330 likes | 784 Views
CMOS Digital Integrated Circuits. Lec 6 CMOS Inverters: Static Characteristics. CMOS Inverters – Static Design. Goals Understand the basic definition of basic circuit-level parameters. Understand the VTC of a CMOS inverter.
E N D
CMOS Digital Integrated Circuits Lec 6 CMOS Inverters: Static Characteristics
CMOS Inverters – Static Design • Goals • Understand the basic definition of basic circuit-level parameters. • Understand the VTC of a CMOS inverter. • Understand in detail static analysis of the CMOS inverter including circuit parameters, VOL, VOH, VIL, VIH, NMH, NML, and Vth
A B=A’ Symbol Vout VDD Logic “1” output Logic “0” output Vin 0 VDD/2 VDD Voltage Transfer CharacteristicThe Ideal Transfer Characteristic Truth Table
Generic InverterVoltage Transfer Characteristic (VTC) • Static characteristics of an inverter represented by its voltage transfer characteristic, a plot of Vout versus Vin over the range 0 to Vdd. An example transfer characteristic of an CMOS inverter is typical Vout dVout / dVin= -1 VOH Vout = Vin dVout / dVin= -1 VOL Vin VIL VIH VOH Vth
Voltage Transfer CharacteristicOperating Points • If Vin is at or near 0V (logic 0), then Vout is at or near 5V (logic 1). If Vin is at or near 5V (logic 1), then Vout is at or near 0V (logic 0). Thus, an inverter. • Points at which the input and output voltages would lie after passing through a long chain of like inverters. • Equivalent to the long chain is a loop of two inverters. • In the loop, each Vinis Vout of the other inverter, giving Vin1 =Vout2 and Vin2 =Vout1 . • Solve graphically by positioning a second VTC over the given one such that these relationships hold on the axes (The second curve is mirrored about the 45 degree line). • The intersection of these curves at a high output level gives output HIGH voltage, VOH, and at the low level gives output LOW voltage, VOL. • These two intersections are stable operating points. The intersection at the mid-level is an unstable operating point. This is the point at which Vout = Vin for the single inverter and is the inverter threshold voltage, Vthor Vinv.
Vout VOH VOL Vin Voltage Transfer CharacteristicOperating Points (Cont.)
Maximum allowable voltage Minimum allowable voltage VIH VOL VOH VIH VOL VIL Interconnect Interconnect Noise Noise Voltage Transfer CharacteristicNoise Margins • Propagation of digital signals under the influence of noise
Voltage Transfer CharacteristicNoise Margins (Cont.) • Ability to tolerate “noise” on its inputs is an important aspect of static behavior. • To quantify this concept, we define two additional voltages levels, VIH and VIL as the HIGH and LOW levels of Vin at which: dVout/ dVin = -1 • What is the justification for this? For any positive voltage superimposed on operating point value VOL,the voltage gain as represented by the |slope| is less than 1 on the entire curve upVIL. This implies that the total large signal gain over the same range is also less than 1. Thus, the negative output voltage change superimposed on VOH is smaller than the positive input change superimposed on VOL that causes it. • So a change due to noise propagating down a chain of inverters will diminish in magnitude at it propagates as long as the magnitude of the
OUTPUT INPUT H VOH H NMH VIH Transition Region VIL NML L VOL L Voltage Transfer CharacteristicNoise Margins (Cont.) change is no more than VIL-VOL. This value is called the noise margin (for LOW signal levels), NML. • Similarly, for HIGH signal levels, the noise margin is NMH=VOH-VIH. This can be represented in terms of voltage ranges as
Voltage Transfer CharacteristicNoise Margins (Cont.) • Nominal output Vout = f(Vin) • Output under noise Vout = f(Vin+Vnoise) Vout = f(Vin) + (dVout /dVin)Vnoise + hight order term (neglected) Perturbed Output = Nominal Output + GainExternal Perturbation • The nominal operating region is defined as the region where the gain is less than unity! ׳ ׳
CMOS Static ParametersThe Inverter Circuit and Operating Regions VDD S • The CMOS inverter is constructed by a pMOS and a nMOS • The pMOS source S and substrate B are both at VDD, so no body effect for either MOS. For the pMOS, VGS=Vin-VDD and VDS=Vout-VDD. • Operating Regions • nMOS Cutoff: Vin < VTn Linear: Vin≥ VTn, Vout < Vin– VTn Saturation: Vin ≥ VTn, Vout ≥ Vin– VTn • pMOS Cutoff: Vin > VDD + VTp Linear: Vin ≤ VDD + VTp, Vout > Vin – VTp Saturation: Vin ≤ VDD + VTp, Vout ≤ Vin – VTp VGS,p=Vin-VDD G VDS,p=Vout-VDD D Vin Vout D G VGS,n=Vin VDS,n=Vout S
Vout A Vout= Vin –VT0,p VDD B Vout= Vin –VT0,n nMOS in saturation C pMOS in saturation both in saturation D E Vin VIL VIH VDD+VT0,p VDD VT0,p 0 VT0,n CMOS Static ParametersVTC and Parameters IDn= IDp, VGSp= VGSn-VDD = Vin–VDD VDSp = VDSn-VDD = Vout-VDD.
CMOS Static ParametersVTC and Parameters (Cont.) • The Parameters • Due to cutoff of the nMOS and pMOS respectively, VOH = VDD VOL = 0 • VIL is at the -1 slope point. The VIL can be found by simultaneous solutions, but without iteration necessary since there is no body effect. The equations used: -IDn(SAT) = IDp(LIN) (kn/2)(VGS,n-VT0,n)2 = (kp/2 )[2(VGS,p-VT0,p)VDS,p-VDS,p2] -Substitute VGS,n=Vin=VIL, VDS,n =Vout–VDD, and kR=kn/kp into the above equation, we get kR(VIL-VT0,n)2 =2(VIL-VDD-VT0,p)(Vout-VDD)-(Vout-VDD)2
CMOS Static ParametersVTC and Parameters (Cont.) -Differentiate both sides of the above with respect to Vin, and dVout/dVIL = -1, we obtain VIL=(2Vout+VT0,p-VDD+kRVT0,n)/(1+kR) -This equation can be solved by combining with the KCL equation to obtain the numerical value of VIL. • For VIH, the equations used are: -IDn(LIN) = IDp(SAT) (kn/2)[2(VGS,n-VT0,n) VDS,n-VDS,n2] = (kp/2 )(VGS,p-VT0,p)2 • By the similar way, we can solve VIH by the following two equations kR[2(VIH-VT0,n) Vout-Vout2] =(VIH-VDD-VT0,p)2 VIH=[VDD+VT0,p+kR(2Vout+VT0,n)]/(1+kR)
CMOS Static ParametersVTC and Parameters (Cont.) • Vth is solved by equating the currents in saturation for the two devices in terms of Vin, and solving or Vth=Vin=Vout: IDn,SAT(Vin) = IDp,SAT(Vin) (kn/2)(VGS,n-VT0,n)2 = (kp/2 )(VGS,p-VT0,p)2 (kn/2)(Vin-VT0,n)2 = (kp/2 )(Vin-VDD-VT0,p)2 • Note that, in the region C, if we neglect the channel-length modulation effect, the VTC is vertical which implies infinite voltage. Therefore, the output voltage can be any value between (Vth-VT0,n) and (Vth-VT0,p), without violating the voltage conditions used in the analysis. • Actually not, If channel length modulation is included, the slope is high, but less than infinite.
CMOS Static ParametersVTC and Parameters (Cont.) • Special case: Vth=VDD/2. For this case, the CMOS VTC approach the ideal VTC. In this case, for VT0,n=|VT0p| • Therefore, which does give a quite good noise margin since, NML=VIL NMH=VDD-VIH=VIL (symmetric inverter)
CMOS Static ParametersVTC and Parameters (Cont.) • As we know to achieve this near ideal situation, kR=1. To make kR=1 • Recall these W and L are effective values of the dimensions, not drawn. Assuming equal L’s, with μn = 2 to 3 μp, Wp is typical 2 to 3 times Wn. • How to choose the kR ratio to achieve a desired inversion threshold voltage:
CMOS Static ParametersStatic (DC) Power Dissipation and Area • Static (DC) Power Dissipation • Finally, the power dissipation of CMOS is quite small, PDC = VDD (Ileaage+Isubthreshold) • Area • In terms of area, a CMOS primitive gate with n inputs require 2n devices whereas an NMOS gate requires only n+1 devices! Plus, there are more complex local interconnections. The integration density of fully-complementary MOS is governed by greater area requirements whereas NMOS density is governed by power dissipation and heat problems.