Mean of AC signal

Discussion in 'General Electronics Chat' started by Beginer, Jun 12, 2007.

  1. Beginer

    Thread Starter New Member

    Jun 12, 2007

    I am a real beginer in electronics it's why my question.
    What kind of equipment can measure mean of the AC signal (f<1MHz)? In my circuit this mean signal will be an input of some electronic block.

    Another question: How can I create a DC signal which amplitude value is the same as amplitude of periodic pulse? (Amplitude of pulse is changing!)

    Thank for all respons
  2. beenthere

    Retired Moderator

    Apr 20, 2004
    Don't think there is any equipment that will output the mean of an AC signal. I assume you have searched Google for such equipment. That value is not particularly useful for determining power, etc.

    Lacking the specific meter, digitizing and calculating the mean is a possibility. If the signal is a sinewave, the mean will be a percentage of the RMS value. A op amp could be trimmed to take in the RMS value and output the mean.

    You second question is a bit unclear. If it's the peak value you want, there are peak detector circuits that will hold the peak of any input. Depending on what you are dioing with the signal, digitizing the pulse might be the best solution.
  3. John Luciani

    AAC Fanatic!

    Apr 3, 2007
    Analog Devices and Linear Technology make RMS-to-DC converters. Checkout
    the specifications for the AD636 and AD637.

    (* jcl *)
  4. Ron H

    AAC Fanatic!

    Apr 14, 2005
    Mean is the same as average. The average value of a pure AC signal, with no DC component, is zero. If it has a DC component, that is the mean. You can use a lowpass filter to eliminate the AC component.
  5. gootee

    Senior Member

    Apr 24, 2007
    As Ron said, a lowpass filter will give you the mean (average) of a signal. How quickly the filter's output will reach the mean, and how well it will track the mean of a signal with a changing DC component, will depend on the filter's parameters, for a given type of signal. You can use a simple passive RC low-pass filter, or an active filter, such as one made with one or more opamps.

    If you don't want to do it with a digital solution, you could probably use an analog circuit called an "Envelope Follower".

    An envelope follower is similar to a peak detector (which holds the highest amplitude seen), except that it also can follow the peaks downward, when the amplitude decreases. A simple example might be a diode in series with the signal, followed by a capacitor and resistor in parallel which are connected between the diode's output and ground. (With an infinitely-large resistance and a perfect capacitor, that would be a peak detector.)

    Somewhat like the averaging low-pass filter, how quickly and how well the envelope detector can acquire the peak value, and how quickly it can track a changing peak value in either the up or down direction, will depend on the design.

    For signals with amplitudes that are smaller than a diode's voltage drop, you will probably need to use an opamp-based "ideal rectifier" circuit, instead of a simple diode.

    Note that it will be more difficult to always precisely acquire the actual peak value if the shape of the signal can vary, as well as the amplitude.

    Also note that if the accuracy is increased, by using a larger resistor in parallel with the capacitor, then the ability to track the peaks of signals with decreasing amplitudes is degraded (i.e. slowed).

    One solution to that problem might be to use a voltage-controlled resistance of some sort, so that the trade-off between accuracy and speed could be made "on-the-fly". But then you'd need additional circuitry, to perform the control function. If the envelope follower was in the feedback loop of a control system or servo, then perhaps the error signal could be used to derive the control signal that varied the resistance, to give a faster response/tracking speed (i.e. lower resistance) when the error was larger and better accuracy (i.e. higher resistance) when the error was smaller. You might also have to add additional low-pass filtering, after the envelope follower, depending on how much ripple you can tolerate in the quasi-DC output.

    A lot might also depend on whether or not the signal has a DC component, and exactly what information you need, regarding the envelope peaks versus the amplitude with or without the DC component. For example, you may want to also use an averaging filter (or an integrator configured as a "DC Servo") and then an opamp circuit to subtract the average from the signal, before the envelope follower, if you want only the amplitude of the AC component. Or, you might be able to use a simple AC-coupling (i.e. DC-blocking) capacitor, in the signal path, before the envelope follower.

    You will also need to think about whether or not you want to track positive peaks, negative peaks, or both.

    There are application notes at sites like,, and, that have more information about peak detectors and envelope followers. Note, when searching, that envelope followers are often mis-labeled as peak detectors.

    ALSO: I think that if you DO know the exact "SHAPE" of the signal waveform, in advance, and the shape will not vary, then you can use any of the many "True-RMS"-measuring circuits and multiply the RMS value by a fixed factor (which depends only on the waveform shape) to get the peak value. In my opinion, that might be easier to do well than trying to accurately track the peaks, themselves.

    Good luck.

    - Tom Gootee