Week 14: 30-03-2020: Electronics

EE3C11 2020 Electronics

Zoom Web Lectures

Please study the topics listed in the presentations section below.

color coded resistors

Demonstration of frequency instability

Practicum: TL081c cable driver

Presentation

The presentation “DIYTL081amplifier.pdf ” shows the evaluation of the dymamic performance of a cable driver with a TL081c operational amplifier.

Frequency compensation

Download the poster: “Frequency compensation

Introduction to Frequency Compensation

After the bandwidth of a negative feedback amplifier has been designed, the poles of the transfer are not necessarily in the desired positions.

Presentation

The presentation “Introduction to Frequency Compensation defines the term frequency compensation and presents strategies and methods for frequency compensation.

Video

Introduction to frequency compensation (9:01)

Study

Chapter 12.1

The Phantom Zero

Phantom zero frequency compensation is the most powerful frequency compensation technique.

Presentation

The presentation “Frequency Compensation: the Phantom Zero introduces the concept of the phantom zero.

Video

Phantom Zeros (0:00 - 4:30)

Study

Chapter 12.2.1

Phantom Zero Compensation of a 2nd-order System

Presentation

The presentation “Frequency Compensation: the Phantom Zero Compensation of a 2nd-order System applies the concept of phantom zero frequency compensation to the compensation of a second order system.

Video

Phantom Zeros (4:30 - 18:20)

Study

Chapter 12.2.2

Implementation of Phantom Zeros

Practical implementation of phantom zeros can be accomplishes in two ways:

  1. Active phantom zeros
  2. Passive phantom zeros

Active implementation requires the use of active differetiating circuits in the feedback loop of the amplifier.

Passive compensation requires the insertion of loop gain zeros in:

  1. The feedback network
  2. Coupling networks between the signal source and the input of the amplifier
  3. Coupling networks between the output of the amplifier and the load.

Such passive zeros are called effective if:

  1. They do not significantly affect the initial pole positions (before compensation) of the loop gain
  2. They do not introduce new dominant poles

This is usually the case if, before compensation, these feedback networks or coupling networks introduce a large attenuation in the loop gain at the phantom zero frequency.

Presentation

The presentation “Implementation of Phantom Zeros presents passive implementation techniques for phantom zeros and discusses the effectiveness of the frequency compensation.

Study

Chapter 12.2.4, 12.2.5, 12.2.6

Examples Phantom Zero Frequency Compensation

Presentation

The presentation “Examples Phantom Zero Compensation presents Examples 11.8, 12.8 and 12.9.

Video

Examples of implementation of phantom zeros (15:23)

Study

Examples 11.8, 12.8 and 12.9

Phantom zero compensation and interaction with other performance aspects

Presentation

The presentation “Phantom zero compensation and interaction with other performance aspects briefly discusses the interaction between frequency compensation with phantom zeros and other performance aspects, such as, noise, bandwidth, weak distortion, energy storage, power dissipation and overdrive recovery.

Study

Chapter 12.2.8.

SLiCAP source files Chapter 12

Download the Chapter 12 SLiCAP files

Homework

  1. If a feedback amplifier does not have an MFM characteristic, can it then always be compensated with the aid of one or more phantom zero’s? Motivate your answer.

    Answer

    • This is not true if we restrict ourselves to phantom zeros in the left half plane.
    • This is not true if we restrict ourselves to the application of passive phantom zeros, and if the source impedance, the load impedance and the feedback network do not establish an attenuation in the loop gain.
  2. The loop gain of a negative feedback amplifier has a DC value of \(-10^4\) and two poles: one at \(-10\) kHz and one at \(-50\) MHz. The ideal gain of the amplifier equals its asymptotic gain and does not depend on frequency. Our aim is to give the servo function an MFM characteristic with the largest possible bandwidth.

    1. How large is this bandwidth?

      Answer

      • The first-order bandwidth would be: \(\left|(1-L_{DC})p_1\right| =\left| -10^8\right|=100\) MHz
      • The second-order bandwidth would be: \(\sqrt{\left|(1-L_{DC})p_1p_2\right|}=\sqrt{\left| 5\cdot 10^{15} \right|}=70.7\) MHz
      • Since the second-order bandwidth is less than the first-order bandwidth, the system has a second-order low-pass transfer with a bandwidth of 70.7MHz.
    2. Is frequency compensation possible with a phantom zero in the left half of the complex plane?

      Answer

      • The sum of the two dominant poles is about -50MHz, which is less than \(-\sqrt{2}\) times the achievable bandwidth. Hence, the poles are in not MFM positions and the system can be compensated with a passive phantom zero.
    3. If so, what should be the frequency of the zero?

      Answer

      • The frequency of the zero should be (calculation in MHz):
      \[z=-\frac{(70.7)^2}{100-50}=-100\text{MHz}\]
    4. Verify your results with SLiCAP.

      Answer

      You are invited to do so!

  3. A passive feedback transimpedance amplifier converts the current of a signal source of a grounded capacitive source with a source capacitance of \(2\) pF into a voltage across a grounded resistive load with a resistance of \(2\) k \(\Omega\). The magnitude of the transimpedance should be \(50\) k \(\Omega\). The \(-3\) dB MFM bandwidth of the amplifier should at least be \(2\) MHz. An operational amplifier of which the specifications have been given below, is preferred as controller.

    1. The DC voltage gain equals \(0.5\times10^{6}\).
    2. The voltage gain has two poles, one at \(-200\) Hz and one at \(-100\) MHz.
    3. The DC output resistance equals \(50\Omega\).
    4. The input capacitance (between the inverting input and the ground) is \(3\) pF.

    Please answer the following questions, and verify your answer with SLiCAP.

    1. Can the required bandwidth be achieved with the preferred operational amplifier as controller?

      Answer

      • The poles of the loop gain are:

        \[p_1=-\frac{1}{2\pi R_f(C_i+C_s)}=-640\text{kHz}\]
        \[p_2=-200\]
        \[p_3=-100\cdot 10^6\]
      • The DC loop gain is:

        \[L_{DC}=-0.5\cdot 10^6 \frac{R_{\ell}}{R_{\ell}+R_o}=-488\cdot 10^3\]
      • The first-order bandwidth is:

        \[B_1=|(1-L_{DC})p_2| = 97.6\text{MHz}\]
      • The second-order bandwidth is:

        \[B_1=\sqrt{\left|(1-L_{DC})p_2p_1\right|} = 7.9\text{MHz}\]
      • The third pole at -100MHz is not dominant

      • Hence a bandwidth of 2MHz seems to be feasible with the given operational amplifier.

    2. If so, is frequency compensation required?

      Answer

      • Yes it is, the absolute value of the sum of the poles is less than \(\sqrt{2}\) times the achievable bandwidth.
    3. If so, give at least two ways for implementation of this compensation.

      Answer

      1. A capacitor in parallel with the feedback resistor
      2. A resistor between the source and the input of the amplifier
      3. An inductor in series with the load
    4. Implement the frequency compensation, using the most promising method.

      Answer

      • The most promising solution is a capacitor in parallel with the feedback resistor. This introduces a new non-dominant pole at a very high frequency.

      • The zero needs to be implemented at:

        \[z=-\frac{(7.9)^2}{7.9\sqrt{2}-0.64}=-5.9\text{MHz},\]
      • from which we obtain:

        \[C_{\mathrm{phz}}=\frac{1}{2\pi\times 5.9\cdot 10^6\times 50\cdot 10^3}=0.54\text{pF}.\]