VBIC MODEL REFERENCE FOR SIMULATIONS IN SPECTRE

Size: px

Start display at page:

Download "VBIC MODEL REFERENCE FOR SIMULATIONS IN SPECTRE"

Barnard Richard
6 years ago
Views:

1 VBIC MODEL REFERENCE FOR SIMULATIONS IN SPECTRE Compiled by Siddharth Nashiney This section includes: Review of the VBIC Model 1 Thermal Modeling 2 VBIC Model Instantiation 3 Conversion of Gummel-Poon Model Parameters to the VBIC Model 4 VBIC Model Parameters in Spectre 6 REVIEW OF THE VBIC MODEL 1

2 Figure 1: Equivalent Circuit-VBIC [1] The thermal network is added to the four terminal transistor network (collector, base, emitter and substrate) to model the local temperature rise. Self-heating is modeled by including the effects of local temperature rise on the branch constituent relationships for each network element at the four terminal network. This model is known to address many problems of the SPICE Gummel-Poon model. The model addresses several issues in the Gummel-Poon model: Improved Early effect modeling Quasi Saturation modeling Avalanche multiplication in collector junction modeling Electrothermal or self heating modeling Parasitic substrate transistor modeling Modulation of collector resistance Parasitic capacitances of base emitter overlap in double poly BJTs THERMAL MODELING q Devices at temperature different from ambient HSPICE [2] DTEMP instance definition parameter - A default of 0 indicates device at ambient (circuit) temperature SPECTRE [3] Trise model parameterða default of 0 indicates device at ambient (circuit) temperature q Enabling Self Heating in VBIC SPECTRE SIMULATOR[3] Selft self-heating flag model parameter- Make it 1 to enable self-heating 2

3 VBIC MODEL INSTANTIATION Hspice Instantiation [2]: VBIC (Vertical Bipolar Inter-Company) model can be invoked by specifying level = 4 in HSPICE for the bipolar transistor model. The advanced version VBIC 99 can be invoked by specifying level = 9. When self heating is accounted for the device syntax becomes Qxxx nc nb ne <ns><nt>mname<regular parameters><tnodeout> The parameter nt is the node for temperature. If this node is given but ns is not specified, then a flag <tnodeout> must be specified to indicate the fourth node is temperature node and not substrate node. To turn on self-heating, in addition to giving the T node, the model parameter, R th, must be non-zero in the model card. The SpectreS Instantiation [3]: Instance Definition Name c b e [s] [dt] [tl] ModelName parameter=value... q q ÔtlÕ node is the local temperature ÔdtÕ node is the rise above the local temperature caused by the thermal power dissipated by the device being modeled by VBIC. Consequently, the ÔtlÕ node can be connected to a thermal network that models heat flow through the substrate and/or between devices. If the self-heating flag is turned on and dt is not given, an internal node is created for self-heating. 3

4 CONVERSION OF GUMMEL-POON PARAMETERS TO THE VBIC MODEL Need for Conversion: The model formulation available for the transistors being the Gummel - Poon (GP) model, an appropriate conversion table had to be developed. The advanced VBIC model parameters have to be disabled, as their values are not known through the available model parameters. The following empirical conversion was used based on the work done by Fujian lin, et al [1]. Empirical Conversions [1]: Empirical Conversion needs to be carried out for the Gummel-Poon Parameters IS, VAF, VAR, IKF and IKR. The following points give a brief overview of the same. The transport saturation current value is to be reduced while using the VBIC model due to the different q b (base charge) modeling in the VBIC model. The implementation of the Early Effect in the VBIC model is also different compared to that in the Spice Gummel-Poon (SGP) model. In the SGP Model, the early voltages come from DC data only. In the VBIC Model, the early voltage is obtained from the forward and reverse output characteristics based on the junction capacitance parameters [4]. High Current roll-off is less effective in VBIC and hence the values of IKF and IKR need to be reduced. Correction to the empirical adjustments on saturation current and early effect values could be made once the measurements are obtained [1]. VEF new sim = VEF. ( Ic 2 Ð Ic sim mea 1 ) / ( Ic 2 Ð Ic mea 1 ) IS new = IS. ( Ic mea / Ic sim ) CONVERSION TABLE [1]: Direct Substitution of parameter values to be done unless a conversion factor is specified. 4

5 TABLE 1:GP ÐVBIC Conversion Table [1] VBIC GP Conversion to be used IS IS IS*0.9 NF NR NF NR VEF VAF VAF*0.5 VER VAR VAR*0.5 IKF IKF IKF*0.9 IKR IKR IKR*0.9 IBEI NEI IBEN NEN IBCI NCI IBCN NCN RE RBX RCX CJE PE XIS XII XIN KFN AFN IS/BF NF ISE NE IS/BR NR ISC NC RE RBM RC CJE VJE XTI XTB XTB KF AF ME MJE CJC CJC PC VJC MC MJC CJEP XCJC (1-XCJC)*CJC CJCP CJS PS VJS MS MJS FC FC TF TF XTF XTF ITF ITF VTF VTF TR TR TD PTF TREF (TNOM TREF (TNOM used in used in SPECTRE) SPECTRE) EA EG RTH CTH Additional SPECTRE model parameters selft - Selft = 1 to enable self-heating trise trise *SGP Parameters left unused: RB, IRB *Other VBIC parameters use default values to disable advanced effects 5

6 VBIC MODEL PARAMETERS IN SPECTRE [3] Instance Parameters 1 area Transistor area factor. 2 m Multiplicity factor. 3 region Estimated operating region. Possible values are off, fwd, rev, sat, or breakdown. 4 trise Temperature rise from ambient. Model Definition model modelname vbic parameter=value... Model Parameters Structural Parameters 1 type=npn Transistor type. Possible values are npn or pnp. Saturation Current Parameters 2 is Transport saturation current (*area). 3 ibei Ideal B-E saturation current. (*area). 4 iben Nonideal B-E saturation current (*area). 5 ibci Ideal B-C saturation current. (*area). 6 ibcn Nonideal B-C saturation current (*area). 7 isp Parasitic transport saturation current. (*area). 8 ibeip Ideal parasitic B-E saturation current (*area). 9 ibenp Nonideal parasitic B-E saturation current (*area). 10 ibcip Ideal parasitic B-C saturation current (*area). 11 ibcnp Nonideal parasitic B-C saturation current (*area). 12 vo Epi drift saturation voltage. 13 gamm Epi doping parameter. 14 hrcf High current RC factor. 15 wbe Portion of Ibei from Vbei. 6

7 16 wsp Portion of Iccp from Vbep. Emission Coefficient Parameters 17 nf Forward emission coefficient. 18 nr Reverse emission coefficient. 19 nei Ideal B-E emission coefficient. 20 nen Nonideal B-E emission coefficient. 21 nci Ideal B-C emission coefficient. 22 ncn Nonideal B-C emission coefficient. 23 nfp Parasitic forward emission coefficient. 24 ncip Ideal parasitic B-C emission coefficient. 25 ncnp Nonideal parasitic B-C emission coefficient. Current Gain Parameters 26 ikf Forward knee current 27 ikr Reverse knee current 28 ikp Parasitic knee current Early Voltage Parameters 29 vef Forward Early voltage. 30 ver Reverse Early voltage. Breakdown Voltage Parameters 31 avc1 B-C weak avalanche parameter. 32 avc2 B-C weak avalanche parameter. Parasitic Resistance Parameters 33 rbi Intrinsic base resistance (/area). 7

8 34 rbx Extrinsic base resistance (/area). 35 re Emitter resistance (/area). 36 rs Substrate resistance (/area). 37 rbp Parasitic base resistance (/area). 38 rcx Extrinsic collector resistance (/area). 39 rci Intrinsic collector resistance (/area). Junction Capacitance Parameters 40 cje B-E zero-bias capacitance (*area). 41 pe B-E built-in potential. 42 me B-E grading coefficient. 43 aje B-E capacitance smoothing factor. 44 fc Forward-bias depletion capacitance limit. 45 cbeo Extrinsic B-E overlap capacitance (*area). 46 cjc B-C zero-bias capacitance (*area). 47 cjep B-C extrinsic zero-bias capacitance (*area). 48 pc B-C built-in potential. 49 mc B-C grading coefficient. 50 ajc B-C capacitance smoothing factor. 51 cbco Extrinsic B-C overlap capacitance (*area). 52 qco Epi charge parameter. 53 cjcp S-C zero-bias capacitance (*area). 54 ps S-C built-in potential. 55 ms S-C grading coefficient. 56 ajs S-C capacitance smoothing factor. Transit Time and Excess Phase Parameters 57 tf Forward transit time. 58 tr Reverse transit time. 8

9 59 td Forward excess-phase delay time. 60 qtf Variation of tf with base width modulation. 61 xtf Coefficient of tf with bias dependence. 62 vtf Coefficient of tf dependence on Vbc. 63 itf Coefficient of tf dependence on Ic. Temperature Effects Parameters 64 selft Flag denoting self-heating. Possible values are no or yes. 65 tnom Parameters measurement temperature. Default set by options. 66 trise Temperature rise from ambient. 67 rth Thermal resistance, must be given for self-heating. 68 cth Thermal capacitance. 69 xis Temperature exponent of Is. 70 xii Temperature exponent of Ibei, Ibci, Ibeip, and Ibcip. 71 xin Temperature exponent of Iben, Ibcn, Ibenp, and Ibcnp. 72 tnf Temperature coefficient of Nf. 73 tavc Temperature coefficient of Avc2. 74 ea Activation energy for is. 75 eaie Activation energy for Ibei. 76 eaic Activation energy for Ibci/Ibeip. 77 eais Activation energy for Ibcip. 78 eane Activation energy for Iben. 79 eanc Activation energy for Ibcn/Ibenp. 80 eans Activation energy for Ibcnp. 81 xre Temperature exponent of re. 82 xrb Temperature exponent of rb. 83 xrc Temperature exponent of rc. 84 xrs Temperature exponent of rs. 9

10 85 xvo Temperature exponent of vo. 86 dtmax Maximum expected device temperature. Noise Model Parameters 87 kfn B-E flicker (1/f) noise coefficient. 88 afn B-E flicker (1/f) noise exponent. 89 bfn B-E flicker (1/f) noise dependence. REFERENCES [1] Fujian Lin, et al, ÒExtraction of VBIC Model for Si-Ge HBTs Made Easy by going through Gummel-Poon ModelÓ, Singapore Microelectronics Modeling Center, International Symposium on Microelectronics and Assembly, November 2000, web-link: This paper discusses obtaining the VBIC model parameters from the Gummel-Poon model for a Si-Ge HBT and develops the conversion table. It does empirical curve fitting based on measurements carried out for a Si-Ge HBT. It suggests typical values of advanced VBIC model parameters. [2] Star-Hspice Manual, Release , December 2001 [3] Cadence Documentation, Affirma Spectre Circuit Simulator Reference, Product Version 4.4.6, 2000 [4] G.W. Huang, K.M. Chen, J.F.Kuan, Y.M.Deng, S.Y.Wen, D.Y.Chiu, M.T.Wang, ÓSilicon BJT Modeling using VBIC ModelÓ, Proceedings of APMC2001, Taipei, Taiwan The paper discusses the differences in the VBIC modeling of certain parameters with respect to the Gummel-poon model. 10

Agilent EEsof EDA.

Agilent EEsof EDA. Agilent EEsof EDA This document is owned by Agilent Technologies, but is no longer kept current and may contain obsolete or inaccurate references. We regret any inconvenience this may cause. For the latest