Connect with us

Noise problem in existing circuit.

Discussion in 'Electronic Design' started by Jamie, Sep 7, 2012.

Scroll to continue with content
  1. Jamie

    Jamie Guest

    Please look at circuit in question.

    Side Band detector

    || audio out
    | 8.4 + | ||
    | --- |
    | --- 0.022|
    + | |
    7.8Mhz RX sig .87 |/ GND |
    inject here +----+----------+| .-.
    | | |> | |
    .-. .-. 0.226 | |
    | | | | |7.8mhz carrier '-'
    10k| | | |100k .-.inject here |
    '-' '-' | | 9.4V
    | | | |
    GND '-'
    9.47 |
    .-. |
    | | ---
    680 | | ---1u
    '-' |

    THe above circuit is also used in the detection section of the AM
    receiver and that works fine there because the 7.8Mhz carrier is not
    active at the emitter and there is no noise.

    Problem, everything in this circuit seems to be fine with the manual
    etc.. The problem is, while in a SSB mode and very low 8.7mhs RX signal,
    the injected 8.7Mhz carrier at the emitter is generating a level of
    white noise, which can he heard in the audio section when advancing
    gain and reducing RF gain to nothing, it seems to be excessive. This
    noise would remind you of listening to FM with no signal, rush noise.

    Using the Tek 485, I see the carrier (BFO if you want), at 2.1VPP
    with a .5us sweep. It matches exactly what should be there, according to
    the manual, however, It appears that this carrier is phase shifting some
    how? The 485 is displaying it in a widen fat line view, which tells me
    it is being Phase/Freq modulated some how with possibly some foreign
    noise ?
    I am on vacation so I don't have access to the lecroy I would normal
    use to measure this. I do have a Regal I could connect up but I have a
    hard time getting away from my Tek485 :)

    Looking at the circuit above, the audio out, my scope at 50mv input
    with the 10:1 probe hardly shows any noise but it is there because the
    next audio stages amplify it nicely. I do have active fet probes with a
    preamp if that is needed. I also have other RF tools if I have to get
    out the big guns.

    THis noise is reminding me of what you'd see at the base of a BJT in
    R.F. surroundings or just lots johnson noise, but I can't seem to find
    noise at the base, but a little barely showing at the emitter.

    Could there be a problem with the 7.8mhz inject carrier at the emitter
    , producing this noise, due to some other issue else where? Like the
    7.8mhz shifting being detected at this stage. Do you think I would be
    able to detect this as actual audible noise with this type of config?

    I have not yet sub the signal with the use of the HP gen, that's next.

    As far as I know, the 485 should show a stable view of that at a .5us
    sweep. It has never failed me before.

    I do have a other tools to work with here if needed. It's beginning to
    look like a modification to the circuit may be needed.

  2. Joerg

    Joerg Guest

    Two things come to mind:

    Could it be that you are feeding too much amplitude in? What does
    "inject here" mean at the emitter? Current feed? How much? 2.1Vpp
    (assuming into 50ohms) sound like a ton of bricks to me.

    Listen to the carrier with something else. Do you have a communications
    receiver available at the place of vacation? Maybe somewhere behind the
    shelf with the tequila and the margarita glasses :)

    Don't spend too much time with electronics or ham radio during the
    vacation or you wife might not be happy :)

  3. Joerg

    Joerg Guest

    Ah, a vacation in Front-porchia :)

    Those can be the best, finally you get around to doing all those things
    you've always wanted to tackle.
  4. Jamie

    Jamie Guest

    My wife has been reporting to work every day this week so that's a
    bust.. My father did come down from the intelligent state (Maine) to
    visit ;)

  5. Tauno Voipio

    Tauno Voipio Guest

    A Gilbert cell is a four-quadrant analog multiplier. It makes a good
    doubly-balanced modulator or demodulator.

    For quadrature demodulation, you have to split the IF into two branches,
    and feed the BFO to the branches in 90 degree phase difference. The
    resulting audio channels (usually called I and Q) have to be combined
    with a wideband 90 degree phase difference network (Hilbert
  6. Tim Williams

    Tim Williams Guest

    What? Those aren't sidebands. Those are harmonics. You can't even get
    basic terminology right!

    Noise sidebands could be represented as a variation in amplitude and phase
    or frequency over time (i.e., the sidebands need not be symmetrical, but
    they will be distributed normally for small increments of any of the three
    quantities). This could be written as fb + epsilon, for an epsilon which
    is a normally distributed free variable with FWHA of sigma. Then, sigma
    is the bandwidth of your BFO.

    Simply, if you reduce the mean frequency of the BFO by mixing it with a
    nearby frequency, and selecting the lowest product, sigma remains the same
    magnitude, but it's a larger percentage of the total. Mixed down to 0Hz
    (baseband), the carrier disappears completely and the noise bandwidth is
    simply band-limited white noise. Typically, the distribution is peaked on
    the carrier, so the baseband version is peaked at 0Hz, i.e., it's at least
    pink noise, but more likely brown noise -- which has a "rushing" sort of
    False, only one of those frequencies are present after selecting a
    particular sum or difference with a suitable filter. Typically in SSB, fb
    = fx, so the difference frequency lands in the base band, and a lowpass
    over the audio range excludes other products.

    If you had written, fb +/- delta fn, then it would be obvious that the
    (fb +/- delta fn) - fx
    when fb ~= fx is
    +/- delta fn
    in other words, a variation in carrier frequency produces a variable tone,
    rather than a zero beat. If that variable tone follows some distribution,
    like a normal distribution, then the noise will have a random sound, and a
    shape corresponding to some color of noise, like brown noise.

  7. Jamie

    Jamie Guest

    -. ,-
    output xform )|(
    + )|(
    +-----------------------+-' '-
    7.8Mhz Filter + |
    | |
    .--------. IF amp + +
    | | || |/ \| TX switch
    | |-||+--------+-| |+------+------+-----+
    | | || | |> <| | + +
    '--------' .-. | + | --- .-.
    | | + | + --- | |
    | | | | .-. | | |
    '-' | | | | === '-'
    | +---------+ | | GND |
    +-----------------+ | | '-' ===
    --- .-. | GND
    0.039uf --- | |
    + | | 330 xmitt signal level negator
    | '-'

    The 0.039uf film poly ws the problem. Using the S/A I found what
    looked like noise at the emitters. If I biased the TX switch a few uAs,
    I could get it to stop however, the design is not intended to work that
    way. So, did a in circuit test on the cap and didn't get anything form
    it. It's hard to get at so I broke it off from the board and then solder
    a new one on the bottom side.

    The noise has diminish to a sencible level now and it has also
    increased the gain in the IF. So I guess we're golden now :)

  8. Tim Williams

    Tim Williams Guest

    Syntax warning: low base impedance.

    Caps on bases make good oscillators. Suggest a few ohms (22-100ish) in
    series with TX switch base.

  9. Tim Williams

    Tim Williams Guest

    OP didn't mention distortion, neither did you. You mentioned sidebands,
    which you still got wrong.

    Like a previous argument: sidebands are like how a waveform's envelope and
    phase change over time, harmonics define how "lumpy" the waveform looks
    within a cycle. Harmonics are farther spaced in frequency, which means
    they affect things at small time scales, and sidebands vice-versa.
    Fourier knows this. And please learn to spell...

Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day