Connect with us

VHF oscillator precision

Discussion in 'Electronic Design' started by Apparatus, Oct 13, 2005.

Scroll to continue with content
  1. Apparatus

    Apparatus Guest


    I am working on building a MURS band (154.600 MHz, 20 KHz) FSK
    transmitter. The issue that I'm running into is oscillator stability.

    My current idea is to use a crystal based colpitts oscillator with a
    JFET for the feedback element, adding a varactor in parallel for a
    small amount of tuning. However, crystals themselves have typical
    precision ratings of +/- 50ppm. At 150 MHz, that's +/- 7.5 kHz
    tolerance, nearly the size of my band. Then the additional issues are:
    the precision of the varactor, stray JFET capacitance, etc.

    I have thought about also using a mixer, with two oscillators as above,
    one of which is tunable via a varactor, both at around 75 MHz.
    Probobalistically, the tolerances should somewhat negate (assuming much
    higher mixer precision).

    Without a manual tuning element, how can I hit 154.600 MHz and keep the
    FSK tuning within my 20 kHz bandwidth? I feel as if I am trying to get
    a hole in one from a mile away. Any insights would be much appreciated.

  2. Temperature compensate the oscillator. And if that doesn't work,
    put the thing in a crystal oven.

  3. Joerg

    Joerg Guest

    Hello Chris,
    Michael's suggestion would work. If that costs too much you will need
    tuning. Either by hand or, if it is a mass product, by an auto-cal
    procedure at final QC where the setting is stored in flash (assuming you
    have a uC with flash on there).

    Regards, Joerg
  4. Don Bowey

    Don Bowey Guest

    First, you don't have 20 kHz.

    I don't perceive a problem. The maximum channel bandwidth is, per the FCC,
    11.25 kHz and permissible modulation deviation is +/- 2.5 kHz. If you build
    the oscillator using insight from the crystal supplier and achieve the (very
    doable) +/- 7.5 kHz stability, The signal will always be contained within
    the 11.25 kHz limit. A decent receiver will have no problem capturing the
    signal at it's frequency tolerance boundaries.


  5. I don't see a sound basis for the "probabilistic solution". Surely some
    units would be as bad as some single crystals ?

    Once your straight crystal spec is unsuitable, the next step is TXCO -
    packaged oscillator units with a varactor and thermistor setup inside to
    tune out the shift. These are stable enought for your application, however,
    note the aging spec. These do have a trim pin for fine frequ shift, though
    it is probably not for FM. If you can't FM such an oscillator, you can use
    it as a reference for a PLL and modulate the oscillator control voltage. At
    that stage you might want to save money by choosing a low frequency
    oscillator at sub 20 MHz and using the PLL to multiply it up.

    There are chips around which make all this this terribly easy. Analog
    Devices RF have some parts for FSK transmitters. Quite a few others.

    Possibly you can bypass the whole thing by buying complete modules with an
    data input pin !

    Roger Lascelles
  6. Mark

    Mark Guest

    you either have to pay for a better osciallor i.e. 5 ppm or you have to
    include a trimmer and pay for someone in the factory to tune it.

  7. budgie

    budgie Guest


    The 11.25kHz channel bandwidth has to contain 2*(highest mod frequency before
    roll-off) + 2* (Max deviation). That means audio roll-off would need to start
    at (11.25 - 5)/2 = 3.125kHz. That's assuming the carrier is on frequency. As
    soon as the carrier is off-frequency by (say) 1kHz the sum of (max dev + mod
    frequency) has to drop by that amount to remain strictly within that allowed

    Having said that, the O/P appears to have misunderstood the +/- 50ppm crystal
    tolerance. This doesn't mean it wanders around by that much in operation.
    Rather it is the tolerance band for the crystal out-of-the-box and before any
    frequency trimming circuit is applied. Depending on the resonant mode chosen, a
    simple trim-cap or coil will normally suffice. Let's face it, commercial
    two-way radios have existed in such narrowband environments for decades using -
    in many cases - no temperature compensation other than a carefully chosen
    ceramic cap with a known tempco.

    If the O/P wants to reinvent the wheel, that's fine. For my money, I'd employ
    an existing two-way off ebay and save my design time for a project where there
    isn't an easy cheap off-the-shelf solution.
  8. And for a Really Cheap(TM) crystal oven, you mount an appropriate value
    resistor (180 ohm 1 watt for a 12 volt system) next to the crystal on long
    leads so that the resistor can touch the crystal case on one side and a
    cheap (1N4148) diode on the other side of the case as a temperature sensor.
    Shrink sleeve both the resistor and diode so that they are in intimate
    contact with the crystal case. Use an op amp comparator to turn on a
    saturated driver transistor and adjust the bias on the opamp so that it
    switches off when the temperature of the diode gets to 50°C or so. Use a
    bit of hysteresis on the comparator so that it doesn't chatter during the
    switching period.

  9. doug dwyer

    doug dwyer Guest

    The biggest problem will be to pull the crystal frequency to nominal,
    20kHZ in 200MHz is 200ppm so 50ppm at roomtemp will be within this.
    Always place the varicap in series with the crystal at this frequency
    and tune out external reactance to zero by a an inductance.
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