I Need 10MHz – how hard can it be!

It all started when I wanted to calibrate my HP 53131A universal counter, which as it turns out probably has one of the crappiest and most disappointing standard oscillators ever put into a frequency counter, HP you should bow your head in shame….oh of course I forgot, a half reasonable oscillator is an “optional extra” when you by HP/Agilent – of course it is….anyway, on with the job at hand

If you have or want to play with an FE-5680A Rubidium Frequency Standard or an OSCILLOQUARTZ OCXO 8663-XS or a HP 53131A Counter or a Racal Dana 1999 counter or similar then this video will most likely be of interest 🙂 what I am trying to get is a predicable and reliable frequency and standard for my home lab.

I guess I will let the video do the talking on this one….

Here are a whole bunch of useful links that relate to this video (there are many more too if you search around the web)

Thanks to all of the authors and content creators for the above information. Thanks for watching.

This content is published under the Attribution-Noncommercial-Share Alike 3.0 Unported license.

37 thoughts on “I Need 10MHz – how hard can it be!

  1. Hi Garry, this is really helpful!! We will recommend this blog to others! Thanks a lot! We look forward to the teardown. ——Sasha

    • Thanks for these, thats really useful. I can see that my hard power switch mod occupies the two holes that hold one end of the OCXO board! The HP OCXO is massive, so big…now I know what all that space in there is for! and I am jealous, you also have the high frequency option, that was another thing I was thinking about cloning !) With your permission I will stick these into my media library and use them in a future blog post when I do the DIY OCXO mod? The OCXO I have is a double-oven one and quite accurate as you might have seen in the video. I think it should be easy to clone this option. I am also going to make the board compatible with a lower-cost (and lower power) TCXO option. Once again, thank you for taking the time to take photos and post them, thats really ver kind of you. Gerry

    • Hi, I will of course attribute them to you, is your YouTube name also SwitchingPower? I have seen the clone high frequency option before. For EUR240 its quite expensive, I think I might have more fun making my own clone 🙂 Thanks for the pointer though…Gerry

      • Gerry – Nice video – a couple of points.
        1. In my experience ( comparing against a trimble thunderbolt ) the fe5680a is often set 0.1hz low.

        2. I have designed a pcb for the 3Ghz option , it’s on the way back from China now. Once I’ve confirmed there aren’t too many issues with my design – you are welcome to one of the pcbs.

        • Hi Andy,

          Thanks for the heads up, I will keep it in mind. My problem is that I have nothing else to reference it to, perhaps I should get me a GPS locked reference too. I would be interested in one of the 3GHz PCB’s for sure – I was thinking about designing one but if you have done it already and it works there is no point me re-inventing the wheel, thanks for the offer.


  2. Gerry,
    I followed a few of those links on the reference modules, I noted that one of them reckoned almost
    double the life expectancy if the device is kept fairly cool. Would be intresting to see if the current consumption were to go up much if the module is kept cool as of course the heating element would need
    to work a little harder to keep the internal bits at the right tempriture, same would go for the ovened
    xtal osc.

    • Span, yeah I recall reading that too. I am not sure, I think the physics package needs to run at that temperature so cooling the rest of the device would only help the electronics. In any case they seem to have packed the innards with sponge to restrict airflow, there is even a thermister soldered to a standard XTAL with current making that hot (make shift oven), I suspect you would get more life by maintaining a temperature, even if its a hot one. I can’t remember now if there are electrolytic caps in there, from memory there is not that many. I will do a teardown soon on video of both the broken ones. I was well impressed by how close the OCXO and the RFS were tracking each other though. Gerry

  3. Hi, I already made some work with Rubidium frequency standards. Here are some pictures, text unfortunately is in Lithuanian language, but pictures are self explanatory. If something will be confusing do not hesitate to ask.

    http://www.laidukas.lt/blog/?p=485 Here I investigate that 10MHz signal from internal coaxial connector has greater spurious noise then near to DB9 connector so I chose to use DB9 pins to solder my coaxial cable.

    http://www.laidukas.lt/blog/?p=521 Here I measure EMI of rubidium standard because one would often use these standards with RF gear. Noise is quite bad in UHF region. I would recommend to enclose this standard in “Faradays cage”.

    I plan to do some sheelding to standard and

  4. Hi.

    Very useful information, this answered a few questions I had about the output from the rubidium references. Thanks. 🙂

    I’m a bit surprised about the current consumption of the devices.
    Have you measured how much current the 5V is using?

  5. Hello Gerry, Interesting to watch your video – I went through an almost identical process a few months back!

    Started by a desire to double check the accuracy of my venerable HP5316B Universal Counter, with Option 001 Oven Crystal. I had tweaked and locked this to the “free” old 64uS (15.625kHz) line frequency of old TV PAL transmissions, which the BBC maintained to with half a gnat’s bollock of spot-on. Sadly, that easy reference has now gone, with the closure of analogue TV transmission!

    I bought a rhubidium reference. Yes, it calls itself FE-5680A. But I guess like Ford Mondoes, the “same” car comes in a wide range of specifications. Caveat Emptor.

    I agree that this unit runs warm, too warm to embed into most other equipment, and indeed too warm to be left on its own. It needs some sort of cooling. I also wanted to make the “bare” reference a bit more portable and presentable, so I have shoe-horned one into a nice matt-black extruded aluminium case from Hammond, which helps with cooling and looks the part. This one:


    This is a really nicely made case; very accessable and easy to work with.

    An external 15V 3A power supply feeds it via a 2.1mm power socket on the back of the case, which also serves as the mount for a 7805 regulator. The front panel supports a 1pps and 10 MHz socket, and to LEDs – 1 PPS and “locked”. A small daughter board fits inside the case to buffer the signals from the FE5680, and also to drive the LEDs. The 1PPS signal is so short that I hade to make a simple monostable pulse extender, so that you can see it on the LED…

    A nice, neat solution which now serves as my lab reference to several other bits of kit. The old HP OCXO is still within parts per hundred million of spot-on, however!

    EXACTLY HOW accurate? I suppose I have no absolute answer, but as I use it for calibrating clocks and there are ~30 million seconds a year, anything better than one part in ten milion is fine fo rmy needs. I reckon the FE5680 delivers that.

    Post Script: Maynuo eLoad and PC software now all working just fine, thanks! 🙂

    • Hi Lawrence,

      Thanks for your comments, glad your Maynuo software is working now – good to hear. I have been pointed at a really great enclosure for my home-brew frequency standard, I am going to post a video on it once I have it.


  6. Hi Gerry, I found your rubidium videos very interesting and have been experimenting with one of the 1pps re-programable ones that you bought originally. It is actually a very useful and flexible device well worth playing with. You mentioned that there is no 1pps signal coming from yours and I originall thought mine was faulty as well but it turns out that the 1pps pulse is very narrow and almost impossible to see on a scope. I tried driving a TTL counter and it worked fine. The other thing I wanted to pass on to you is about the second rubidium you recieved. You said it would not lock, I found some info on the internet about this, sorry cannot remember where right now, and apparently the PLL (FLL) may have just drifted out of range due to age etc. It is possible to adjust something inside to bring it back into locking range again. There is something about it here but its not the original info 🙁 http://www.febo.com/pipermail/time-nuts/2012-January/062780.html

    • Hi Mark,

      Without trimming they are well within 100Hz generally but they do need calibrating. The key thing is though, once calibrating they are very stable and very accurate easily down to 100th of 10 cycle.


  7. Hi Gerry,

    This is Masoud, studying mechanical engineering at UNC Charlotte. I want to buy a rubidium frequency standard from ebay. Just saw your video on Youtube, I realized that it might be better for me to buy from the same seller that you bought yours, since they have been nice and they have sent you three to make sure that it works. Would you please let me know the seller’s ebay ID?


Leave a Reply

Your email address will not be published. Required fields are marked *