EE 5 Lecture eferences
. eview from last lecture. Consider oltage eferences DD M BIAS oltage eference Circuit EF EF M DD, T reference EF DD T 0 WL WL W L W L Observation ariables with units olts needed to build any voltage reference
. eview from last lecture. oltage eferences BIAS oltage eference Circuit EF Observation ariables with units olts needed to build any voltage reference What variables available in a process have units volts? DD, T, BE (diode), Z, BE, t??? What variables which have units volts satisfy the desired properties of a voltage reference? How can a circuit be designed that expresses the desired variables?
. eview from last lecture. oltage eferences I D D I J D J S S ~ J Ae SX D t T m e - G0 t Consider the Diode t kt q. 8x0. 60x0 8. 6x0 K G 0. 06 pn junction characteristics highly temperature dependent through both the exponent and J S G0 is nearly independent of process and temperature k q 5 9 o termed the bandgap voltage o K
. eview from last lecture. oltage eferences BIAS oltage eference Circuit EF Observation ariables with units olts needed to build any voltage reference What variables available in a process have units volts? DD, T, BE (diode), Z, BE, t, G0??? What variables which have units volts satisfy the desired properties of a voltage reference? G0 and?? How can a circuit be designed that expresses the desired variables? G0 is deeply embedded in a device model with horrible temperature effects! Good diodes are not widely available in most MOS processes!
. eview from last lecture. Standard Approach to Building oltage eferences Negative Temperature Coefficient (NTC) X N X OUT Positive Temperature Coefficient (PTC) X P K X OUT X N KX P Pick K so that at some temperature T 0, XN KX P T T T 0 0
. eview from last lecture. Standard Approach to Building oltage eferences Negative Temperature Coefficient Positive Temperature Coefficient T 0 T X N +KX P X N KX T P T T 0 T 0 T
. eview from last lecture. Bandgap oltage eferences Consider two BJTs (or diodes) Q Q BE BE BE At room temperature BE If ln(i C /I C )= BE BE BE Δ T BE BE k I ln q I k q I ln I C C C C T 5 8.6x0 x00 5.8m BE T BE o TT 00 K 0 8.6x0 5 86μ/ o C The temperature coefficient of the PTAT voltage is rather small
. eview from last lecture. Bandgap oltage eferences Consider two BJTs (or diodes) Q Q BE BE I C (T) ~ I BE ln I t SX T m e - G0 t e BE t (T) lnj A C G0 t ~ SX E mlnt If I C is independent of temperature, it follows that BE T T BE TT 00 K 0 o k q - m 8.6x0 5 BE t G0 0.65.. 5m.m/ o C
BE and Δ BE with constant I C. 0.8 olts. eview from last lecture. BE 0.6 0. 0. 0 0 50 00 50 00 50 00 50 00 Temperature Δ BE
. eview from last lecture. First Bandgap eference (and still widely used!) DD EF Q Q P.Brokaw, A Simple Three-Terminal IC Bandgap eference, IEEE Journal of Solid State Circuits, ol. 9, pp. 88-9, Dec. 97.
. eview from last lecture. First Bandgap eference (and still widely used!) DD Current ratios is / Current not highly dependent upon T EF Q Q BE EF BE BE BE Δ BE
Thanks to Damek for the following assessment!
Most Published Analysis of Bandgap Circuits T kt T0 kt J EF= G0+ BE0 - G0 + m- ln +K ln T0 q T q J where K is the gain of the PTAT signal Negative Temperature Coefficient (NTC) X N X OUT Positive Temperature Coefficient (PTC) X P K
First Bandgap eference (and still widely used!) DD I E BE BE EF I I E E BE I C DD C OS Q Q EF I C C DD I α I I α C E I E C β α β I E I E OS EF α BE BE BE OS α α
First Bandgap eference (and still widely used!) DD Q Q EF EF BE BE α BE BE BE α I α I C I α I C E I E E I E lni ln A J ~ t lni ln A J ~ t C C G0 G0 t t E E SX SX mlnt mlnt BE BE Δ BE k ln q A A E E T
First Bandgap eference (and still widely used!) DD EF EF BE BE α BE BE BE α lni ln A J ~ t lni ln A J ~ t C C G0 G0 t t E E SX SX mlnt mlnt Q Q BE BE Δ BE k ln q A A E E T From the expression for BE and some routine but tedious manipulations it follows that k α BE G0 ~ A E m tlnt tln ln q A EJSX AE
First Bandgap eference (and still widely used!) DD EF α BE BE BE α BE BE Δ BE k A ln q A E E T EF Q Q k α BE G0 ~ A E m tlnt tln ln q A EJSX AE It thus follows that: EF α tln k A T ln q A E E G0 t ~ k A E α lni mlnt ln T SX q A E α
First Bandgap eference (and still widely used!) DD EF Q Q BE BE BE EF α α T α α A A ln q k mlnt I ln A A ln q k T α ln E E SX t G0 E E t EF ~ TlnT c T b a EF GO a SK E E E E I A A ln α q k ln A A ln α α q k b ~ m q k c
First Bandgap eference (and still widely used!) DD EF Q Q TlnT c T b a EF GO a SK E E E E I A A ln α q k ln A A ln α α q k b ~ m q k c 0 lnt c b dt d EF c b T INF e INF lnt c b INF EF T c a m q kt INF G0 EF
First Bandgap eference (and still widely used!) DD EF a bt ctlnt EF G0 kt q INF m EF Q Q.0000 Bandgap oltage Source.9500 EF.9000.8500.8000 GO.06 TO 00 BEO 0.65 m-. k/q 8.6E-05.7500.7000 00 50 00 50 00 Temperature in C
Temperature Coefficient Max NOM Min T T T TC MAX T T MIN TC ppm MAX NOM (T MIN T ) 6 0
TC of Bandgap eference (+/- ppm/c).5.5 TC.5.5 0.5 0 0 50 00 50 00 50 Temperature ange
Bamba Bandgap eference DD M M M I I I θ EF D 0 I D I D D D D [7] H. Banba, H. Shiga, A. Umezawa, T. Miyaba, T. Tanzawa, A. Atsumi, and K. Sakkui, IEEE Journal of Solid-State Circuits, ol., pp. 670-67, May 999.
Bamba Bandgap eference DD I 0 BE I = I =I BE 0 I =I 0+I I =KI K is the ratio of I to I M M M I I I θ D 0 I D I D D D D I EF EF =θi Substituting, we obtain BE BE EF=θK + 0 Δ =θk + Δ EF BE BE 0 EF a bt ctlnt
Kujik Bandgap eference EF I I X 0 I D D I D D D D [] K. Kuijk, A Precision eference oltage Source, IEEE Journal of Solid State Circuits, ol. 8, pp. -6, June 97.
Kujik Bandgap eference EF I I I 0 BE 0 0 X I =I0 I D D I D D EF=I +BE D D solving, we obtain = + EF BE BE 0 a b T c TlnT EF
End of Lecture