Loading ...
Sorry, an error occurred while loading the content.

1568Re: Timing Question

Expand Messages
  • vk4iu
    Nov 16, 2009
    • 0 Attachment
      Hi Dave,

      I am not sure I understand what you mean by "mult-freq unit", but I think I have solved my problems with time and Faros.

      I'm at a rural QTH with a fast Internet connection. But the connection is not that suitable for the needs of Faros.

      I have been working with Faros for several weeks, addressing the problem of "time precision". On the Internet, knowing the time is a complex thing. It works around a few servers knowing the precise time, and most others working on an "average time", usually obtained from some other server somewhere on the Internet. One Internet server, no matter how fast, does not necessarily know the correct time.

      My solution uses nothing but Faros and the Internet and has proved "good enough" for reliable beacon observations - depsite the unsuitable nature of the Internet connection.

      First, I implemented my own time servers. Bad move - it made things worse, not better. So I began work to understand the problem better. At one stage, I had five copies of Faros running with four local NTP time servers all talking to ten to fifteen Internet time servers. I even attached my own GPS to one of my own time server! All to little result. I clearly did not understand the problem well enough.

      Clearly, Faros needs to know the time, to a reasonable precision, so it can be ready to listen when the beacons transmit. But, much more importantly for us amateur radio enthusiasts, it needs a "consistent" measure of time. If the time is "consistent", then we can use the "delay adjustment" on the Details panel, to make up for all the time lost elsewhere, and Faros will not only manage to listen at the right time, but also work out whether the signal came short path, or long path. Without consistent time, Faros will miss many worthwhile observations, and be unable to tell short path from long path.

      Faros starts work before the "beacon time". It asks the sound card for 10 seconds of sound, and then does an analysis of the sound stream, using "Faros time" to find the beacon signal in the sound stream. It then calculates the "delay" for the signal.

      You need more than one Internet time server - just ONE will not do. Faros will work, but for best results, more than one Internet server of which to ask the time is needed to calculate reliable "Faros time". Faros makes no use of the PC clock - it is not consistent enough, for this work.

      My Internet time server delays range from 90ms to 500ms. There is no need to change the INI file - Faros will do that. My suggestions to help with your problem are:

      Faros already has a list of time servers, and they, most probably, will do. Use View, UTC Clock info, Test Servers, Click on the Test Servers button. But, you must do that repeatedly, at least 10 times (its important) - click the button again immediately after Faros gets to the end of the the list (that's also important). While its all happening, keep on eye on the server delays, and note those that have delays mostly below 128ms (the faster the better) - test the servers until you are satisfied - given your values at 25ms you will have no problems doing that. The server delay times may vary a lot. Note the servers that are consistently below 128ms - and tick them - and "apply the selection". You are now ready for Faros to start work.

      But, do that all again in a few hours, tomorrow, next week. Note the servers that are consistent, day after day, week after week. The immediate goal is to find seven to ten servers from the list already in Faros, that result in "Faros time" flat lining. If you have less than 10, that's OK. Start with as many as you can, you need at least 2, the more the better. But, find a list of servers and stick with it for a week before using a different set of servers.

      Now you need to do two things - just wait for time to pass, and most importantly, leave the PC running continuously with your Internet connection active. If there is a storm, disconnect the antenna, but try to leave Faros, the PC, and the Internet connection running. Clearly, that may not always be possible. You need days and hours, not hours and minutes, to enjoy all the features of Faros.

      What happens now depends greatly on the packet delay quality of your Internet connection, as compared to its speed, and how consistent the chosen servers are in providing time. Its all related.

      My experience is:

      With a good fast, broadband Internet connection, Faros will have consistent time within 15 to 30 minutes. But ... I have a fast broadband Internet connection, permanently ON, and it often takes THREE HOURS, for Faros time to "flat line" and every few hours it "moves a bit". That's because my broadband connection is via "broadband 3G wireless" and is not consistent in terms of "packet delay" when Faros asks the servers for time. Each and every Internet connection will be different. When my daughter arrives for a weekend visit, Faros finds my connection extremely slow for that weekend!

      After 24 hours or so of continuous operation of Faros, late in the UTC day, look at View, UTC Clock Info, Clock Accuracy. Move the slider on the right until the 150ms bar is near the top on the right. You may need to wait 10 to 15 minutes until the dialogue box is full of data. The red line of "Faros time" should be running through the middle of the screen, with lots of dark blue dots straddling the red line, and lots of light blue dots scattered around - most will have a light blue vertical line running through the dot. If the red line is not "flat", we need "plan B" - for now that has to wait for later.

      If you have a flat red line for "Faros tme", close the dialogue box and go to the Details screen. In the set of beacons to the top left, click on a beacon you know is strong at your location and has a high probability of being short path - any band will do. Where is the row of "dots"? The row should be approximately straight and flat - but may be anywhere on the screen. If the row of "dots" is not on the SP line, adjust the "delay correction" at the bottom of the screen until the row is ON the SP line. For a fast broadband connection the "delay correction" will be of the order of 20ms, more or less. My correction is 60ms.

      Now wait another day or two and recheck. The beacons should still be on the SP line, and Faros should be making good observations of SP/LP. If not, we need "plan B".

      If Internet time is a problem you will see ...

      On the Details panel - the dots wander up and down the screen in large amounts. Normal variation is plus or minus 10ms or so on the scale on the right - centered on the SO line.

      On the UTC Clock Accuracy dialogue - there are no, or few, dark blue dots anywhere, and the red line wanders up and down. The dark blue dots are servers with time that varyies little, less than 60ms or so. The light blue dots are servers are with large variations in reported time and are given much less weight by Faros.

      If you want, sit and watch Faros at work. Its very informative. Display the Monitor panel, and then click on View, UTC Clock Info, Clock Accuracy. You can leave the dialogue open for hours on end. It does not seem to bother Faros. Watch the beacon measurements and note the "delay" - if you have everything set correctly it should be around 0 plus or minus 10ms. It will vary with the propagation and "Faros time", which of course is related to the Internet time. Look in View, Delay Statistics. The graph should narrow and sharp - the sharper the better your "system" for Internet time is.

      Peter VK4IU


      --- In dxatlas_group@yahoogroups.com, "davemynatt" <dave@...> wrote:
      >
      > Hi group- Wondering what problems folks have with timing inside Faros and how they resolved the problem. Is an external timing app best, or, in the case of COMCAST, what NTP server are you using? Did you adjust the timing INI file and select different servers? What is a typical timing delay; I see anywhere from 25 to 100ms delay. Of course I chose the 25ms server, but I can't seem to use a multi-freq unit and instead settle upon a single freq and get good results. Any suggestions appreciated.
      >
      > Dave
      > KA0SWT
      >
    • Show all 29 messages in this topic