from the second harmonic of theTEXT: H larger frequency.TEXT: HTEXT: H      Distortion analysis is supported for the following non-TEXT: H linear  devices: diodes (DIO), BJT, JFET, MOSFETs (levels 1,TEXT: H 2, 3, 4/BSIM1, 5/BSIM2, and 6) and MESFETS.  All linear dev-TEXT: H ices are automatically supported by distortion analysis.  IfTEXT: H there are switches present in the circuit, the analysis con-TEXT: H tinues  to  be  accurate provided the switches do not changeTEXT: H state under the small excitations used for distortion calcu-TEXT: H lations.TEXT: HSUBJECT: Sensitivity AnalysisTITLE: Sensitivity AnalysisTEXT: HTEXT: H _1._1._6.  _S_e_n_s_i_t_i_v_i_t_y _A_n_a_l_y_s_i_sTEXT: HTEXT: HTEXT: H      Spice3 will calculate  either  the  DC  operating-pointTEXT: H sensitivity  or the AC small-signal sensitivity of an outputTEXT: H variable with respect to all  circuit  variables,  includingTEXT: H model  parameters.   Spice  calculates  the difference in anTEXT: H output variable (either a node voltage or a branch  current)TEXT: H by  perturbing  each parameter of each device independently.TEXT: H Since the method is a numerical approximation,  the  resultsTEXT: H may  demonstrate  second  order  affects in highly sensitiveTEXT: H parameters, or may fail to show very low but non-zero sensi-TEXT: H tivity.   Further, since each variable is perturb by a smallTEXT: H fraction of its value, zero-valued parameters are not analy-TEXT: H ized  (this  has  the  benefit of reducing what is usually aTEXT: H very large amount of data).TEXT: HSUBJECT: Noise AnalysisTITLE: Noise AnalysisTEXT: HTEXT: H _1._1._7.  _N_o_i_s_e _A_n_a_l_y_s_i_sTEXT: HTEXT: HTEXT: H      The noise  analysis  portion  of  SPICE  does  analysisTEXT: H device-generated noise for the given circuit.  When providedTEXT: H with an input source and an output port, the analysis calcu-TEXT: H lates the noise contributions of each device (and each noiseTEXT: H generator within the device) to the output port voltage.  ItTEXT: H also  calculates  the input noise to the circuit, equivalentTEXT: H to the output noise referred to the specified input  source.TEXT: H This  is done for every frequency point in a specified rangeTEXT: H - the calculated value of the noise corresponds to the spec-TEXT: H tral  density of the circuit variable viewed as a stationaryTEXT: H gaussian stochastic process.TEXT: HTEXT: H      After  calculating  the   spectral   densities,   noiseTEXT: H analysis  integrates  these  values  over the specified fre-TEXT: H quency range to arrive at the  total  noise  voltage/currentTEXT: H (over   this   frequency   range).   This  calculated  valueTEXT: H corresponds to the variance of the circuit  variable  viewedTEXT: H as a stationary gaussian process.SUBJECT: ANALYSIS AT DIFFERENT TEMPERATURESTITLE: ANALYSIS AT DIFFERENT TEMPERATURESTEXT: HTEXT: H _1._2.  _A_N_A_L_Y_S_I_S _A_T _D_I_F_F_E_R_E_N_T _T_E_M_P_E_R_A_T_U_R_E_STEXT: HTEXT: HTEXT: H      All input data for SPICE is assumed to have been  meas-TEXT: H                                     oTEXT: H ured  at a nominal temperature of 27 C, which can be changedTEXT: H by use of the TNOM parameter on the  .OPTION  control  line.TEXT: H This  value  can  further be overridden for any device whichTEXT: H models temperature effects by specifying the TNOM  parameterTEXT: H on the model itself.  The circuit simulation is performed atTEXT: H                    oTEXT: H a temperature of 27 C, unless overridden by a TEMP parameterTEXT: H on  the  .OPTION  control  line.   Individual  instances mayTEXT: H further override the circuit temperature through the specif-TEXT: H ication of a TEMP parameter on the instance.TEXT: HTEXT: H      Temperature dependent support is  provided  for  resis-TEXT: H tors,  diodes,  JFETs,  BJTs, and level 1, 2, and 3 MOSFETs.TEXT: H BSIM (levels 4 and 5) MOSFETs have an alternate  temperatureTEXT: H dependency  scheme which adjusts all of the model parametersTEXT: H before input to SPICE.  For details of the BSIM  temperatureTEXT: H adjustment, see [6] and [7].TEXT: HTEXT: HTEXT: H      Temperature appears explicitly in the exponential termsTEXT: H of  the BJT and diode model equations.  In addition, satura-TEXT: H tion currents have a built-in temperature  dependence.   TheTEXT: H temperature  dependence of the saturation current in the BJTTEXT: H models is determined by:TEXT: HTEXT: H                              XTITEXT: H                          |T |        | E q(T  T )|TEXT: H                            1            g   1  0TEXT: H          I (T ) = I (T ) |--|     exp|-----------|TEXT: H           S  1     S  0TEXT: H                          |T |        |k (T  - T )|TEXT: H                            0              1    0TEXT: HTEXT: HTEXT: HTEXT: H where k is Boltzmann's constant,  q  is  the  electronicTEXT: H charge, E  is the energy gap which is a model parameter,TEXT: H          GTEXT: H and XTI is the saturation current  temperature  exponentTEXT: H (also a model parameter, and usually equal to 3).TEXT: HTEXT: HTEXT: HTEXT: H      The temperature dependence of forward and reverse  betaTEXT: H is according to the formula:TEXT: HTEXT: H                                      XTBTEXT: H                                  |T |TEXT: H                                    1TEXT: H                    B(T ) = B(T ) |--|TEXT: H                       1       0TEXT: H                                  |T |TEXT: H                                    0TEXT: HTEXT: HTEXT: HTEXT: H where T  and T  are in degrees  Kelvin,  and  XTB  is  aTEXT: H        1      0TEXT: H user-supplied  model  parameter.  Temperature effects onTEXT: H beta are carried out by appropriate  adjustment  to  theTEXT: H values  of  B , I  , B , and I   (spice model parametersTEXT: H              F   SE   R       SCTEXT: H BF, ISE, BR, and ISC, respectively).TEXT: HTEXT: HTEXT: HTEXT: H      Temperature dependence of the saturation current in theTEXT: H junction diode model is determined by:TEXT: HTEXT: H                             XTITEXT: H                             ---TEXT: H                              NTEXT: H                         |T |        |  E q(T  T ) |TEXT: H                           1             g   1  0TEXT: H         I (T ) = I (T ) |--|     exp|-------------|TEXT: H          S  1     S  0TEXT: H                         |T |        |N k (T  - T )|TEXT: H                           0                1    0TEXT: HTEXT: HTEXT: HTEXT: H where N is the emission coefficient, which  is  a  modelTEXT: H parameter,  and  the other symbols have the same meaningTEXT: H as above.  Note that for Schottky  barrier  diodes,  the