You are on page 1of 17

MOSFET High Frequency Model and Amplifier Frequency Response Objectives

To review the small signal MOSFET and BJT models at low frequencies To study the high frequency MOSFET and BJT models To estimate the MOSFET and the BJT unity-gain frequency

Introduction
In the last two lectures we examined the time and frequency response of the STC circuits. As we said the importance of studying the STC circuits is that the analysis of a complex amplifier circuit can be usually reduced to the analysis of one or more simple STC circuits. The frequency response of the amplifier circuits will be explored starting next lecture. Before starting this study it will be constructive to review in this lecture the MOSFET and BJT small-signal models, and examining the highfrequency models. Also, the transistor cut-off frequency which is considered a figure of merit at high frequency operation will be estimated for both transistors.

The MOSFET small signal model

gm Vgs

+ Vgs

ro

+ B Vbs

Fig-1 in (Lec_03_Ver_01.vsd)
Figure 1 MOSFET small signal model

The small signal model of the MOSFET is shown in figure 1. Since the gate is insulated from the channel by the gate-oxide the input resistance of the transistor is infinite. The small-signal parameters are controlled by the Q-point (operating point).

The NMOS small-signal parameters may be summarized in Table 1 For PMOS the small-signal model will be the same as NMOS after replacing the electron mobility with the hole mobility and the NMOS threshold by the PMOS threshold

Table 1 MOSFET small-signal parameters

gmb Vbs

Symbol

Parameter

Value

gm =

gm

Transconductance

VTN is the NMOS threshold voltage Kn = n CoxW/L n is the electron mobility Cox is the gate oxide capacitance/area= ox/tox ox is the oxide permittivity tox is the oxide thickness W is the transistor width L is the transistor length

ID = 2K n I D V GS TN V 2

ro

Output resistance

1 +V DS ro = 1 ID I D
g mb = I D V BS

is the channel length modulation parameter Back-gate transconductance

gmb

= g m
Q po int

0<<1 is called the transconductance parameter.

back-gate

The MOSFET high frequency model


The MOSFET small-signal model discussed earlier is valid at low frequencies. This model will fail at high frequencies. At high frequencies the transistor internal capacitances should be considered. The MOSFET's internal capacitances limit the high-frequency performance of the MOSFET that means: o Limit the switching speed of the circuits in digital applications

o Limit the frequency at which useful amplification can be obtained in the amplifiers. MOSFET capacitances depend on operation region and are non-linear functions of voltages at device terminals. The MOSFETs internal capacitances are associated to either reverse biased junctions or changes of charges due to voltage difference. A physical capacitance may also exist formed by a dielectric between two conductors. Refer to Figure 2.

Source C OX
+

Gate

Drain COX

CGSO n

CGDO n+

n-type channel p-type channel Bulk CDB

C SB

NMOS device in saturation

Fig-2 in (Lec_03_Ver_01.vsd) Figure 2 NMOS Transistor Capacitances in the Saturation Region

The complete high frequency MOSFET model in the saturation region is shown in Figure 3-a.

Cgd Cgs

gmb Vbs

gm Vgs

+ Vgs

ro

Vbs +

Csb B

Cdb

Fig-3a in (Lec_03_Ver_01.vsd) Figure 3-a MOSFET high-frequency model

The different high frequency internal capacitances of MOSFET are summarized in Table 2.

Table 2 MOSFET internal capacitances

Symbol Cgs

Name
Gate-Source capacitance Gate-Drain capacitance

Origin
Physical oxide) (gate 2 WLC ox 3

Value

Cgd

Physical (drain diffusion overlap under gate oxide)

W Lov C ox Lov is the overlap length between the drain and the gate C sb = C sb 0 V 1 + SB Vo

Csb

Source-Substrate Capacitance

Depletion capacitance between the C is the source-substrate sb0 source and the capacitance when V = 0 SB substrate Vo is the junction built-in potential (0.6V to 0.8V) C db 0 C db = Depletion V 1 + DB capacitance Vo between the is the draindrain and the Csb0 substrate capacitance substrate when VDB = 0

Cdb

Drain-Substrate Capacitance

Symbol

Name

Origin
C sb =

Value
C sb 0 V 1 + SB Vo

Csb

Source-Substrate Capacitance

Depletion capacitance between the C is the source-substrate sb0 source and the capacitance when V = 0 SB substrate Vo is the junction built-in potential (0.6V to 0.8V) C db 0 Depletion C db = V capacitance 1 + DB between the Vo drain and the C is the drain-substrate sb0 substrate capacitance when VDB = 0

Cdb

Drain-Substrate Capacitance

The model in Figure 3-a may be simplified when we have the source and the substrate connected (no body effect) to the model shown in Figure 3-b

Cgd Cgs S

gm Vgs

+ Vgs

ro
Cdb

Fig-3b in (Lec_03_Ver_01.vsd)
Figure 3-b Simplified MOSFET model when the source and the substrate are connected

Further simplification may be achieved by neglecting Cdb as shown in Figure 3-c.

Cgd Cgs S

gm Vgs

+ Vgs

ro

Fig-3c in (Lec_03_Ver_01.vsd)
Figure 3-c Simplified MOSFET high-frequency model

The MOSFET Unity-Gain Frequency (fT)


As an application to the MOSFET high frequency model, let us calculate the MOSFET unity-gain frequency fT. fT is considered a figure of merit for the high-frequency operation of the MOSFET as an amplifier. fT is defined as the frequency at which the short-circuit currentgain of the common-source amplifier becomes unity. Figure 4 shows the circuit used to calculate fT, in which we used the simplified model in Figure 3c.

Please note that since we are now dealing with quantities (currents and voltages) which are frequency dependent. We will use the s-domain analysis discussed in Lecture 2. Also, we will use capital letters with lowercase subscripts to identify these quantities.

Ii

+ Vgs

Cgd Cgs

gm Vgs

Io ro

Fig-4 in (Lec_03_Ver_01.vsd)
Figure 4 Determining the short-circuit current gain

Applying the nodal analysis at the output node we can write


I o = g mV gs sC gdV gs

Recalling that Cgd (overlap drain capacitance) is small, at the frequencies of interest we can neglect the second term

I o = g mV gs

Applying the nodal analysis at the input node we may write


I i = sV gs (C gs + C gd )

Combining the input and output current equations we can write


Io gm = I i s (C gs + C gd )

hfe

g m sC Ic = 1 Ib + s (C + C ) r g m r 1 + s (C + C )r g m >> sC

hfe

hfe

o s 1+ p

1 (C + C ) r gm T = o p = (C + C )

p =

hfe

g m sC Ic = 1 Ib + s (C + C ) r g m r 1 + s (C + C )r g m >> sC 1 (C + C )r

hfe hfe

o s 1+ p

p =

T = o p =

gm (C + C )

For physical frequencies s=j, it can be seen that the magnitude of the current gain becomes unity at the frequency
fT =

T gm = 2 2 (C gs + C gd )

As we can see from the last equation. Higher fT means higher gm and lower internal MOSFET capacitances which means better amplifier operation. Typically, fT is ranging from about 100MHz for older technologies (e.g., a 5-m CMOS technology) to many GHz for

current high-speed technologies (e.g., a 0.13-m CMOS technology).

The BJT small-signal model


The small signal model of the BJT amplifier is shown in figure 5. Figures 5-a,b are for the -model, where Figures 5-c,d are for the T-model. These models are valid for both NPN and PNP transistors. For the same operating point, the BJT has higher transconductance and higher output resistance that the MOSFET.

(a)

(b)

(c)

(d)

r
E

gm V

+ V

ro

ib

r ib
E

ro

(a)
B

(b)
B

+ V

gm V re
E

ro

+ V

re

i i
E

ro

(c)

(d)

Fig-5 in (Lec_03_Ver_01.vsd)
Figure 5 small signal-models of the BJT

The small-signal parameters are controlled by the Q-point (operating point). The BJT small-signal parameters may be summarized in Table 3 Table 3 BJT small signal parameters Symbol Parameter Value

Transconductance gm

I gm = C VT VT is the thermal voltage = kT/q, which equals 25mV at room temperature. k is Boltzman's constant T is the absolute temperature in Kelvins q is the electron charge
r = V VT = T = IB IC g m

Base input resistance

Symbol

Parameter is the current gain


re =

Value common-emitter

re

Emitter input resistance

V VT = T = IE IC g m

is the common-base current gain

ro

Output resistance

ro =

V A +V CE

IC

VA

IC

VA is the early voltage.

The BJT high-frequency model

gm V

rx r

B' + C V C
E

ro

Fig-6 in (Lec_03_Ver_01.vsd)
Figure 6 The high-frequency hybrid-

model of the BJT

The high frequency hybrid- model for the BJT is shown in

Figure 6. This model is useful for signal frequencies up to a several tens of megahertz, after which a more detailed model becomes necessary.
Typically, the base-emitter junction capacitance C is in the

range of few pF to few tens of pF, while the collector-base junction capacitance C is in the range of fraction of pF to few pF The base resistor rx is added partly to account for the comparatively long internal connection from the base external connection and the actual internal base connection. A representative resistance value for this lumped resistor is in the range of 50 to perhaps 200. This resistor ordinarily can be neglected for hand estimates. Note that rx becomes the dominant input resistance for frequencies so high that C effectively short-circuits r. A second base-width modulation effect, characterized by a resistor connected between the base and collector is omitted; its influence is dominated by the collector junction reverse-bias capacitance C. The emitter junction (diffusion) capacitance C represents the charge store to support the current flow across the base.

The BJT Cutoff frequency As defined earlier, it is the frequency at which the current gain of the transistor becomes one. (i.e. no more active element). It is calculated by finding the short circuit collector current in terms of the base current. Using the high frequency model of BJT we can draw the circuit to estimate the cut-off frequency of the BJT as shown in Figure 7.

Ib
B

gm V

rx r

Vb
E

B' + V

sC V

I c = ( g m s C )V ro
C

C C

Fig-7 in (Lec_03_Ver_01.vsd)
Figure 7 Circuit used to estimate the BJT cutoff frequency

Applying nodal analysis at the input and output nodes as we did earlier. We can estimate the cut-off frequency as follows:
V + s C V + s C V r V I b = + s (C + C ) V r g m sC I hfe c = 1 Ib + s (C + C ) r Ib = Assuming Assuming hfe g m >> sC g m r 1 + s (C + C ) r g m r g m >> sC hfe 1 + s (C + C ) r 1 where p = (C + C )r hfe

o s 1+ p

Unity gain bandwidth gm (T ) = o p = (C + C )

hfe

g m sC Ic = 1 Ib + s (C + C ) r g m r 1 + s (C + C )r assuming g m >> sC 1 (C + C )r gm (C + C )

hfe hfe

o s 1+ p

where p =

Unity gain bandwidth (T ) = o p =

We can observe from the last analysis that the common-emitter current gain (hfe) frequency response is similar to a simple pole with p as the pole frequency. This may be drawn as shown in Figure 8

|hfe| [dB]
o

3 dB -6 dB/octave

[log scale]
0 dB

Fig-8 in (Lec_03_Ver_01.vsd)
Figure 8 Bode plot of |hfe|

As we can see from the last equation. Higher T means higher gm and lower internal BJT capacitances which means better amplifier operation. Typically, fT is ranging from about 100MHz to Tens of GHz.

You might also like