|
Site Visited 502733 times | Page Visited 50 times | You are in : Etantonio/EN/Universita/4anno/Microelettronica/ |
Characterization of the circuits and estimates of the performances 1) Esteem of the resistance: The resistance introduced from a layer of material
conductor is A device MOS moreover can often be characterized from its
resistance
2) Condenser CMOS: Condenser CMOS is formed exclusively from substrate p-doped, SiO2 and gate metallic or in polisilicio, can be introduced the following situations: to) accumulation:
to the gate ones it comes applied a negative tension, it attracts gaps
to oxide therefore the ability is only that one of the
c) inversione : a advanced positive tension to the Vt recalls electrons from the substrate which come to constitute a channel, such electrons but do not succeed to follow variations of mark them of income to advanced frequencies to 100Hz therefore for inferior frequencies the ability are C0 while for advanced frequencies it will be necessary to consider also ability Cdep and therefore the total ability is Cgb that it is also the possible minim.
3) gate Abilities to to a MOS:
In particular the ability to the gate ones is Cg = Cgb Cgs Cgd where Cgs and Cgd are understandings between the gate ones and the channel therefore if this last one is absent is absent also they therefore to second of the region of operation of the MOS is had: to) interdiction : the channel is absent therefore the only ability is givenC gb from the series of 0C and Cdep b) non-saturazione :
the channel is formed therefore Cgb = 0 while c) saturazione :
the channel is not formed from the side of drain
therefore Cgd = 0 while To the aim to calculate the delays in the digital circuits them C g =C 0 canbe approximated .
It is Cja that Cjp is but function of the present
tension to the heads of the splice second
4) Analysis and solutions for distributed circuits RC: In the case of logons in polisilicio many long or logons
in metal heavy loaded, are had of the forts delay of propagation that
can be study by means of the theory of the lines thinking the line to
you subdivided in infinite having drafts infinitesimal thickness,
resistance to length unit r and ability to unit of length c, derive
that endured delayt x from
one marks them that a discreet analysis crosses a distance along The quadratic course of the delay regarding the distance is much heavy one, much to render advisable in some cases to insert of the buffer along the line which regenerate mark them, the delay that is obtained also considering an inner delay to the buffer of some ns, is always inferior to the delay that is had in absence of the buffer. Alternatively the source in chip so that it tos be distant little from the adressee, the this can be decided is the reason for which the clock sovente it is found to the center of the chip. A model for the calculation of the delay is: ![]() 5) regarding characteristic Times the commutation of one door: tr necessary time so that the escape passes from 10% to 90% of its stationary value tf necessary time so that the escape passes from 90% to 10% of its stationary value td time that elapses between the variation of 50% of the income and the variation of 50% of the escape, comes also defined the times tdr and tdf relati you to the increase and the reduction 6) analytical Model of delay of inverter a CMOS:
Evidently therefore the calculation of the time of
reduction is subdivided in two is made, in which the nMOS it is in
saturation (…is resolved for separation of variable the
(…with the expression ofthe DS for the region of triodo and
integrating between VDD â "
Vtn and 0.1VDD is obtained As far as the times of delay they are simply the half of
respect times to you of climb or reduction and can be expressed in the
shape
7) Delay of one logical door: The delay of simple a logical door can be calculated constructing inverter an equivalent in which that is the dimensions of pull-down and the those of the pull-up reflect the distances that effectively are active in the pull-down and in
it is had that for the pull-down it is necessary
to consider the series of the 3 nMOS and therefore of their
conductances therefore for bn1 = bn2 = bn3
8) Infuence of the slope of the shape of wave of income on the delay of one logical door: The slope of the shape of income wave can modify the delay
of a door, in particular if the foreheads are the much steep then
delay are determined mostly from the distances of load and unload
while if the varied income slowly then contributes to the delay of the
escape, in particular In analogous way
9) Models switch in order to calculate the delay of one door: They are models for the calculation of delays of complex doors, the MOS like resistances that they load and they unload of the abilities, in the fattispecie are based on considering are assert following the 3 models to you: to) RC
b) Penfield-Rubenstein
c) Slope It defines the intrinsic time of climb like the time of climb that would be had if in income came applied a step. The time of effective climb comes subdivided for the intrinsic time of climb. d) Penfield-Rubenstein Slope Delay Combination of the Slope and the Penfield-Rubenstein is one.
10) Model in order to calculate the delay of circuits many complexes: The delay of one given door comes determined by means of a
simulator from which an equation of the type
11) Body â " effect in the delays of the logical doors: NMOS the more neighbors to the escape of a door nand have V sba ¹ necessarily 0, that ago that they are more discs of a valve in the commutations effect that becomes obvious if the escape ability is comparabile to the inner abilities. We consider the two following doors nand: the nand up he has all the nMOS ignited except that one up, of it he derives that its source is to mass and therefore when the income passes to 1 the escape commuta quickly to 0. In nand low instead all the nMOS up they are active while that one low is initially interdetto, follows some that the abilities will be all charges to the high level and when the income high-pass filter, will have to be unloaded all before in order to low concur the commutation of the escape from the high level with the level that therefore happens in a time longer than how much is not had up for the nand. In order to diminish the effects of the body â " effect the following strategies can be used: to) diminishing the abilities to the inner nodes b) to place the having MOS it marks them that they arrive for last more close possible to the escape
12) Dimensioning of the transistors in doors CMOS: In kind W p= 2 is made Wn so as to to equalizzare the times of it loads and discharge, but in such a way is had an increase of the occupied area and the dynamics dissipation therefore when possible the following strategies are applied: to) a brace of inverter places in having series withW p = 2W n from place to the same delay generated from a brace of inverter Wp = Wn , in order to demonstrate it is R and Ceq respective the resistance and the ability equivalent of a having MOS unitary dimensions, and considers the following delays for configurations:
Being Wp = 2Wn the ability to the nMOS is
equal to that one of the pMOS therefore will be had 2Ceq , of it derives that the time of
reduction is b) In order to pilot cargos a lot elevates to you can be inserted of the inverter in cascade of gradually increasing dimensions in way but to diminish the delay, the area and the dissipation. It is to the increase percentage of the dimensions of a
inverter regarding the previous one and td the introduced medium delay from inverter to minimal
dimensions, then a delay introduced from every stage is atd while the delay total is natd that if we place 13) Power dissipated from a circuit CMOS:
to) b) Pd is the dynamics power dissipated in order to load and to
unload the cargo ability, considering in income a having wave ideal
quadrant period tp has c) Psc is the power dissipated in short circuit when the income is
not a wave ideal quadrant but more really it is introduced like a
repetition of trapezes like evidenced in figure, has The calculation of the dissipated power must be carried out for every ability to the frequency for which it is operating.
14) Charge - sharing:
15) Yield of a wafer: Draft of the relationship between the n° of chip good presents on wafer and n° a total of chip present on the same one, it is function of the area To of the chip and of the density of D defects, the following models are had: to) Seed b) Murphy The yield can be increased increasing the redundancy of the circuits. |