Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Basic crystal oscillator specifications question

  1. Jul 21, 2012 #1
    I need to create a 10 MHz square wave clock signal which gets terminated into a 50 ohm resistor for use by a waveform generator. The specifications for the waveform generator suggest a 12 to 14 dBm power level for the input clock. ie. a power level of ~16-25mW.

    So I've been looking to buy an oscillator circuit component online of which there are plenty but I can't find anything at all which meets those power specifications. The way I'm calculating the power is by:

    [tex] p_{rms} = \dfrac{v_{rms}^2}{R} [/tex]
    [tex] v_{rms} = V_H \sqrt{D}[/tex]

    Which means i need a high level of [tex] V_H = 1.12V [/tex]

    Is the power output calculated in this way for RF specifications of this sort or am I doing something wrong? Or are Oscillator ICs just generally not available with these power levels? If so what are my options? I'm a physics student currently doing a bunch of electronics stuff so I'm still not incredibly familiar with a lot of specifications given on data sheets but I'm definitely willing to do some reading.
  2. jcsd
  3. Jul 21, 2012 #2

    jim hardy

    User Avatar
    Science Advisor
    Gold Member

    theer are 50 ohm line drivers like SN74138

    might be overkill but if you want something to experiment with,,,,,
  4. Jul 22, 2012 #3
    Ahh yes that looks like it could be useful. Since I don't really have many requirements other than generating a clean clock signal overkill is probably fine. I guess will probably just buy a few components and try to see what works best.
  5. Jul 22, 2012 #4


    User Avatar
    Science Advisor

    Getting a square wave at 10 MHz is more than power levels. It needs a bandwidth close to 100 MHz to achieve this.

    Even if you got it, you would not be able to see it as a square wave on an oscilloscope unless it had a bandwidth like this.

    So. I would be more worried about bandwidth than exact power levels.

    One way of doing this is to use very fast frequency dividers (like the 74 F family) and divide from 100 MHz to 10 MHz or even from 20 MHz to 10 MHz.

    Crystals you see on Ebay vary a lot in accuracy and stability. If you need an accurate reference frequency, be prepared to pay $20 or more for a suitable oscillator. The "Bliley" brand is pretty good and they make oven controlled oscillators which give a crystal a stable temperature. Of course the oven uses some power (about 500 mA at 5 volts).
    Computer grade oscillators usually have very poor accuracy, (+/- 100 Hz) but they are cheap and may be good enough for a signal generator.
    Last edited: Jul 22, 2012
  6. Jul 23, 2012 #5

    jim hardy

    User Avatar
    Science Advisor
    Gold Member

    i once made myself an oscillator for calibrating oscilloscope, used Epson SPG8640 which is a crystal oscillator and frequency divider in one IC. It lets you set the divisor so is quite versatile.

    But i couldn't locate an IC at Digikey anymore
    surely somebody still makes the little fellow.

    I took it into work and compared it against our frequency standard and it was good to five digits.

    What a tinkerer's delight.
    Maybe one of the electronics professionals here knows of something similar.

    old jim

    edit aha found the datasheet indeed it's discontinued
    http://www.epsontoyocom.co.jp/discon/epsondiscon/2000catalog/spg_8640series_e.pdf [Broken]
    but it didn't go to 10mhz.

    if they still use same oscillator in a more modern product i think you'll be surprised at its performance.

    If you need the precision of an oven look for a surplus HP 10544A oven contolled crystal oscillator Ebay lists several this morniing. But, it outputs 1volt sinewave you'll need a buffer behind it.
    Last edited by a moderator: May 6, 2017
  7. Jul 23, 2012 #6


    User Avatar
    Science Advisor

    I have one of these:


    which will give 0 to 5 volt pulses out, but probably not into 50 ohms. Ideal for driving counter chips though.

    There is a control pin which can be connected to the wiper of a 10 turn pot, with 5 volts across it, so that the frequency can be set to exactly 10 000 000 Hz. Note that is 8 figure accuracy.

    I have a Rubidium standard to calibrate against and I use the excellent PC program Spectran to do the calibration. Once warmed up it will hold calibration within 1 Hz.

    So, if this is used with a DDS signal generator, I can set a frequency to within 3 Hz or so from 0 to 30 MHz.

    It gets better if I use the Rubidium standard.

    This is equipment that would have cost thousands a few years ago but is now available to anyone.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook