模电设计-电流镜负载的差分放大器

模拟集成电路课程设计报告电流镜负载的差分放大器摘要:差分放大器是最重要的电路创造之一，它可以追溯到真空管时代。有于差动放大具有很多有用的特性，像对差模输入信号的放大作用和对共模输入信号的抑制作用，所以它已经成为当代高性能模拟电路和混合信号电路的主要选择。电流源在差分放大器中广泛应用，电流源起一个大电阻的作用，但不消耗过多的电压余度。在模拟电路中，电流源的设计是基于对基准电流的“复制”，稳定的基准电流那么由一个相对复杂的电路来产生。在电流镜中，只需调整MOS管的W/L就能获得不同的、精确的复制电流。在本课程设计中，将根据典型电流镜负载差动对中，增益、带宽与MOS管W/L之间的关系，获得满足要求的放大器。一.设计目标-1-二．单个MOS管的的特性-2-2.1、NMOS特性仿真-2-2.2、PMOS特性仿真-4-三．电路设计与参数推导-6-3.1电路设计：-6-3.2手工推导参数-7-四．差分放大器仿真-9-4.1、HSPICE仿真：-9-4.2、器件参数修改-10-4.3仿真波形-11-4.2、共模电平的范围：-13-4.3数据比照-16-五.总结-17-一.设计目标设计一款差分放大器，要求满足性能指标：负载电容对管的m取4的倍数低频开环增益>100GBW(增益带宽积)>30MHz输入共模范围>3V功耗、面积尽量小参考电路图：二．单个MOS管的的特性MOS管是金属(metal)—氧化物(oxid)—半导体(semiconductor)你场效应晶体管，或者称是金属—绝缘体(insulator)—半导体。MOS管的source和drain是可以对调的，他们都是在P型backgate中形成的N型区。在多数情况下，这个两个区是一样的，即使两端对调也不会影响器件的性能。这样的器件被认为是对称的。2.1、NMOS特性仿真电路图如下：HSPICE仿真:*ProjectNMOS*Inifile:*Options:-h-d-n-m-z-x-c6*Levels:*.prot.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'tt.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'res.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'cap.unprotM1I1VDVB00NVNL=1UW=10UM=1VBSVB01VDSVD05*DICTIONARY1*GND=0.optionspostlist.dcVDS050.1.op.printi1(M1I1).END仿真波形：仿真得出的数据：subcktelement0:m1i1model0:nvnregionSaturatiid18.6184uibs-4.227e-22ibd-23.6496avgs1.0000vds5.0000vbs0.vth830.1150mvdsat125.6460mvod169.8850mbeta1.4749mgameff894.5056mgm192.1882ugds1.2418ugmb72.1958ucdtot12.5800fcgtot24.0149fcstot31.6174fcbtot34.8211fcgs18.4311fcgd2.8784f参数计算：由仿真结果可以算出：=0.0352.2、PMOS特性仿真电路图如下：HSPICE仿真：*ProjectPMOS*Inifile:*Options:-h-d-n-m-z-x-c6*Levels:*.prot.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'tt.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'res.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'cap.unprotM1I1VDSVGSVDDVDDNVPL=1UW=10UM=1V1I2VGS04V1I3VDS05V1I4 VDD0 5*DICTIONARY1*GND=0.optionspostlist.dcV1I3050.1*V1I23.550.1.op.printi1(M1I1).END仿真得出的数据：subcktelement0:m1i1model0:nvpregionLinearid0.ibs0.ibd0.vgs-1.0000vds0.vbs0.vth-899.3391mvdsat-136.0660mvod-100.6609mbeta471.1383ugameff384.0716mgm0.gds46.9930ugmb0.cdtot29.4825fcgtot30.8144fcstot30.3463fcbtot40.6913fcgs17.7584fcgd12.8808f参数计算：由仿真结果可以得出=0.0729三．电路设计与参数推导3.1电路设计：3.2手工推导参数由库文件可以得到上述除了λn、λp外的器件参数，λn、λp可以由mos管的仿真得到。由性能指标低频开环增益>100，GBW(增益带宽积)，CL=1pf可得求得我们设计中取。另一方面求得。因此ID可取的范围为。所以我们取单边电流ID=25mA。忽略沟长调制效应，求得。这里我们取下一步应该确定M3，M4，M5，M6的宽长比。由可得取负载管M3，M4的过驱动电压Vod=300mV可得。这里我们取同理取电流源管M5，M6的过驱动电压Vod=500mV可得。这里我们取理论开环增益:理论单位增益带宽:理论共模输入电平最小值:理论共模输入电平最大值:理论共模输入电平范围最大值为V=4.5171-1.4231=3.0940V四．差分放大器仿真电流源负载的差分放大器整体电路图：4.1、HSPICE仿真：*ProjectDC_NMOS*Inifile:*Options:-h-d-n-m-z-x-c6*Levels:*.prot.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'tt.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'res.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'cap.unprotcoutvout01pfM3M3dg-M4g-M1dM3dg-M4g-M1dvddvddNVPL=1UW=8UM=1M4voutM3dg-M4g-M1dvddvddNVPL=1UW=8UM=1M1M3dg-M4g-M1din1M1s-M2s-M6d0NVNL=1UW=16UM=1sM2voutin2M1s-M2s-M6d0NVNL=1UW=16UM=1M6M1s-M2s-M6dM6g-M5dg00NVNL=1UW=4UM=1M5M6g-M5dgM6g-M5dg00NVNL=1UW=4UM=1V1vdd05VI10M6g-M5dgDC=50uAVin1in202.5Vac=0.5vVin2in102.5Vac=0.5v180*DICTIONARY1*GND=0.OPTIONSPROBE.OP.dcV1050.1.acdec101k100meg.optionslistnodepost.printacvdb(vout).end4.2、器件参数修改仿真后波形图如下图：发现波形锁呈现出来的增益带宽积为43.1M，已经到达题目的要求，但是低频开环增益为39.6db左右，即放大倍数小于100倍，达不到题目要求，所以尝试加大M1,M2的M值，因为题目要求对管的m为4的倍数，所以各自增大到4.修改后的hspice仿真如下：*ProjectDC_NMOS*Inifile:*Options:-h-d-n-m-z-x-c6*Levels:*.prot.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'tt.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'res.lib'D:\ePD\05model\05model\h05hvcddtt09v01.lib'cap.unprotcoutvout01pfM3M3dg-M4g-M1dM3dg-M4g-M1dvddvddNVPL=1UW=8UM=1M4voutM3dg-M4g-M1dvddvddNVPL=1UW=8UM=1M1M3dg-M4g-M1din1M1s-M2s-M6d0NVNL=1UW=16UM=4M2voutin2M1s-M2s-M6d0NVNL=1UW=16UM=4M6M1s-M2s-M6dM6g-M5dg00NVNL=1UW=4UM=1M5M6g-M5dgM6g-M5dg00NVNL=1UW=4UM=1V1vdd05VI10M6g-M5dgDC=50uAVin1in202.5Vac=0.5vVin2in102.5Vac=0.5v180*DICTIONARY1*GND=0.OPTIONSPROBE.OP.dcV1050.1.acdec101k100meg.optionslistnodepost.printacvdb(vout).end4.3仿真波形图中明显能看出放大器的单位增益带宽超过42db即放大倍数有125.89，同时带宽也增大了，增大为59.1M，明显已经满足设计中的带宽和增益的要求。仿真的.lis文件局部数据如下：*******projectdc_nmos******acanalysistnom=25.000temp=25.000******xfreqvoltdbvout1.00000k42.33431.25893k42.33431.58489k42.33421.99526k42.33422.51189k42.33423.16228k42.33413.98107k42.33405.01187k42.33386.30957k42.33357.94328k42.333110.00000k42.332412.58925k42.331215.84893k42.329519.95262k42.326625.11886k42.322231.62278k42.315139.81072k42.303950.11872k42.286263.09573k42.258379.43282k42.2144100.00000k42.1458125.89254k42.0392158.48932k41.8754199.52623k41.6279251.18864k41.2625316.22777k40.7397398.10717k40.0218501.18723k39.0824630.95734k37.9154794.32823k36.53731.00000x34.98101.25893x33.28621.58489x31.49041.99526x29.62422.51189x27.71073.16228x25.76583.98107x23.80065.01187x21.82196.30957x19.83427.94328x17.840210.00000x15.841212.58925x13.837215.84893x11.827519.95262x9.810125.11886x7.781331.62278x5.735639.81072x3.664350.11872x1.554863.09573x-609.6717m79.43282x-2.8499100.00000x-5.18714.4、共模电平的范围：那么在上面的hspice语言中最后的修改如下：.OPTIONSPROBE.OP*.dcVin1050.1*.printacvM(vout).printi1(M3).optionslistnodepost.END出来的波形如下：可以看出当输入共模电平到达1.6V时所有的MOS管已处于饱和状态。输出电流根本比拟接近25mA了。以下数据是当共模输入电平为Vincom=Vgsm1+Vdsm6=1.6V时仿真得到的数据****mosfetssubcktelement0:m30:m40:m10:m20:m60:m5model0:nvp0:nvp0:nvn0:nvn0:nvn0:nvnregionSaturatiSaturatiSaturatiSaturatiSaturatiSaturatiid-23.3008u-23.3008u23.3008u23.3008u46.6016u50.0000uibs1.185e-211.185e-21-16.2402a-16.2402a-1.058e-21-1.135e-21ibd4.6881a4.6881a-113.8455a-113.8455a-1.0139a-2.5571avgs-1.2392-1.23921.06351.06351.35211.3521vds-1.2392-1.23923.22433.2243536.4570m1.3521vbs0.0.-536.4570m-536.4570m0.0.vth-898.4776m-898.4776m1.02021.0202840.6927m839.6145mvdsat-325.6130m-325.6130m71.2561m71.2561m310.9569m311.4916mvod-340.7521m-340.7521m43.3089m43.3089m511.4315m512.5097mbeta359.5878u359.5878u9.5044m9.5044m576.1957u576.2126ugameff384.0653m384.0653m908.7034m908.7034m894.5198m894.5210mgm119.6053u119.6053u402.3115u402.3115u159.6314u174.0003ugds2.0864u2.0864u990.6995n990.6995n8.9999u2.0796ugmb29.2127u29.2127u120.0342u120.0342u56.6258u61.4295ucdtot11.5100f11.5100f83.0119f83.0119f7.0004f6.2662fcgtot20.3759f20.3759f131.8157f131.8157f9.6359f9.6096fcstot26.4459f26.4459f163.4455f163.4455f13.1528f13.1500fcbtot28.1764f28.1764f199.0483f199.0483f16.2186f15.5264fcgs17.7019f17.7019f91.4478f91.4478f7.7484f7.7420fcgd1.6795f1.6795f18.1567f18.1567f1.0386f1.0019f以下数据时当共模输入电平为Vincom=Vgsm1+Vdsm6=5V时仿真得到数据：****mosfetssubcktelement0:m30:m40:m10:m20:m60:m5model0:nvp0:nvp0:nvn0:nvn0:nvn0:nvnregionSaturatiSaturatiSaturatiSaturatiSaturatiSaturatiid-26.0692u-26.0692u26.0692u26.0692u52.1384u50.0000uibs1.326e-211.326e-21-24.5995a-24.5995a-1.184e-21-1.135e-21ibd4.7720a4.7720a-28.2928a-28.2928a-6.1486a-2.5571avgs-1.2614-1.26141.74961.74961.35211.3521vds-1.2614-1.2614488.1626m488.1626m3.25041.3521vbs0.0.-3.2504-3.25040.0.vth-898.4814m-898.4814m1.60661.6066837.0766m839.6145mvdsat-343.1824m-343.1824m144.8909m144.8909m312.7809m311.4916mvod-362.9668m-362.9668m142.9751m142.9751m515.0476m512.5097mbeta358.2060u358.2060u2.3586m2.3586m576.2513u576.2126ugameff384.0646m384.0646m955.8307m955.8307m894.5218m894.5210mgm125.7337u125.7337u289.6421u289.6421u180.5302u174.0003ugds2.2816u2.2816u4.1017u4.1017u764.2799n2.0796ugmb30.7391u30.7391u45.7557u45.7557u63.5969u61.4295ucdtot11.4719f11.4719f20.1149f20.1149f5.6403f6.2662fcgtot20.3750f20.3750f35.3235f35.3235f9.7477f9.6096fcstot26.4458f26.4458f37.8057f37.8057f13.1491f13.1500fcbtot28.1257f28.1257f40.1024f40.1024f14.7630f15.5264fcgs17.7131f17.7131f28.2358f28.2358f7.7386f7.7420fcgd1.6796f1.6796f3.8744f3.8744f1.1407f1.0019f所以由以上数据我们可以得出实际仿真出输入共模电平范围为V=5-1.6=3.4V。这个值比理论计算出来的值大，这是由于实际中我们忽略了放大管M1，M2的衬偏效应。实际上可以看出在Vin=1.6V时，M1，M2管已经存在衬偏效应，此时放大管的阈值电压VTH=1.0202，当共模电平增大时，VTH也随之增大，直到输入共模电平增大到5V，此时的阈值电压增大到VTH=1.6066V，并且Vds6从536.4570mV增大到3.2504V。从而导致即使输入共模电平增大到5V，负载管仍没有进入线性区。一下数据是仿真得到功耗：****voltagesourcessubcktele