Single frequency oscillator,vhf range, 0.1hz precision

Discussion in 'The Projects Forum' started by AlgoryThm, May 18, 2013.

  1. AlgoryThm

    Thread Starter Member

    Apr 23, 2008
    22
    0
    Hello! Well I need something very specific and although I did some searching on the forum and I found some similar topics,generally speaking, I didn't find anything close enough to what I need, so I'm posting my question here.

    I need to make an oscillator that generates a very specific frequency, in the vhf range (let's say for example 192,353,452.3hz), accurate -and stable- down to the last digit. That is with a resolution of 0.1hz. It will be used in a lab for nuclear magnetic resonance study, so you realize it has to be extremely precise with the less possible instabilities I could get. Because I need a single very specific frequency only, I thought it would be pointless to spend a couple of thousands of $ on an expensive rf signal generator and that it would be better if I made one myself. The problem is I'm not sure about the path I should follow.

    I don't have the necessary skills to write the code for a PIC DDS (yet, I'm learning fast though) and I'm not sure if there is anything like AD9851 that goes up to 200 MHZ anyway. If you are aware of any project though, with the code already written, that I could try and replicate, please inform me about it.

    Solution no2 is a simple crystal oscillator. But I rejected that idea too because of the kind of precision I need. I don't think this kind of approach would give me a 0.1hz resolution at almost 200Mhz (or maybe not? what do you think?).

    So is there any solution other than ready made expensive rf signal generators that I could use? If the oscillator is absolutely stable, I could settle for one with 1hz resolution.

    Thank you in advance for any help given!
     
  2. Shagas

    Active Member

    May 13, 2013
    802
    74
    0.1 hz resolution at Mhz Range , is that even possible?
    I mean even with fractions of a degree temperature change even the most stable oscillator would drift by more than 1hz no?
     
  3. GopherT

    AAC Fanatic!

    Nov 23, 2012
    6,023
    3,789
    To get this resolution, you need a heated (temperature controlled) oscillator. And yes, that heater must be controlled within a refraction of a degree. Signal generators with heated oscillators with this precision (0.1 hz) are very specialized pieces of laboratory grade equipment and will cost you in excess of $2000 used (e.g. Agilent N9310a).
     
    Last edited: May 18, 2013
  4. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    Iam no RF guru, buf from watching many teardowns and reviews of lab gear, everything starts with a high-precision tempeature compensated or oven controlled 10Mhz crystal oscillator. This is then fed into a PLL that generates the required frequency.

    But anywaz, the AD9851 goes up to 180MHz and you theoretically could multiply the output ferquency, but you still haven't said exactly what frequency range you require.

    Also look at these two
    http://www.youtube.com/watch?v=oUiEeOp8ynY
    http://www.youtube.com/watch?v=mWHiXTfsUP8
    Maybe some compannies can lend you a proper generator?
     
    absf likes this.
  5. MrChips

    Moderator

    Oct 2, 2009
    12,440
    3,360
    Let's look at the specs.

    2Hz in 200MHz is 1:10^8 or 10 ppb
    1Hz in 200MHz is 5:10^9 or 5 ppb
    0.1Hz in 200MHz is 5:10^10 or 0.5 ppb

    A standard crystal oscillator will give about 5 ppm with short term stability of about 1 ppm.
    A temperature controlled crystal oscillator will give about one order of magnitude improvement, 0.1 ppm.

    A microcomputer compensated crystal oscillator pushes the stability down another order of magnitude to 1:10^8. We're still not close to your desired specs.

    Your bet is a GPS synchronized oscillator which will get you to your specs of 1:10^10.

    Either that or you're looking at an atomic clock.
     
    absf likes this.
  6. Dodgydave

    Distinguished Member

    Jun 22, 2012
    4,979
    744
  7. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
  8. AlgoryThm

    Thread Starter Member

    Apr 23, 2008
    22
    0
    Thank you all very much for your quick replies! I really appreciate it! Well, yes I already knew about temperature controlled oscillators but it's something I want to avoid, because of the cost. Kubeek, one frequency I need is around 172Mhz. The spectrum I need is 171 to 173Mhz and 172 to 173kHz, the last precise down to third digit eg. 172,151.556hz. I think the khz generator is much easier to be made, I will try a very accurate DDS I already have with a spectrum from 1hz to 1Mhz. But for the VHF region I didn't see any solution until you gave me some and thank you for that. I will check all the links and get back to you.
     
  9. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    Remember all those things are only as accurate and stable as the primary frequency source is. Also, how do you plan to verify that you are getting out what you want? Again if the meter is not very stable and accurate the measured results will be useless.
     
  10. AlgoryThm

    Thread Starter Member

    Apr 23, 2008
    22
    0
    $80 + shipping??? That's a gift!! So what might the rest of the circuit be? A regular microcontroller circuit taking advantage of the crystal's precision or perhaps something more special to make it work? Can I use it as a regular two pin crystal?
     
  11. AlgoryThm

    Thread Starter Member

    Apr 23, 2008
    22
    0
    About the measurements, I really don't need any because it will either work or not. If it doesn't, I will know it is from the oscillator and not from the testing material. I had the luck to have a friend in MIT and he sent me the exact frequencies I need, after a lot of testing and experimenting. So he did all the hard work, I just need to replicate it and confirm it works.
     
  12. GopherT

    AAC Fanatic!

    Nov 23, 2012
    6,023
    3,789
    $80 just gets you a 10.0000000000 MHz standard signal.
     
  13. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    But wiht a pll that multiplies this clock 30 times he should be able to use that ad9851 chip to generate whatever frequency he wants. But the jittter in the pll could be what breaks the precision of the whole setup i guess.
     
    AlgoryThm likes this.
  14. AlgoryThm

    Thread Starter Member

    Apr 23, 2008
    22
    0
    Ok I get it now...I will try it with a pll circuit and I'll let you know how it went. Thank you both and all the previous guys for this help!
     
Loading...