Transcript Chapter 5 Low-Noise Design Methodology

Chapter 6
Low-Noise Design Methodology
1
• Low-noise design from the system designer’s viewpoint is concerned
with the following problem: Given a sensor with known signal, noise,
impedance, and response characteristics, how do we optimize the
amplifier design to achieve the lowest value of equivalent input noise?
• The answer is the amplifier portion of the system must be matched to
the sensor. This matching is the essence of low-noise design.
Circuit Design
• When designing an amplifier for a specific application, there are many
specifications to be met and decisions to be made. These include gain,
bandwidth, impedance levels, feedback, stability, dc power, cost and
signal-to-noise ratio requirements. Amplifier designers can elect one
of 2 paths:
• The wrong approach is to worry about the gain and bandwidth first and
later in the design power they check for the noise.
• Alternatively, one can design the system with initial emphasis on noise
performance.
• Although there are many low-noise devices available they do not
perform equally for all signal sources.
2
• To obtain the optimum noise performance , it is necessary to select the
proper amplifying device (FET, BJT, or IC) and operating point for the
specific sensor or input source.
• Feedback and filtering can then be added to meet the additional design
requirements
Design Procedure
• Select the input stage device, discrete or IC, BJT or FET
• Select the operating point
• If preliminary analysis shows that the noise specification can be met, a
circuit configuration (CS, CG, CD) can be selected and the amplifier
designed to meet the remainder of the circuit requirements
• Noise is essentially unaffected by circuit configuration and overall
negative feedback. Therefore the transistor and its operating point can
be selected to meet the circuit noise requirements.
• Then the configuration or feedback can be determined to meet the
gain, bandwidth, and impedance requirements.
• This approach allows the circuit designer to optimize for the noise and
other circuits requirements independently.
3
• After selecting a circuit configuration, analyzing it for non-noise
requirements may indicate that it will not met all the specifications. If
the bandwidth is too narrow, more stages and additional feedback can
be added, the bias current of the input transistor can be increased or a
transistor with a larger fT can be selected. The noise can be
recalculated to see if it is still within specifications. This iterating
process ensures obtaining satisfactory noise performance and prevents
locking in on a high-noise condition at the very start of the design.
• The ultimate limit of equivalent input noise is determined by the sensor
impedance and the first stage, Q1, of the amplifier.
• The source impedance Zs(f) and noise generators En(f) and In(f)
representing Q1 are each a different function of frequency. Initial steps
in the the design are the selection of the type of input device, such as
BJT, FET or IC, and the associated operating point to obtain the
desired noise characteristic as described above.
• In the simplest case of a resistive source, match the amplifier’s
optimum source resistance Ro to the resistance of the source of sensor.
• If the amplifier is operated over a band of frequencies, the noise must
be integrated over this interval.
4
• By changing devices and/or operating points, theoretical performance
can approach an optimum.
• The noise of the first stage must be low to obtain overall low system
noise.
• The following stages cannot reduce noise no matter how good.
• Subsequent stages can add noise so the design of these stages must be
considered for a low-noise system. The usual problem is that there is
not enough gain-bandwidth in the first stage to provide high gain with
the bandwidth needed. Then the noise of the following stages may
contribute.
• After selecting the input stage the circuit is designed. Set up the
biasing, determine the succeeding stages, the coupling networks and
the power supply.
• Then analyze the noise of the entire system, including the bias-network
contributions to ensure that you can still meet the noise specifications.
• Finally add the overall negative feedback to provide the desired
impedance, gain and frequency response.
5
Optimum Source Resistance
• The point at which the total
equivalent input noise
approaches closes to the
thermal noise curve in the
figure is significant
• At this point the amplifier
adds minimum noise to the
thermal noise of the source.
This optimum source
resistance is called Ro and is
defined as Ro  En / I n ,
where En  I n Rs .
6
Selection of an Active Device
• An active input device can be an IC with a bipolar or FET input stage
or a discrete transistor. Selection depends primarily on the source
impedance and frequency range. To assist in decisions in choosing the
right active input devices, a general guide is shown below:
• At the lowest values of source resistance, it is usually necessary to use
transformer coupling at the input to match the source resistance to the
amplifier Ro. Bipolar transistor s and bipolar input ICs are most useful
at midrange impedances. Adjustment of Ro to match the source
impedance is made by changing the transistor collector current with
higher currents for lower Ro as given by Ro  (0.025  0 ) / I C
7
• At higher values of source resistance, FETs are more desirable because
of their very low noise current In. In some instances, they are even
preferred when a low En is desired.
• When operating with a very large range of source resistance such as in
an instrumentation amplifier application, a JFET is generally preferred
for the input stage.
• A typical JFET has an En slightly larger than a BJT, but its In is
significantly lower.
• Another advantage of JFET is its higher input resistance and low input
capacitance; thus it is particularly useful as a voltage amplifier.
• For the highest source resistances the MOSFET with its extremely low
In has an advantage. The MOSFET may have up to 10 to 100 times the
1/f voltage En of a JFET or BJT. As processing techiques have
improved the MOSFETs are becoming more attractive as low noise
devices.
• The advantages of include low cost and compatibility with digital IC
process. It is often desirable to combine a MOSFET input signal
amplification stage with DSP on the the same chip.
8
• IC amplifiers are usually the first choice for amplifier designs because
of their low cost and ease of use. In selecting the one IC for your
design, the source impedance and the transistor type must still be
considered. All of the preceding statements about input stages apply
when selecting the IC for your design.
• For lower source impedances, select a bipolar unit IC and for higher
impedances a FET input IC.
• If state-of-the-art performance is need the use a discrete BJT or JFET
stage ahead of the IC.
Transformer Coupling
• To couple the detector and amplifier in an electronic system, it is
sometimes better to use a coupling or input transformer. When it is not
possible to achieve the necessary noise figure using device selection,
transformer coupling may be the solution. Very low resistance sources
can cause this problem.
• The use of an input transformer between the detector and amplifier
improves the system noise performance by matching the sensor
resistance with the amplifier’s optimum source resistance Ro.
9
• Consider the secondary circuit as shown below that contains noise
generators En and In. We have identified Ro  En / I n . When these
quantities are reflected to the primary as En’ and In’ we obtain En'  En / T
where T is the turn ratio.
10