Negative-TC and zero-TC current generators

Một phần của tài liệu operational amplifier speed and accuracy improvement analog circuit design with structural methodology pdf (Trang 57 - 61)

Negative-TC current generators can be used for temperature measurements in the protection units of power OpAmps (chapter 8), and for trimming of the offset voltage temperature drift (chapter 5).

The most accurate variable with negative TC that one can find in IC components is the bipolar transistor emitter-base voltage, Vbe. It has TC that is approximately -2 mV/oC. The temperature dependence of Vbe is not perfectly linear, and has a bow-like curvature that can cause about 4 mV (0.7%) of error in the -40 to 125oC range. The process variation of Vbe

between wafer lots can be as large as 20 mV (3%).

This voltage is transformed into a current with negative TC. Fig. 3-7a shows the simplest generator of a current that is equal to Vbe/R0. Nonlinearity of the resistor TC is another source of error.

The minimum supply voltage for this generator is (Vgs + Vbe +Vsat)≈ 2 V.

The current I0 should have a linear dependence on temperature (PTAT, zero- or negative TC).

For smaller supply voltages the circuits of fig. 3-7b or 3-7c may be recommended.

Figure 3-7. Vbe/Rcurrent generators

The circuit of fig. 3-7 b has three gain stages in the negative feedback loop (M1-M4-Q1), and needs compensation (C0 and gain-degeneration diode M3). The minimum supply voltage is (Vgs + Vsat + VR1) ≈ 1.3 V.

The circuit of fig. 3-7c is utilizing the current-input amplifier (M5/M6/M10/M11) in the negative feedback loop and does not need compensation. The minimum supply voltage is (Vgs +Vsat +VR23)≈1.3 V.

The reader can verify that the circuits of fig. 3-7b and c are built using the idea of two matched currents produced from the same diode-connected transistor. These two currents appear in two corresponding loops, and the difference of the currents is amplified with a high gain amplifier.

The circuits of fig. 3-7b and 3-7c also need a start-up that can be provided by a small current via Rstart.

Zero-TC current sources are used for biasing in the applications where stability of the OpAmp quiescent current, Iq, versus temperature is more important than stability of the bandwidth. They can also be used for trimming of the voltage offset and offset temperature drift.

An obvious way to get a zero-TC current is to combine PTAT and Vbe/R currents. These currents can be provided by two separate current generators, as well as combined within one circuit as shown in fig. 3-8.

Figure 3-8. Zero-TC current generator

The drain current of M6 should be a combination of PTAT current with Vbe/R currents. Due to the transistor matching, M5 and M6 should have equal

drain currents. To keep this current as required the designer has to add an amplifier that amplifies the difference between the current in M5 and an arbitrary current (the collector current of Q0). If the amplifier gain is sufficiently high, and the amplifier output provides a finite voltage at the diode-connected transistor so that this voltage can be used to develop the matched currents in M5 and M6 then the required current in M6 will be obtained. It is not surprising that the circuit of fig. 3-8 will be described by the same graph as circuit in fig. 3-2, only with a different transfer function in the negative feedback loop. An exact equation describing the temperature behavior of this circuit may be available, yet a modern designer usually relies on physical principle of circuit operation, and the optimal operation point is found using simulations. Using this approach (and that is the one that we are trying to develop), the designer finds that the circuit can provide very good results (about 0.1% current variation with temperature) as shown on the simulation plot with N=8, R0=52 k, R1=710 k.

The circuit of fig. 3-8 is operational with supply voltage as small as 1V.

It has three gain stages in the negative feedback loop and requires compensation (C0/M7).

A zero-TC current generator having a similar structure but using a current-input amplifier (and, as a result, does not require compensation) is shown in fig. 3-9. In the circuit of fig. 3-9 the current of M12 is zero-TC. It is combined from the current though R1, which is Vbe/R1 (i.e., negative-TC) and the currents through R2and R3 that are both PTAT.

Both circuits of fig. 3-8 and fig. 3-9 need a start-up, for example, by a small current between VDD and g5 for the circuit of fig. 3-8 or between VDDand the gates of M0,M3for the circuit of fig. 3-9 as shown by the dashed line resistors.

Figure 3-9. Zero-TC current generator stable without compensation capacitors

Fig. 3-10 shows a single cell that generates all three types of current biasing generators by adding two current input amplifiers. Two new feedback loops do not require compensation.

The first additional feedback loop uses PTAT voltage across R2as the input signal. Transistors M3 and M5 with the current mirror M6,M9 and M10

form the current-input amplifier. The current through R5, if the gain of the amplifier M6/M10 is high enough, is a PTAT current as well. But the drain current of M6, and so of M9, are PTAT currents. Hence, the current of M10 is PTAT, and so is the current of M11 that can be sourced to any load. If the current of M5 is small then the current ID10 is proportional to the current through R2 with R5/R2ratio and can be mirrored to the any load by M11.

The second feedback loop uses a fraction (R1/(R1+R1A) of the Vbe voltage as an input signal for the current-input amplifier M15/M16/M17. If the amplifier biasing currents ID14 = ID13 << Vbe/(R1+R1A) then the current through M15 is proportional to Vbe:

)

( 1 1

6 1 15

A A be

D R R R

V R

I = + .

Figure 3-10. PTAT, Vbe/Rand zero-TC current generator

The minimum supply voltage for the circuit of fig. 3-10 is (VgsPMOS+Vsat+Vs1s2)≈1.2V.

Using the current-input amplifiers allows one to avoid the compensation capacitors while preserving low supply voltage capability. The biasing currents of these amplifiers cause errors in the circuit transfer coefficients. If these errors are not acceptable, than the voltage-input variants should be used with the drawbacks of requiring compensation capacitors and/or larger supply voltage.

Một phần của tài liệu operational amplifier speed and accuracy improvement analog circuit design with structural methodology pdf (Trang 57 - 61)

Tải bản đầy đủ (PDF)

(204 trang)