Loading ...
Sorry, an error occurred while loading the content.

Re: Phase Noise

Expand Messages
  • Scotty
    Hi Sam and All, My expertise on Phase noise? Ok, phase noise is bad, naughty, and no good. Do not use it. It messes up good systems. All kidding aside, my
    Message 1 of 4 , Feb 17, 2006
    • 0 Attachment
      Hi Sam and All,
      My expertise on Phase noise? Ok, phase noise is bad, naughty, and
      no good. Do not use it. It messes up good systems.
      All kidding aside, my philosophy is that, even though there are
      several varieties of phase noise, they all do exactly the same thing
      to a receiving system. They decreases the signal to noise ratio of
      the signal and the receiving detector is faced with the problem of
      processing a noisy signal. Any and all methods are used to minimize
      phase noise.
      Phase noise, by definition, is ANY noise that is in close proximity
      to a carrier frequency. Phase noise is a combination of noise
      spectra. This includes oscillator noise and uncorrelated white noise,
      whether generated in the atmosphere or within the electronic
      processing hardware. We normally attribute oscillator phase noise to
      be the combination of amplitude modulated noise, phase modulated
      noise, and uncorrelated electrical noise. These three noises will
      modulate the oscillator in different ways, but the result is, a
      carrier with noise sidebands.
      Oscillators are normally nonlinear amplifying devices, and are
      susceptable to AM and FM (phase) modulation. Some modulated noise is
      created by noise entering the oscillator via the power supply pins or
      ground. This can be minimized by good layout design and proper power
      supply filtering. The fact that the oscillator is an active device
      means that is will create and amplify uncorrelated electrical noise
      all by itself. This includes flicker noise, 1/f noise, and base
      noise, however small. Without going into these types of noises, they
      all result in AMing and PMing the carrier with noise sidebands. To
      make matters worse, the voltage controlled oscillator (VCO) is
      inherently noisy due to the fact it is a wide band device, aka, low Q
      circuit. Most of the VCO's, we deal with, use a varactor diode for
      tuning. Diodes are terribly noisy devices and nothing can be done to
      prevent their ability to convert noise to phase noise sidebands.
      As Sam stated, there is a difference between amplitude modulated
      phase noise and frequency modulated phase noise. Can a spectrum
      analyzer measure amplitude modulated phase noise and frequency
      modulated phase noise as separated noise entities? Normally, it
      cannot. The reason is, both types of phase noise occupy the same
      frequency domain and their total power will combine in the spectrum
      analyzer to display a single area of noise density. In reality, these
      two phase noises don't need to be differentiated because a final
      receiving detector system cannot tell the difference, anyway. Poo poo
      from a camel or a cow is different but they both do the same thing.
      They make a mess of things and stink up the place! And, if either was
      stuck in your face, you couldn't tell the difference. So much for the
      analogy.
      Yes, AM noise on a carrier can be reduced by hard limiting.
      However, in all cases, if there is any AM noise, it will create an AM
      to PM conversion in an active device. And, that PM noise cannot be
      reduced by limiting.
      Ok, so this is the phase noise we are talking about, amplitude
      modulated phase noise and frequency modulated phase noise. How do we
      measure them? The best way is with a narrow band spectrum analyzer.
      Bandwidth narrow enough to differentiate signals that are only a few
      hertz apart. Let me stop here and say that I never intended my
      spectrum analyzer project to be a phase noise measuring system. Due
      to the wide tuning range of my SSA, the phase noise of the first Local
      Oscillator is too high to measure most other systems' phase noises,
      unless they are very high. The nominal phase noise of the SSA is on
      the order of -90 to -96 dBc/Hz.

      Sam's link to vectron is a good one.
      http://www.vectron.com/products/appnotes/phase.htm

      I will make a few comments on their noise measurement that may not
      be quite so obvious in their app note. Looking at their figures 3 or
      4, there are two oscillators shown. They say, "If the noise of one
      oscillator dominates, its phase noise is measured directly." This
      means that the phase noise of one of the oscillators must be at least
      10 dBc better than the other, preferably 20 dB better. Let's pretend
      that oscillator 2 is called the reference oscillator and is perfect.
      That is, it has NO phase noise. (I want one of those!). Oscillator 1
      is the unit under test (UUT) and we want to measure its phase noise.
      If osc 2 is phase locked, in quadrature (90 degrees), to osc 1, the
      output of the mixer will output 0 volts (DC) and any phase noise will
      be AC. Any phase noise measured, will be from osc 1 since osc 2 is
      "perfect". Phase noise is always specified as a single sideband
      measurement. Since oscillator phase noise is on both sides of the
      carrier, both noise sidebands will be converted to baseband noise.
      Since they are converted to the same baseband frequency, they will add
      directly. They do not cancel each other out. With this type of
      measurement system, any noise that is measured in baseband is assumed
      to be double sideband. This is why they later show a correction
      factor of -6 dB when converting to single sideband.
      Since it is unlikely that anyone has a "perfect" reference
      oscillator, it is common practice to use two identical oscillators, as
      shown in the app note. The total phase noise, that is measured, is
      the combination of the two oscillators. Assuming the characteristics
      of the two oscillators are identical, each oscillator's phase noise
      are equal in power, but are not coherent and do not directly add.
      Therefore, by subtracting 3 dB from the total noise measurement, a
      single noise contribution can be obtained.
      Cheers, Scotty
    • Sam Wetterlin
      Scotty, The reason I was thinking the two AM noise sidebands would cancel upon conversion to baseband is that when noise at frequency n modulates the signal f,
      Message 2 of 4 , Feb 19, 2006
      • 0 Attachment
        Scotty,

        The reason I was thinking the two AM noise sidebands would cancel upon
        conversion to baseband is that when noise at frequency n modulates the
        signal f, one sideband is -cos(f+n)t and the other is cos(f-n)t,so
        while they are a "matched set", one on each side of f, their
        amplitudes are negatives of each other--a fact which won't show up on
        a spectrum analyzer. When they are ultimately added together when the
        spectrum folds when mixed with frequency f, it seems to me the two
        sidebands should add to zero.

        But maybe something in the final mixing with f causes reversal of one
        of their amplitudes before they are summed, so they no longer cancel.
        If so, I can't figure out what it is.

        I agree that when you step in manure you don't really care what kind
        it is. But if it turns out the manufacturer claims its product to be
        clean, but its test only detects one type of manure, that would be a
        good heads up to watch your step.

        Sam Wetterlin


        --- In spectrumanalyzer@yahoogroups.com, "Scotty" <wsprowls@...> wrote:
        >
        > Hi Sam and All,
        > My expertise on Phase noise? Ok, phase noise is bad, naughty, and
        > no good. Do not use it. It messes up good systems.
        > All kidding aside, my philosophy is that, even though there are
        > several varieties of phase noise, they all do exactly the same thing
        > to a receiving system. They decreases the signal to noise ratio of
        > the signal and the receiving detector is faced with the problem of
        > processing a noisy signal. Any and all methods are used to minimize
        > phase noise.
        > Phase noise, by definition, is ANY noise that is in close proximity
        > to a carrier frequency. Phase noise is a combination of noise
        > spectra. This includes oscillator noise and uncorrelated white noise,
        > whether generated in the atmosphere or within the electronic
        > processing hardware. We normally attribute oscillator phase noise to
        > be the combination of amplitude modulated noise, phase modulated
        > noise, and uncorrelated electrical noise. These three noises will
        > modulate the oscillator in different ways, but the result is, a
        > carrier with noise sidebands.
        > Oscillators are normally nonlinear amplifying devices, and are
        > susceptable to AM and FM (phase) modulation. Some modulated noise is
        > created by noise entering the oscillator via the power supply pins or
        > ground. This can be minimized by good layout design and proper power
        > supply filtering. The fact that the oscillator is an active device
        > means that is will create and amplify uncorrelated electrical noise
        > all by itself. This includes flicker noise, 1/f noise, and base
        > noise, however small. Without going into these types of noises, they
        > all result in AMing and PMing the carrier with noise sidebands. To
        > make matters worse, the voltage controlled oscillator (VCO) is
        > inherently noisy due to the fact it is a wide band device, aka, low Q
        > circuit. Most of the VCO's, we deal with, use a varactor diode for
        > tuning. Diodes are terribly noisy devices and nothing can be done to
        > prevent their ability to convert noise to phase noise sidebands.
        > As Sam stated, there is a difference between amplitude modulated
        > phase noise and frequency modulated phase noise. Can a spectrum
        > analyzer measure amplitude modulated phase noise and frequency
        > modulated phase noise as separated noise entities? Normally, it
        > cannot. The reason is, both types of phase noise occupy the same
        > frequency domain and their total power will combine in the spectrum
        > analyzer to display a single area of noise density. In reality, these
        > two phase noises don't need to be differentiated because a final
        > receiving detector system cannot tell the difference, anyway. Poo poo
        > from a camel or a cow is different but they both do the same thing.
        > They make a mess of things and stink up the place! And, if either was
        > stuck in your face, you couldn't tell the difference. So much for the
        > analogy.
        > Yes, AM noise on a carrier can be reduced by hard limiting.
        > However, in all cases, if there is any AM noise, it will create an AM
        > to PM conversion in an active device. And, that PM noise cannot be
        > reduced by limiting.
        > Ok, so this is the phase noise we are talking about, amplitude
        > modulated phase noise and frequency modulated phase noise. How do we
        > measure them? The best way is with a narrow band spectrum analyzer.
        > Bandwidth narrow enough to differentiate signals that are only a few
        > hertz apart. Let me stop here and say that I never intended my
        > spectrum analyzer project to be a phase noise measuring system. Due
        > to the wide tuning range of my SSA, the phase noise of the first Local
        > Oscillator is too high to measure most other systems' phase noises,
        > unless they are very high. The nominal phase noise of the SSA is on
        > the order of -90 to -96 dBc/Hz.
        >
        > Sam's link to vectron is a good one.
        > http://www.vectron.com/products/appnotes/phase.htm
        >
        > I will make a few comments on their noise measurement that may not
        > be quite so obvious in their app note. Looking at their figures 3 or
        > 4, there are two oscillators shown. They say, "If the noise of one
        > oscillator dominates, its phase noise is measured directly." This
        > means that the phase noise of one of the oscillators must be at least
        > 10 dBc better than the other, preferably 20 dB better. Let's pretend
        > that oscillator 2 is called the reference oscillator and is perfect.
        > That is, it has NO phase noise. (I want one of those!). Oscillator 1
        > is the unit under test (UUT) and we want to measure its phase noise.
        > If osc 2 is phase locked, in quadrature (90 degrees), to osc 1, the
        > output of the mixer will output 0 volts (DC) and any phase noise will
        > be AC. Any phase noise measured, will be from osc 1 since osc 2 is
        > "perfect". Phase noise is always specified as a single sideband
        > measurement. Since oscillator phase noise is on both sides of the
        > carrier, both noise sidebands will be converted to baseband noise.
        > Since they are converted to the same baseband frequency, they will add
        > directly. They do not cancel each other out. With this type of
        > measurement system, any noise that is measured in baseband is assumed
        > to be double sideband. This is why they later show a correction
        > factor of -6 dB when converting to single sideband.
        > Since it is unlikely that anyone has a "perfect" reference
        > oscillator, it is common practice to use two identical oscillators, as
        > shown in the app note. The total phase noise, that is measured, is
        > the combination of the two oscillators. Assuming the characteristics
        > of the two oscillators are identical, each oscillator's phase noise
        > are equal in power, but are not coherent and do not directly add.
        > Therefore, by subtracting 3 dB from the total noise measurement, a
        > single noise contribution can be obtained.
        > Cheers, Scotty
        >
      • william sprowls
        Hi Sam, You have the right idea, there. And you answered your own question. Just as carrier modulation causes phase reversal in a sideband, so does
        Message 3 of 4 , Feb 19, 2006
        • 0 Attachment
          Hi Sam,
          You have the right idea, there. And you answered your
          own question. Just as carrier modulation causes phase
          reversal in a sideband, so does demodulation by
          mixing. Remember, demodulation by "mixing" is also AM
          modulation. It's just that the resulting carrier is 0
          Hz. The sidebands revert to their original baseband
          frequency. The final phase reversal of one of the
          sidebands causes it to directly (in phase) add to the
          other sideband. The formuli are a tad more
          complicated than you show, but you are close enough.

          --- Sam Wetterlin <swetterlin@...> wrote:

          > Scotty,
          >
          > The reason I was thinking the two AM noise sidebands
          > would cancel upon
          > conversion to baseband is that when noise at
          > frequency n modulates the
          > signal f, one sideband is -cos(f+n)t and the other
          > is cos(f-n)t,so
          > while they are a "matched set", one on each side of
          > f, their
          > amplitudes are negatives of each other--a fact which
          > won't show up on
          > a spectrum analyzer. When they are ultimately added
          > together when the
          > spectrum folds when mixed with frequency f, it seems
          > to me the two
          > sidebands should add to zero.
          >
          > But maybe something in the final mixing with f
          > causes reversal of one
          > of their amplitudes before they are summed, so they
          > no longer cancel.
          > If so, I can't figure out what it is.

          > Sam Wetterlin



          __________________________________________________
          Do You Yahoo!?
          Tired of spam? Yahoo! Mail has the best spam protection around
          http://mail.yahoo.com
        Your message has been successfully submitted and would be delivered to recipients shortly.