Single Ended and Differential Operation Basic Differential Pair Common-Mode Response Differential Pair with MOS loads

Differential Amplifiers •Single Ended and Differential Operation •Basic Differential Pair •Common-Mode Response •Differential Pair with MOS loads Ha...
Author: Horatio Green
4 downloads 2 Views 424KB Size
Differential Amplifiers

•Single Ended and Differential Operation •Basic Differential Pair •Common-Mode Response •Differential Pair with MOS loads

Hassan Aboushady University of Paris VI

References • B. Razavi, “Design of Analog CMOS Integrated Circuits”, McGraw-Hill, 2001.

H. Aboushady

University of Paris VI

1

Differential Amplifiers

•Single Ended and Differential Operation •Basic Differential Pair •Common-Mode Response •Differential Pair with MOS loads

Hassan Aboushady University of Paris VI

Single Ended and Differential Operation • Single Ended Signal: - Measured with respect to a fixed potential, usually ground. • Differential Signal: - Measured between 2 nodes that have equal and opposite excursions around a fixed potential. - The center potential is called “Common Mode” (CM).

H. Aboushady

University of Paris VI

2

Rejection of Common Mode Noise • Single Ended Signal: - Due to capacitive coupling, transitions on the clock line corrupt the signal on L1 .

• Differential Signal: - If the clock line is placed midway, the transitions disturb the differential signals by equal amounts, leaving the difference intact.

H. Aboushady

University of Paris VI

Rejection of Power Supply Noise

• Maximum Output Swing:

Vout max = VDD − (VGS − VTH )

H. Aboushady

• Maximum Output Swing:

VX max − VY max = 2[VDD − (VGS − VTH )]

University of Paris VI

3

Differential Pair Differential circuit sensitive to the input CM level.

if

Differential pair minimal dependence on input CM level.

Vin1 ≠ Vin 2

⇒ I D1 ≠ I D 2 ⇒ g m1 ≠ g m 2

I SS = I D1 + I D 2 if

Vin1 = Vin 2 I SS 2 I Output CM = VDD − RD SS 2

⇒ I D1 = I D 2 =

H. Aboushady

University of Paris VI

Differential Pair: Qualitative Analysis Vin1 > Vin 2 I D1 = I SS

H. Aboushady

M1 OFF, M2 ON

Vout1 = VDD Vout 2 = VDD − I SS RD M1 ON, M2 ON

Vout1 = Vout 2 = VDD −

I SS RD 2

M1 ON, M2 OFF

Vout1 = VDD − I SS RD Vout 2 = VDD

University of Paris VI

4

Differential Pair: Qualitative Analysis

• Maximum and minimum levels are well-defined and independent of the input CM: VDD and VDD - RD ISS • The small signal gain (the slope of Vout1-Vout2 vs Vin1-Vin2) is maximum for Vin1=Vin2 (equilibrium).

H. Aboushady

University of Paris VI

Differential Pair: Common-Mode Behavior To study Common-Mode For proper operation: • M3 in saturation • M1 & M2 in saturation

H. Aboushady

Vin1 = Vin 2 = Vin ,CM Vin ,CM ≥ VGS 1 + (VGS 3 − VTH 3 ) I Vin ,CM ≤ VDD − RD SS + VTH 2

University of Paris VI

5

Differential Pair: Output Voltage Swing For M1 & M2 in saturation:

Vout − VP ≥ Vin ,CM − VP − VTH Vout ≥ Vin ,CM − VTH Output Voltage Swing:

VDD ≥ Vout ≥ Vin ,CM − VTH To increase output swing, we choose a low

H. Aboushady

Vin ,CM

University of Paris VI

Differential Pair: Quantitative Analysis VP = Vin1 − VGS1 = Vin 2 − VGS 2 Vin1 − Vin 2 = VGS1 − VGS 2

1

Assuming M1 & M2 in saturation:

(VGS − VTH ) 2 =

VGS =

2I D

µ nCox

2I D W µ nCox L

W L

+ VTH

From eq. 1 & 2 :

Vin1 − Vin 2 =

2

Squaring the 2 sides, and since:

(Vin1 − Vin 2 ) 2 = H. Aboushady

2 I D1 2I D 2 − W W µ nCox µ nCox L L

I SS = I D1 + I D 2 2

W µ nCox L

( I SS − 2 I D1 I D 2 ) University of Paris VI

6

Differential Pair: Quantitative Analysis Previous equation can be written:

µ nCox W 2

L

(Vin1 − Vin 2 ) 2 − I SS = −2 I D1 I D 2

Squaring the 2 sides, and since:

4 I D1 I D 2 = ( I D1 + I D 2 ) 2 − ( I D1 − I D 2 ) 2 2 = I SS − ( I D1 − I D 2 ) 2

We arrive at: 2

1⎛ W W⎞ ( I D1 − I D 2 ) = − ⎜ µ nCox ⎟ (Vin1 − Vin 2 ) 4 + I SS µ nCox (Vin1 − Vin 2 ) 2 4⎝ L L⎠ 2

I D1 − I D 2 =

µ nCox W 2

L

4 I SS − (Vin1 − Vin 2 ) 2 W µ nCox L

(Vin1 − Vin 2 )

H. Aboushady

3

University of Paris VI

Differential Pair: Quantitative Analysis Let ∆Vin = Vin1 − Vin 2

and ∆I D = I D1 − I D 2

Deriving eq. 3 with respect to

∆Vin

4 I SS − 2∆Vin2 ∂∆I D µ nCox W µ nCoxW / L Gm = = ∂∆Vin 2 L 4 I SS − ∆Vin2 µ nCoxW / L For

∆Vin = 0 , Gm = µ nCox

Since:

4

W I SS L

Vout1 − Vout 2 = VDD − I D1 RD1 − VDD − I D 2 RD 2 ∆Vout = Gm ∆Vin RD ∆Vout = ∆I D RD

The small signal differential voltage gain: H. Aboushady

Av =

∆Vout W = Gm RD = µ n Cox I SS RD ∆Vin L University of Paris VI

7

Drain Currents and Overall Transconductance ∆Vin1 is when I D1 = I SS I D1 − I D 2 =

µ nCox W 2

L

∆Vin1 = VGS1 − VTH 1

(Vin1 − Vin 2 )

2 I SS W µ nCox L

∆Vin1 =

4 I SS − (Vin1 − Vin 2 ) 2 W µ nCox L

3

4 I SS − 2∆Vin2 µ C W µ nCoxW / L Gm = n ox 2 L 4 I SS − ∆Vin2 µ nCoxW / L

H. Aboushady

4

University of Paris VI

∆ID vs ∆VD Plot the input-output characteristics of a differential pair as the device width and the tail current vary:

∆Vin1 =

W ↑↑ L

2 I SS W µ nCox L

I SS ↑↑

H. Aboushady

University of Paris VI

8

Differential Pair: Small Signal Gain Av =

∆Vout W = Gm RD = µ n Cox I SS RD ∆Vin L

In equilibrium, we have

I D1 = I D 2 =

Av = g m RD

I SS 2

Where gm is the transconductance of M1 & M2.

H. Aboushady

University of Paris VI

Calculating Small Signal Gain by Superposition Set Vin2 =0 M1 forms a common source stage with a degeneration resistance

Av =

VX g R = − m1 D Vin1 1+ g m1 RS

Neglecting channel length modulation and body effect

RS = 1 / g m 2 VX g m1 RD =− Vin1 1 + g m1 / g m 2

VX RD =− 1 1 Vin1 + g m1 g m 2 H. Aboushady

5 University of Paris VI

9

Calculating Small Signal Gain by Superposition Replacing M1 by its Thévenin equivalent:

VT = Vin1

RT = 1 / g m1 VY RD = 1 1 Vin1 + g m1 g m 2

From eq. 5 & 6, we get:

(V X − VY ) due to V = in1

H. Aboushady

Rin 2 = 1 / g m 2 6

− 2 RD V 1 1 in1 + g m1 g m 2

University of Paris VI

Calculating Small Signal Gain by Superposition (V X − VY ) due to V = in1

Since:

g m1 = g m 2 = g m

− 2 RD V 1 1 in1 + g m1 g m 2

(V X − VY ) due to V = − g m RDVin1 in 1

Similarly we can say that:

(V X − VY ) due to V = g m RDVin 2 in 2

The small signal differential voltage gain: H. Aboushady

(V X − VY ) total Vin1 − Vin 2

= − g m RD University of Paris VI

10

The concept of Half Circuit If a fully symmetric differential pair senses differential inputs then the concept of half circuit can be applied.

• A differential change in the inputs Vin1 and Vin2 is

absorbed by V1 and V2 leaving VP constant

H. Aboushady

University of Paris VI

Application of The Half Circuit Concept Since VP experiences no change, node P can be considered “ac ground” and the circuit can be decomposed into two separate halves

Two common source amplifiers:

VX = − g m RD Vin1

VY = − g m RD Vin 2

VX − VY = − g m RD Vin1 − Vin 2 H. Aboushady

University of Paris VI

11

The Half Circuit Concept : Example Taking into account the output resistance (channel length modulation)

Two common source amplifiers:

VX = − g m (RD // rO1 ) Vin1

H. Aboushady

VY = − g m (RD // rO 2 ) Vin 2

VX − VY = − g m (RD // rO ) Vin1 − Vin 2

University of Paris VI

Arbitrary Inputs to a Differential Pair Conversion of arbitrary inputs to differential and common-mode components:

H. Aboushady

Common Mode

Differential University of Paris VI

12

Arbitrary Inputs to a Differential Pair: Example Calculate VX and VY if Vin1 =Vin2 and λ=0

For differential mode operation:

V X = − g m (RD // rO1 )

(Vin1 − Vin 2 )

2 ( Vin 2 − Vin1 ) VY = − g m (RD // rO 2 ) 2 VX − VY = − g m (RD // rO ) (Vin1 − Vin 2 )

For common mode operation:

I D1 = I D 2 = I SS / 2

Assuming fully symmetric circuit and Ideal Current Source: •ID1 and ID2 independent of VCM ,in •VX and VY independent of VCM ,in

The Differential pair circuit: Amplifies Vin1 − Vin 2 , Eliminates the effect of H. Aboushady

VCM ,in

University of Paris VI

Common Mode Response: Non-Ideal Current Source Assuming fully symmetric circuit with finite output impedance current source, RSS : Equivalent circuit: • Degenerated Common Source

Av ,CM =

Vout RD / 2 =− Vin ,CM 1 /( 2 g m ) + RSS

gm the transconductance of 1 transistor. H. Aboushady

University of Paris VI

13

Common Mode Response: RD Mismatch Effect Assuming M1 & M2 identical:

∆V X = −∆Vin ,CM

RD 1 /( 2 g m ) + RSS

∆VY = − ∆Vin ,CM

RD + ∆RD 1 /( 2 g m ) + RSS

Common mode to differential conversion

Effect of CM noise in the presence of resistor noise

H. Aboushady

University of Paris VI

Common Mode Response: M1-M2 Mismatch Effect g m1 ≠ g m 2

I D1 = g m1 (Vin ,CM − VP ) I D 2 = g m 2 (Vin ,CM − VP ) VP = (g m 2 + g m 2 )(Vin ,CM − VP ) RSS VP =

(g m 2 + g m 2 )RSS V (g m 2 + g m 2 )RSS + 1 in,CM

V X = − g m1 (Vin ,CM − VP )RD

VX =

− g m1 RV (g m1 + g m 2 )RSS + 1 D in,CM

VY = − g m 2 (Vin ,CM − VP )RD

VY =

− gm2 R V (g m1 + g m 2 )RSS + 1 D in,CM

V X − VY =

(g m1 − g m 2 )RDVin,CM (g m1 + g m 2 )RSS + 1

CM to DM Conversion Gain: H. Aboushady

ACM − DM =

V X − VY ∆g m RD = (g University )RParis Vin ,CM 1 m1 + g m 2 of SS +VI

14

Common Mode Rejection Ratio CMRR =

ADM

ADM : Differential Mode Gain

ACM − DM

ACM − DM : Common-Mode to

H. Aboushady

Differential Mode Gain

University of Paris VI

Cascode Differential Pair Current Source Load:

Av = g mN (rON // rOP ) Low gain 10 to 20.

To increase the gain: Cascode Differential Pair

Av = g m1 ( g m3 rO 3rO1 // g m5 rO 5 rO 7 )

H. Aboushady

University of Paris VI

15