Connect with us

active filter question

Discussion in 'Electronic Design' started by Walter Harley, Oct 13, 2003.

Scroll to continue with content
  1. Living near a couple of big TV towers, I get 15.8kHz (or so) horizontal
    retrace coming through on everything I measure. This often interferes with
    distortion measurements in audio circuitry. So, I'd like to build a little
    filter to notch out that specific frequency. Conveniently, my distortion
    analyzer supports plug-in filter cards, with the signal levels all nicely
    normalized and with power supply provided, so all I need is the filter
    circuit itself, not any additional support circuitry.

    My first thought was to use a twin-T filter. That gives me a nice deep
    notch, but at the expense of a very gradual approach. I'd like to not mess
    up any more of my "legit" harmonics than I have to.

    Next thought was to use a state-variable active filter, ideally one nicely
    built into a module for me, like the Burr-Brown (/TI) UA42. It looks like I
    can get a higher Q with this, although I'm not sure how high. Disadvantage
    is probably higher noise, but since I'm dealing with relatively strong
    signal and low source impedance, that shouldn't be a bad problem.

    I haven't any useful experience in this regard. Can anyone lend a bit of
    guidance as to which avenue to pursue? If I go the active filter route, is
    there a module that is preferred over the UA42?

    (Real L's are probably out, as the space available is fairly small.)

    Thanks,
    -walter
     
  2. I mean, of course, UAF42. Sorry.
    -w
     
  3. legg

    legg Guest

    Your measuring equipment is unlikely to be small. I suggest you look
    at simple precautions in emi pick-up via mains, test and supply leads
    in your test set-up.

    15.8KHz isn't the frequency infecting your measurement set-up. More
    likely it will be RF with this as a modulating component. RF (at the
    TVIfrequency) is easy to lose with simple absorbers and beads.

    If you were doing a wideband sweep with a spectrum analyser, you could
    spot the carriers in your baseband, and simply ignore them.

    RL
     
  4. Ban

    Ban Guest

    Walter, I would still recommend a passive filter. this one is easy to build
    and tunable to your exact frequency. The attenuation is around 20dB.
    The coil is a CT181LY-104 by Central Technologies.
    ___
    o-|___|--+-----+----o
    1k00 | |
    | |
    --- --/
    1n --- -/- 50p
    | /|
    | |
    +--+--+
    |
    C|
    C|100mH
    C|
    |
    |
    ===
    GND

    ciao Ban
     
  5. Ian Buckner

    Ian Buckner Guest

    There was a thread here titled "Active Filters" in July last year,
    with an
    exchange between Jim Thompson and Win Hill which covered
    topology and gotchas for a nice gyrator. You could use that to
    replace the inductor in the above scheme.

    Regards
    Ian
     
  6. Ban

    Ban Guest

    I doubt with the active gyrator a resistance of 82 ohms will be possible,
    but probably a factor ten higher.
    The advantage with this approach is: no power supply needed, less expensive,
    less noise, less board space required. The inductor is shielded with 5% tol.
    and should not make any particular trouble.
    We find similar circuits in fm-receivers to suppress the 19kHz pilot tone.
    I have never seen this done with an active circuit.
    The gyrator approach is well suited for lower frequencies, where inductors
    are becoming unhandy. Also with the 3 or 4 components you do not need a PCB,
    just solder everything to the connector and you're done.
    ciao Ban
     
  7. I meant, the space inside the equipment where the filter plug-in sits is
    small. (This is an Audio Precision "Portable One Plus" audio analyzer,
    FWIW).
    It might be, if all the equipment in question were under my control. But if
    I'm fixing someone's hi-fi, and it's picking up RFI through its power cord
    or speaker wires and rectifying it, there may not be a lot I can do about
    that.

    Meaning no disrespect, I have to admit to some frustration with comments
    like "RF... is easy to lose." Living in the delightfully hilly town of
    Seattle as I do, one has essentially two choices: live in a valley, and
    avoid RF; live on a hill, and share it with radio towers. I like hills...
    I am within three blocks of two different towers, with a third and fourth
    another few blocks farther. RF, in my environment, is NOT easy to lose -
    please believe me, I've tried :)
     

  8. Thanks, Ban. That circuit has the obvious advantages of simplicity and low
    noise. However, I calculate a Q around 8.5? I was thinking that I should
    be able to get a skinnier notch with an active filter. Do you think I'm
    wrong?

    Amusingly, I can't figure out how to buy just one of these inductors! I
    called Central Technologies, and said "hey, I need to buy an inductor, but I
    only need one or two, and I'm not ever going to place a big order for 'em."
    They said "request a sample from the web site." I said "really? I mean,
    I'm not a big manufacturer..." They said "yep. click." I guess it's
    cheaper to give away free parts than it is to deal with handling the money.
    Fine by me, as long as I don't have to lie to get 'em.

    -walter
     
  9. Jim Thompson

    Jim Thompson Guest

    See "GyratorFilterNotch.pdf" on the S.E.D/Schematics page of my
    website.

    You can get about any shape you want. Use high GBW OpAmps for good
    notches.

    ...Jim Thompson
     
  10. R.Legg

    R.Legg Guest

    If the interference always appears to be there, it would most likely
    be traced to equipment that is always used (ie the analyser), rather
    than the changing components being tested - as these others will at
    least vary in their susceptibility.

    I guess that if you are satisfied that the interference can be dealt
    with in no other way; that your customers don't mind your 'sort of
    modified' piece of equipment that boasted thd/noise/IMD to 4
    signifigant figures on the nice side of the decimal point and flatness
    to fractions of a dB; then adding active filters, where they work, is
    the answer.

    The question would still be - where do the filters have to be,
    particularly if the interference is being picked up by the analyser
    itself. Have you established that it's coupled into the measurement
    through input or output connections? Do filters in either location,
    that work only above 120KHz, have any effect?

    Have you contacted the manufacturer? Perhaps TVI is a known and
    possibly licked problem with the model.

    RL
     
  11. I'm sorry if I haven't been clear. Here's the situation:

    I design, build, and repair a variety of audio and musical gear. Often in
    the course of taking measurements, I find a 15.8kHz wave riding on top of
    whatever I'm trying to measure. It shows up on various instruments - for
    instance, if I dangle a lead from my multimeter and set it to count
    frequency; or if I see some fuzz around a wave on the 'scope and fool around
    with the sync in order to sync to the fuzz rather than the wave; or, when
    I'm looking at distortion residuals from the distortion analyzer. I believe
    that the 15.8kHz wave is a demodulated horizontal sync signal from the TV
    towers. (I also get a very strong 75MHz signal - e.g., if I simply connect
    the ground lead from the scope probe to the probe tip, it makes a nice loop
    antenna that picks up around a volt of signal IIRC.)

    I'm pretty sure that most of my problem is demodulation in the DUT, rather
    than in the instrumentation - the fact I can see it with the scope supports
    that, I think. I am careful with shielding and use of proper connectors,
    and the distortion analyzer is intended to resist RFI, although it does use
    XLR connectors rather than BNC (since it is dealing with balanced signals).
    I'm sure some 75MHz is leaking into the analyzer itself, but I think that's
    not where the problem is coming from.

    Most of the time this is not a big problem; I can just ignore it. For
    taking distortion measurements, though, it is a problem, because it often is
    much bigger than the rest of the residual. Fortunately, my distortion
    analyzer supports filter plugins. SO: what I am planning to do is build a
    filter plugin for the distortion analyzer, that notches out 15.8kHz (and if
    that helps, I might also try notching out some harmonics of that frequency).

    Thanks,
    -walter
     
  12. JeffM

    JeffM Guest

    I also get a very strong 75MHz signal
    The video carrier for channel 5 is 77.25MHz. That's KING5, right?
    The 15.75kHz modulates that (AM) with video between the sync pulses.

    As has been suggested,
    unless you have land-based submarines using LF transmitters nearby,
    the 15kc is riding on a carrier and being demodulated.
    You have to deal with the RF (ferrites, faraday cage, etc.).
     
  13. R.Legg

    R.Legg Guest

    I'm looking at distortion residuals from the distortion analyzer. I believe
    That is indeed one fine strong TV signal, although I should have
    thought the scope probe reception would reduce considerably in
    circuitry at the 600 ohm impedance levels of test gear. I bow to your
    experience on-site.
    Good luck.

    There are a couple of things I try to keep in mind when setting up a
    quiet lab;

    A personal computer shouldn't be galvanically connected to the test.

    Non-flourescent lighting, by preference, or just switch them off.

    A basement location is better - even on a hill. {Hard to do in an
    apartment though).

    Choice of metal(steel) or wooden bench tops, depending on what you're
    trying to do. The metal bench usually offers a lower noise test
    environment. Steel shelving can also be beneficial, if the back and
    walls are closed in, but getting rare these days. Wood or non-metalic
    benches are generally safer for prototyping, though heat-resistance
    can be a problem. Both types can be tailored for local static control.

    Keep power, grounding and isolation simple, visible and
    reconfigurable.
    .............

    Of course if you're making inadvertent noise, which is a more common
    problem around here nowadays, some of these features may also serve to
    contain it. I sometimes have to faraday-shield metering, just to get
    repeatable indications, when circuits are only just beginning to shape
    up on the safe bench.

    When I was 'lad, we used to 'ave t'faraday shield 'head, just t'stop
    teeth from catchin' on fire. Tell it to the lads nowadays, 'n they
    wo'n't b'lieve ye.

    RL
     
  14. I bow to it too, the better to present the most-affected part of my anatomy
    ;-)
    Hmm, interesting. I'm using wood right now; been thinking of running some
    copper sheet on the bottom side of it, and then maybe continuing the sheet
    up the sides or back of the shelves behind it, to give me a low-L ground
    plane to plug my gear into. Maybe I'll step that up on the priority list.
    I hate when that happens!

    -w
     
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

-