微波EDA网,见证研发工程师的成长!
首页 > 研发问答 > 微波和射频技术 > RFIC设计学习交流 > 手算参数K值选择

手算参数K值选择

时间:10-02 整理:3721RD 点击:
刚接触CMOS的学习设计,通常仿真前需要根据电路性能指标手算预估MOS管的宽长比,我想知道大家是如何选取较精确的本征挂到参数K值的,等于uCox,还有沟道长度调制系数LAMBDA,以及阈值电压Vt的选取?因为仿真发现这些参数都是在变化的,那么咱们手算的时候到底怎么取才较精确呢?至少可以保证计算的准确度高一点点,哪怕只有一点点也好。求解答,谢谢!

书上说从工艺库模型中查到的,下面是我在工艺库文件里找到的NMOS模型,请问是哪些个参数。实在不知道,求解答。
.model mn nmos
+level =49
*
* general parameters
*
+lmin = 5.0e-7lmax = 2.0e-5wmin = 5.0e-7wmax = 2.0e-5
+tref = 25.0
+version = 3.2
+hspver =2000.2
+tox = '1.25e-08+toxn'
+xj =1.0000000e-07
+nch =1.4924999e+17
+lln =1.0000000
+lwn =1.0000000
+wln =1.0000000
+wwn =1.0000000
+lint =0.00
+ll =1.0850657e-14
+lw =-5.0000000e-21
+lwl =-7.4160580e-21
+wint =1.3450000e-07
+wl =-6.5402850e-14
+ww =-3.6158840e-14
+wwl =-2.6667273e-22
+mobmod =1
+binunit =2
+xl = '0.00e+00+xln'
+xw = '0.00e+00+xwn'
+lmlt =1
+wmlt =1
+binflag =0
+dwg =-2.3440163e-09
+dwb =6.3122880e-09
*
*NMOS Noise Parameters
+noimod= 1
+kf= 1.40056062662658E-29
+af= 0.559711233377457
+ef= 0.988859925270081
*
* diode parameters
+acm =2
+ldif = 0.00
+hdif = 5.00e-07
+rsh =60
+rd =0
+rs =0
+rsc =0
+rdc =0
*
* threshold voltage parameters
*
+vth0 = '7.55e-01+vth0n'pvth0 =-2.8450000e-14
+k1 =0.7923000
+k2 =3.3844000e-02
+k3 = -7.6100000lk3 = 1.1000006e-07wk3 =1.1600000e-06
+dvt0 = 9.7212000ldvt0 =2.0210000e-06
+dvt1 =0.7776000
+dvt2 = -6.1170000e-03wdvt2 =-3.0220001e-08
+dvt0w =6.3178490e-02
+dvt1w =0.00
+dvt2w =0.2375000
+nlx =4.2079000e-07
+w0 =0.00
+k3b = 0.6043000pk3b =3.0000002e-13
+ngate =1.0000000e+23
+vfb =-0.8949800
*
* mobility parameters
*
+vsat = 6.4038830e+04pvsat =3.6100000e-10
+ua =-3.5938450e-10
+ub =2.2536685e-18
+uc =4.9139520e-11
+rdsw =8.0636650e+02
+prwb = 5.0100010e-03pprwb =5.0000000e-14
+prwg = 5.8200000e-10pprwg =-8.9052070e-15
+wr =1.0000000
+u0 =4.0425720e-02
+a0 = 1.1484001la0 = -7.0913550e-07wa0 = 9.0000010e-08pa0 =-1.0000001e-13
+keta = -6.0000000e-03lketa =-1.4150000e-08
+a1 =0.00
+a2 =0.9900000
+ags = 0.1360000lags = 1.7782144e-08pags =1.1600000e-13
+b0 =-1.1810000e-08
+b1 =1.0210000e-05
*
* subthreshold current parameters
*
+voff =-0.1516000
+nfactor =0.5101000
+cit =4.9180000e-06
+cdsc =4.6920000e-03
+cdscb =-1.3220000e-04
+cdscd =0.00
+eta0 = 0.1177000peta0 =8.0000000e-14
+etab =-3.4688240e-02
+dsub =0.5830780
*
* rout parameters
*
+pclm =1.8977000
+pdiblc1 =0.1000000
+pdiblc2 =1.7299000e-03
+pdiblcb =-0.1250000
+drout =3.0000000
+pscbe1 =4.9480000e+08
+pscbe2 =4.0500000e-06
+pvag =0.00
+delta =1.0000000e-02
+alpha0 =0.00
+alpha1 =0.00
+beta0 =30.0000000
*
* temperature effects parameters
*
+kt1 =-0.3813780
+kt2 =-3.1836310e-02
+at =2.1657000e+04
+ute =-1.2951473
+ua1 =3.7449790e-09
+ub1 =-5.4665100e-18
+uc1 =-9.7921100e-11
+kt1l =3.5527000e-15
+prt =20.0000000
*
* capacitance parameters
*
+cj = '8.13e-04+cjn'
+mj =0.2999605
+pb =0.6647527
+cjsw = '2.80e-10+cjswn'
+mjsw =0.1652686
+php =0.496524
+cjgate = '4.99e-10+cjgaten'
+tpb =1.780267e-03
+tpbsw =1.749689e-03
+tcj =1.246081e-03
+tcjsw =8.846801e-04
+js = 2.62e-07
+jsw = 0.00
+n = 1.0
+xti = 3.0
+cgdo = '1.83e-10+cgdon'
+cgso = '1.83e-10+cgson'
+cgbo = 1.0e-13
+capmod =2
+nqsmod =0
+elm =5
+xpart =0
+cgsl =2.4300000e-17
+cgdl =2.4300000e-17
+ckappa =0.6000000
+cf =0.00
+clc =5.0000000e-12
+cle =2.3309000
+dlc =-2.43e-17
+dwc =-2.43e-17
+vfbcv =-0.864
+llc =0
+lwc =0
+wlc =0
+wwc =0
+lwlc =0
+wwlc =0
+acde =0.5
+moin =15
+noff =1
+voffcv =0
+tlevc=1

有PDK吗? 有的话仿几个点的Id, 拿w/L一算就得到了啊

呃, 补充一下, 你看到变化应该是因为不同的状态, 你可以选择贴近你应用的状态去仿真, 比如strong inversion, long channel. 毕竟工程师最喜欢的话, 就是能拿等效一阶线性最好了.

弱弱的问一句,什么是PDK?

我之前仿真的二级运放,管子都是工作的饱和区,但是参数还是差异挺大的。那可不可以给定一个管子N管或者P管,以不同的漏源电压和不同的栅源电压进行扫描,然后lis文件列表查看其在各种不同状态下的阈值电压,K值参数,然后取平均,这样得到计算参数,,我觉得这样比较准确点。,。

查了个工艺文档,0.5um 5V nmos的k'=1E-4 A/V^2 , pmos的k'=0.35E-4 A/V^2 ,和自己搭个电流源算出来的差不多,实际上常见的MOS的饱和区输出电阻会随着VDS变化,一阶模型能初略算出来就不错了
用仿真倒算出来的0.5um 5V CMOS参数
CHARTER
nmos K=1.3e-4
pmos K=0.4e-4
UMC
nmos K=1.1e-4
pmos K=3.7e-5

确实可以做这个事情, 但是必要性有待商榷.因为得到这些值只是让designer心里有个方向, 具体参数还是要看tool的结果, 否则就不需要做simulation了, 对吧.
现在MOS model level已经远远超出人肉能直观理解的一阶二阶线性范围了.
个人建议.

Process design kit, which is offered by foundry to designer to smooth the simulation and silicon gap.

thanks you

Copyright © 2017-2020 微波EDA网 版权所有

网站地图

Top