{
  "id": "bode-plot",
  "title": "Bode Plot",
  "lang": "en",
  "source_lang": "en",
  "summary": "A Bode plot represents a system’s frequency response using magnitude and phase curves on a logarithmic frequency axis, supporting analysis of filters, amplifiers, and feedback control systems.",
  "category": "technology",
  "subcategory": "engineering",
  "type": "concept",
  "aliases": [
    "Bode diagram",
    "logarithmic frequency-response plot"
  ],
  "generated_at": "2026-11-04",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/bode-plot/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/bode-plot/",
      "json": "https://aiwiki.page/zh/bode-plot.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "A Bode plot is a graphical representation of the frequency response of a system, consisting of two curves: magnitude and phase plotted against frequency on a logarithmic scale. Magnitude is usually expressed in decibels, while phase is usually expressed in degrees. Bode plots are widely used in control theory, electrical engineering, and signal processing to characterize how a system amplifies, attenuates, and phase-shifts sinusoidal signals at different frequencies.\n\nThe method is named after Hendrik Wade Bode, who developed logarithmic frequency-response techniques while working at Bell Telephone Laboratories in the 1930s. Its importance comes partly from the way logarithms turn multiplication into addition: the responses of cascaded components can be combined by adding their magnitude and phase curves."
    },
    {
      "heading": "Mathematical basis",
      "text": "For a continuous-time linear time-invariant system with transfer function \\(H(s)\\), the frequency response is obtained by evaluating the transfer function on the imaginary axis:\n\n\\[H(j\\omega)=H(s)\\big|_{s=j\\omega},\\]\n\nwhere \\(j^2=-1\\) and \\(\\omega\\) is angular frequency in radians per second. Ordinary frequency \\(f\\), measured in hertz, is related by\n\n\\[\\omega=2\\pi f.\\]\n\nFor a stable system, a sinusoidal input produces a steady-state sinusoidal output at the same frequency. If\n\n\\[x(t)=A\\cos(\\omega t),\\]\n\nthen, after transients have decayed,\n\n\\[y(t)=A|H(j\\omega)| \\cos\\!\\left(\\omega t+\\arg H(j\\omega)\\right).\\]\n\nThe two parts of the Bode plot are therefore:\n\n- Magnitude plot: the size of the output-to-input ratio as a function of frequency.\n- Phase plot: the phase difference between output and input as a function of frequency.\n\nA transfer function can also be evaluated on the imaginary axis for many unstable systems, but the result does not then imply that an actual sinusoidal experiment will settle to a bounded steady-state response.\n\nMagnitude in decibels\n\nFor a dimensionless amplitude ratio, magnitude is written in decibels as\n\n\\[M(\\omega)=20\\log_{10}|H(j\\omega)|.\\]\n\nAn amplitude ratio of 1 corresponds to \\(0\\ \\mathrm{dB}\\); a ratio of 10 corresponds to \\(20\\ \\mathrm{dB}\\); and a ratio of 0.1 corresponds to \\(-20\\ \\mathrm{dB}\\).\n\nThe factor 20 is appropriate for amplitude ratios. Power ratios use \\(10\\log_{10}\\). Converting a voltage ratio into a power-gain statement additionally requires the relevant impedances to be accounted for.\n\nWhen input and output have different physical units, such as voltage and displacement, the magnitude must be stated relative to a defined reference with the same units as the transfer ratio. Taking a logarithm of a dimensional quantity without specifying a reference is incomplete.\n\nLogarithmic frequency axis\n\nEqual horizontal distances represent equal frequency ratios. A decade is a tenfold change in frequency; an octave is a twofold change. Logarithmic spacing allows responses extending across several orders of magnitude to fit on one graph.\n\nA slope of \\(20\\ \\mathrm{dB/decade}\\) is approximately \\(6.02\\ \\mathrm{dB/octave}\\). Frequency must be labeled clearly because a corner frequency given in radians per second differs numerically from one given in hertz."
    },
    {
      "heading": "Construction from elementary factors",
      "text": "A rational transfer function can be factored into a constant gain, powers of \\(s\\), and factors associated with its poles and zeros. If\n\n\\[H(s)=H_1(s)H_2(s)\\cdots H_n(s),\\]\n\nthen\n\n\\[20\\log_{10}|H(j\\omega)| = \\sum_{k=1}^{n}20\\log_{10}|H_k(j\\omega)|,\\]\n\nand, with a consistent phase convention,\n\n\\[\\arg H(j\\omega)=\\sum_{k=1}^{n}\\arg H_k(j\\omega).\\]\n\nThis additive structure makes hand construction possible and helps explain numerical plots.\n\nConstant gain, integrators, and differentiators\n\nA positive constant gain \\(K\\) shifts the magnitude curve vertically by \\(20\\log_{10}K\\) without changing phase. A negative real gain adds a phase shift of \\(180^\\circ\\), modulo \\(360^\\circ\\).\n\nAn ideal integrator \\(1/s\\) contributes:\n\n- A magnitude slope of \\(-20\\ \\mathrm{dB/decade}\\).\n- A constant phase of \\(-90^\\circ\\).\n\nAn ideal differentiator \\(s\\) contributes a slope of \\(+20\\ \\mathrm{dB/decade}\\) and a phase of \\(+90^\\circ\\). Repeated integrators or differentiators multiply these contributions by their order.\n\nFirst-order poles and zeros\n\nFor a first-order pole factor\n\n\\[H_p(s)=\\frac{1}{1+s/\\omega_p},\\]\n\nwhere \\(\\omega_p>0\\),\n\n\\[M_p(\\omega) = -10\\log_{10}\\!\\left[1+\\left(\\frac{\\omega}{\\omega_p}\\right)^2\\right],\\]\n\n\\[\\phi_p(\\omega) = -\\tan^{-1}\\!\\left(\\frac{\\omega}{\\omega_p}\\right).\\]\n\nIts straight-line magnitude approximation is flat below the corner frequency \\(\\omega_p\\) and falls at \\(-20\\ \\mathrm{dB/decade}\\) above it. At the corner, the exact magnitude is approximately \\(-3.01\\ \\mathrm{dB}\\) relative to the low-frequency level, and the phase is \\(-45^\\circ\\).\n\nA left-half-plane zero factor \\(1+s/\\omega_z\\), with \\(\\omega_z>0\\), has the opposite contributions: a high-frequency slope of \\(+20\\ \\mathrm{dB/decade}\\) and phase approaching \\(+90^\\circ\\).\n\nA common straight-line phase approximation spreads the transition from one-tenth to ten times the corner frequency. This is a plotting convention, not an exact boundary: the phase changes continuously at all positive frequencies.\n\nSecond-order factors\n\nA standard second-order low-pass factor is\n\n\\[H(s)= \\frac{\\omega_n^2} {s^2+2\\zeta\\omega_n s+\\omega_n^2},\\]\n\nwhere \\(\\omega_n\\) is the undamped natural frequency and \\(\\zeta\\) is the damping ratio.\n\nAt high frequency its magnitude falls at \\(-40\\ \\mathrm{dB/decade}\\), and its phase approaches \\(-180^\\circ\\). For \\(0<\\zeta<1/\\sqrt{2}\\), the magnitude has a resonant peak. Straight-line approximations can substantially underestimate this peak, making exact evaluation important in lightly damped mechanical and electrical systems."
    },
    {
      "heading": "Example: a first-order low-pass filter",
      "text": "A simple low-pass filter formed from a resistor and a capacitor, with the output taken across the capacitor and negligible loading, has the transfer function\n\n\\[H(s)=\\frac{1}{1+sRC}.\\]\n\nIts corner frequency is\n\n\\[\\omega_c=\\frac{1}{RC}, \\qquad f_c=\\frac{1}{2\\pi RC}.\\]\n\nBelow \\(f_c\\), the magnitude is close to \\(0\\ \\mathrm{dB}\\), so slowly varying signals pass with little attenuation. At \\(f_c\\), the magnitude is approximately \\(-3.01\\ \\mathrm{dB}\\), corresponding to an amplitude ratio of \\(1/\\sqrt{2}\\), and the phase is \\(-45^\\circ\\). Far above \\(f_c\\), the magnitude falls at approximately \\(-20\\ \\mathrm{dB/decade}\\), while the phase approaches \\(-90^\\circ\\).\n\nThis example illustrates why a cutoff frequency is not a sharp dividing line: attenuation and phase shift change gradually."
    },
    {
      "heading": "Feedback stability and stability margins",
      "text": "Bode plots are particularly useful for analyzing negative feedback. For a conventional single-loop system with loop transfer function \\(L(s)\\), the closed-loop characteristic equation is\n\n\\[1+L(s)=0.\\]\n\nThe critical frequency-response value is therefore \\(-1\\): a loop magnitude of unity combined with a phase of \\(-180^\\circ\\), modulo \\(360^\\circ\\).\n\nPhase margin\n\nA gain crossover frequency \\(\\omega_{gc}\\) satisfies\n\n\\[|L(j\\omega_{gc})|=1.\\]\n\nUsing a phase branch near \\(-180^\\circ\\), the phase margin is\n\n\\[\\mathrm{PM} = 180^\\circ+\\arg L(j\\omega_{gc}).\\]\n\nFor example, a loop phase of \\(-135^\\circ\\) at gain crossover gives a phase margin of \\(45^\\circ\\). In a conventional, well-behaved loop, this expresses the additional phase lag at that frequency needed to reach the critical condition.\n\nGain margin\n\nA phase crossover frequency \\(\\omega_{pc}\\) has loop phase equal to \\(-180^\\circ\\), modulo \\(360^\\circ\\). The corresponding gain margin is\n\n\\[\\mathrm{GM}=\\frac{1}{|L(j\\omega_{pc})|},\\]\n\nor in decibels,\n\n\\[\\mathrm{GM}_{\\mathrm{dB}} = -20\\log_{10}|L(j\\omega_{pc})|.\\]\n\nThese margins describe particular gain or phase changes, not arbitrary combinations of uncertainty. Multiple crossover frequencies, unstable open-loop poles, and unusual phase behavior require more careful interpretation. The Nyquist stability criterion provides the general encirclement test underlying these frequency-domain stability arguments; positive-looking margins alone are not a universal proof of closed-loop stability.\n\nBode plots also guide the tuning of a PID controller and other compensators. Controller adjustments reshape loop magnitude and phase to balance tracking, disturbance rejection, noise sensitivity, and robustness."
    },
    {
      "heading": "Phase, delay, and minimum-phase behavior",
      "text": "Phase is defined modulo \\(360^\\circ\\). A displayed curve may therefore contain jumps between equivalent angles, such as from \\(-180^\\circ\\) to \\(+180^\\circ\\). Phase unwrapping removes these display discontinuities by adding or subtracting multiples of \\(360^\\circ\\).\n\nA pure time delay \\(T\\) has frequency response\n\n\\[H(j\\omega)=e^{-j\\omega T}.\\]\n\nIts magnitude is unity at every frequency, but its unwrapped phase is\n\n\\[\\phi(\\omega)=-\\omega T\\]\n\nin radians. Delay can therefore reduce stability margins without changing the loop magnitude plot.\n\nMagnitude alone does not generally determine phase. For example,\n\n\\[1+\\frac{s}{a} \\quad\\text{and}\\quad 1-\\frac{s}{a}, \\qquad a>0,\\]\n\nhave identical frequency-response magnitudes but opposite phase contributions. The second factor has a right-half-plane zero.\n\nFor a minimum-phase system, magnitude and phase are related under appropriate analytical conditions. Additional all-pass factors or delays can change phase without changing magnitude. Consequently, two systems with identical magnitude plots may behave differently in feedback or in their transient responses."
    },
    {
      "heading": "Applications and measurement",
      "text": "Bode plots are used to study:\n\n- The bandwidth and frequency-dependent gain of an electronic amplifier.\n- Passbands, stopbands, and roll-off rates of filters.\n- Resonant modes in mechanical structures and actuators.\n- Loop response in process control and motion control.\n- Regulation dynamics of a switched-mode power supply.\n- Model uncertainty and performance trade-offs in robust control.\n\nPlots can be computed from analytical models or estimated experimentally. A frequency-response analyzer can apply sinusoidal excitation over a frequency range and measure output-to-input amplitude ratios and phase differences. Broadband excitation and spectral estimation offer alternative measurement methods.\n\nExperimental interpretation depends on the operating point, loading, excitation amplitude, and signal-to-noise ratio. A nonlinear device may exhibit an approximately linear small-signal response near one operating condition but a different response elsewhere. Saturation, drift, and excitation of unwanted modes can invalidate a simple transfer-function interpretation.\n\nFor discrete-time systems, the frequency response is evaluated on the unit circle:\n\n\\[H(e^{j\\Omega}),\\]\n\nwhere \\(\\Omega\\) is frequency in radians per sample. If the sampling period is \\(T_s\\), then \\(\\Omega=\\omega T_s\\). The distinct positive-frequency range normally extends only to the Nyquist frequency."
    },
    {
      "heading": "Limitations and interpretation",
      "text": "A Bode plot describes frequency-domain input–output behavior; it is not a direct plot of transient response or internal state dynamics. Bandwidth and phase margin can suggest response speed and damping for simple systems, but they do not uniquely determine overshoot or settling time.\n\nStraight-line approximations are useful for reasoning about slopes and corner frequencies, but exact curves are needed near resonances, closely spaced poles and zeros, and stability crossovers. A plotted transfer function may also hide unstable internal modes through pole–zero cancellation, so apparent input–output behavior does not always establish internal stability.\n\nFor systems with multiple inputs and outputs, separate channel Bode plots do not fully characterize interactions between channels. Matrix-based frequency-response methods and multivariable stability tools are needed when those interactions are significant."
    }
  ],
  "links": [
    "control-theory",
    "electrical-engineering",
    "signal-processing",
    "linear-time-invariant-system",
    "transfer-function",
    "decibel",
    "signal-filter",
    "resistor",
    "capacitor",
    "negative-feedback",
    "pid-controller",
    "electronic-amplifier",
    "process-control",
    "switched-mode-power-supply",
    "robust-control",
    "signal-to-noise-ratio"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}