ad: GNOHF-1

Time sync where there's no/no reliable internet.

Discussion in 'Ham Radio Discussions' started by ON6KE, Sep 11, 2019.

ad: L-HROutlet
ad: l-rl
ad: L-MFJ
ad: Subscribe
ad: Left-2
ad: MessiPaoloni-1
ad: K5AB-Elect-1
  1. NE1U

    NE1U Ham Member QRZ Page

    I use a USB glonass/gps rcvr like this. With any of the syncing software makes it transparent ... as in no touchy ever.

    And I am so sorry to say that a once per 24hr update keeps time synced to less than a few milliseconds. I know, that is excessively unnecessarily precise. But, what do you expect for $14?
     
    NL7W and WA7PRC like this.
  2. KK5JY

    KK5JY Premium Subscriber QRZ Page

    I wrote a couple of scripts that watch the FT8 decodes and use that to discipline the local clock. You only have to get close enough by hand so that the decodes actually happen, and then the scripts slowly pull in the clock as the FT8 cycles proceed, and keep the clock very close to majority of stations heard.

    Right now, it's just a command-line thing, but I wouldn't mind refining it and releasing it if there is any interest.
     
    W0PV, WS4JM, K3XR and 2 others like this.
  3. KP4SX

    KP4SX XML Subscriber QRZ Page

    That sounds very interesting! I guess once you cull out a couple of the outliers you could get a very good average.
     
  4. KK5JY

    KK5JY Premium Subscriber QRZ Page

    Yes, that's essentially what happens. Outlier removal is an art, but given that FT8/4 signals should all be in fairly close agreement, a simple majority-selecting algorithm is good enough on even a moderately busy channel.
     
  5. KA9JLM

    KA9JLM Ham Member QRZ Page

    Question,

    Exactly why does the clock have to be so precise for FT8 to work ?

    In the day, Clocks synced to 60 cycle AC and worked just fine.
     
  6. K0UO

    K0UO Subscriber QRZ Page

    GPS Or WWV
     
    WA7PRC likes this.
  7. KK5JY

    KK5JY Premium Subscriber QRZ Page

    Because WSJT-X modes don't have their own internal clock recovery--that is, the demodulator can't watch the incoming audio and determine where the message or bit edges are. It uses pre-synchronized computer clocks to align the shift registers within a few bits, and then employs a fine-adjustment to slide the messages around a bit to find the most promising match for each one in the presence of noise. There is a fixed bit pattern embedded in the messages to facilitate that latter function, and also to help identify valid messages. Since the messages are fixed length, the amount of work to do to slide the incoming audio a few milliseconds each way to fine-tune the clock alignment is also relatively fixed, as long as the clock alignment is relatively good to start with.

    The guys who worked on FT4 tried to add clock recovery to the demodulator, but it is very tough to compete with the +6dB of signal enhancement one gets by pre-synchronizing the message shift registers.
     
    K6CLS and K4AGO like this.
  8. AG6QR

    AG6QR Subscriber QRZ Page

    A synchronous AC clock still needs an accurate time reference to initially set it. Once set, it will maintain the time with very little long-term drift, thanks to the people operating the power grid.

    Of course, all that precision goes out the window when the power goes out or the clock is unplugged.

    The subject of this thread was maintaining an accurate clock for portable field operations. I'd guess that means no grid power to supply a clock, and even if grid power is available, an accurate initial setting of the clock is needed.
     
  9. K4AGO

    K4AGO Ham Member QRZ Page

    West mountain radio has a GPS receiver that plugs into the USB port on your PC. It sells for less than $25.00 and it will keep the time accurate to milliseconds on your PC. It precludes having to use Dimension 4 or Meinberg when running digital modes requiring millisecond time synchronization.

    The software required to use the UPS receiver is a free download.

    My experience with West Mountain Radio is that they have the best customer service in the industry. If you have a question or a problem with their installed equipment, they resolve the problem. They will talk to you on the phone and help you out; Sort of like TenTec did before they changed hands this last time.

    These little GPS dongles work perfectly in any modern PC and they are dirt cheap. The last one I bought came from Amazon and was less that $30.00 and has worked perfectly.

    The mainstream GPS receivers like Garmin, etc. are overkill for just keeping PC time accurate. They are made for helping you not get lost in unfamiliar territory.

    John
     
    K0UO, WA7PRC and W4NNF like this.
  10. W0IS

    W0IS Ham Member QRZ Page

    Yes. In fact, my device is "Play" without any need to "Plug"!

    This thread reminded me of an experiment that I had started in June and had neglected to look at since then. I wanted to know how accurate the cheapest possible clock I could find would be. I started out by going to the Dollar Store, but they unfortunately didn't have any clocks that day. Instead, I went to Target, which is a U.S. discount store, and found this clock:

    [​IMG]

    https://www.target.com/p/9-round-wa...3/-/A-50417922?preselect=50330031#lnk=sametab

    This is a very nice wall clock with a sweep second hand. It's very large (about 20 cm in diameter) and it's nice and easy to read for those of us whose vision is slowly fading away. It cost a grand total of $3.99. It uses one AA battery, and I believe I used a carbon zinc battery.

    On June 10, I set it to WWV and placed it on the wall. For about the next two weeks, I dutifully compared its display to WWV. It was 3 days before I could detect any difference, and it was 1 second slow. I kept taking readings until June 25, at which time it was 2 seconds slow. Ideally, I should have taken a few more measurements, but it appears that it is losing time at a rate of 1 second per 6.5 days.

    It has now been 90 days. Therefore, if the rate of loss is linear, then it should have lost 90/6.5 seconds = 13.84 or approximately 14 seconds. So whatever time it displays, I should now add 14 seconds to get the true time.

    Comparing that to WWV, the calculated time is now 1 second slow. In other words, when WWV says that it's 01:00:00, my clock says 00:59:45, and I calculate the time as 00:59:59.

    So now, after 90 days, the clock is no longer accurate to within 1 second. But if you are going to be off the grid for 45 days or less, then it would still be accurate enough. In fact, for about 3 days, it would be accurate enough even without applying any correction.

    I suspect that I could have come up with a more accurate correction factor if I had taken more measurements. In fact, going forward, I will use 15 seconds/90 days to do the calculation. Also, I made no special effort to control the temperature. But in answer to your question, as long as you'll be off the grid for less than 45 days at a time, then I would say to go buy a cheap wall clock and occasionally set your computer's clock according to that.

    Having completed this experiment, I believe I'm now eligible to collect my Longitude Reward of £20,000:
    https://en.wikipedia.org/wiki/Longitude_rewards
     

Share This Page