Shop OBEX P1 Docs P2 Docs Learn Events
28506-VPN1513GPS Timemark (1pps) precision ? — Parallax Forums

28506-VPN1513GPS Timemark (1pps) precision ?

jmgjmg Posts: 15,155
edited 2012-02-08 11:54 in Accessories
I have found an older thread
http://forums.parallax.com/showthread.php?131916-Coming-Soon-A-New-GPS-Receiver-Module

that mentions a Timemark pin (that Parallax forgot to connect?!), which has the 1pps, but you can run a worm to this.

"The pin you want is "Timemark" which is pin 20 on the GPS module."

The specs give the somewhat vague 1us accuracy on this, but I was wondering if anyone has actually tested this, on a real system ?.

Other GPS chips can deliver 50ns, 25ns or similar, and given GPS systems are rather generic, I was wondering how this compares... ?

As an example, this is a reference example plot, from a Trimble Resolution T GPS

http://www.leapsecond.com/pages/res-t/1sigma1.gif

Comments

  • David BDavid B Posts: 591
    edited 2012-02-06 13:18
    One question you could ask is how closely does the 1PPS transition agree with the "real" one second change, in other words, what is the error in the phase of the oscillator; that sounds like what your discussion refers to.

    But a different question would be to ask how closely does each one second period of the oscillator match the "real" time interval of one second; in other words, what is the noise in the frequency of the oscillator, not quite the same thing as the first, and that, I think, is what the attached chart is measuring.

    It sounds like you're more interested in the phase, is that right?
  • jmgjmg Posts: 15,155
    edited 2012-02-06 21:47
    Rephrasing, Consistency between pulses, matters more than the difference between (say) an edge on my bench, and an edge on yours.
    The objective is not to measure neutrino-speed, ;) , but to calibrate time-base measurements.

    ie Has anyone hooked up a decent counter onto one of these, and checked the LSB's ?
  • Tracy AllenTracy Allen Posts: 6,658
    edited 2012-02-07 08:28
    I'm curious about that too. Search for "GPS disciplined clock" to find good data sheets on the state of the art. For example, here from Trimble, Mini-T GPS Disciplined Clock Board. 15ns accuracy the 1pps signal, and phase noise on the 10MHz output less than -120dBc. (I'm not sure what the "c" stands for. Also, the following plot from the data sheet, I don't know how to interpret it.)

    Allan plot_GPS.png
    Allan plot_GPS.png
    573 x 472 - 153K
  • max72max72 Posts: 1,155
    edited 2012-02-07 08:57
    there is a nice description of the 1pps pulse on the A1035H datasheet.
    google A1035H vincotech datasheet
    page 30 concisely describes the nature of the 1pps. The delay is dependent on the GPS module, but it is the clearer description I found so far.
    Massimo
  • jmgjmg Posts: 15,155
    edited 2012-02-07 12:25
    Tracy Allen wrote: »
    ... 15ns accuracy the 1pps signal, and phase noise on the 10MHz output less than -120dBc. (I'm not sure what the "c" stands for. Also, the following plot from the data sheet, I don't know how to interpret it.)

    dBc is dB relative to carrier.
    The plot is similar to the link I gave in #1, but I think #1 is someone measuring, and #4 is Trimble measuring.
    #4 is better, and suggests #1 is the sum of measuring system, and clock.

    Given the claimed sigma of under 10^-10 @ 1 second, that is ~80ps of edge sigma precision, and very few 'average' test systems can deliver 80ps in 1 second!.

    The curve also suggests simple edge noise dominates up to nearly 200 seconds, and then time-averages start to gain.

    Contrast that with #1, where that direct curve maps almost exactly to a ~20ns edge sigma precision.
    ie close to what a Generic Prop-Pin would measure #4 as.
  • jmgjmg Posts: 15,155
    edited 2012-02-07 12:55
    max72 wrote: »
    there is a nice description of the 1pps pulse on the A1035H datasheet.
    google A1035H vincotech datasheet
    page 30 concisely describes the nature of the 1pps. The delay is dependent on the GPS module, but it is the clearer description I found so far.
    Massimo

    I see the A1035(0SiRFstar III chipset) is being phased out, and the A2035 (SiRFstar IV chipset) is cheaper, with under $20 systems for Module + Antenna now available.

    http://www.mouser.com/Search/Refine.aspx?N=10789085&Ns=Pricing|0

    This http://www.rell.com/resources/RellDocuments/SYS_27/Maestro%20gps%20Focus%20Products%2006232011.pdf

    claims both have 1pps &- the 1035 data says ~ +- 45ns but I could not find similar data for the new 2035.

    You never know when a drop in price, also delivers a drop in spec...
  • max72max72 Posts: 1,155
    edited 2012-02-08 05:57
    the A1035H simply changed manufacturer name.
    http://it.mouser.com/ProductDetail/Maestro-Wireless-Solutions/A1035-H/?qs=sGAEpiMZZMsa9w4dmI1JYEv5hDt7Ew5Y
    Anyway the Vincotech datasheet is much more complete, so I use it as a reference.
    Massimo
  • Tracy AllenTracy Allen Posts: 6,658
    edited 2012-02-08 11:54
    Given that there are phase variations in the GPS signal due to propagation effects, there has to to be a good local oscillator locked to the GPS. So, for disciplined clocks, as opposed to generic GPS, is it that they have a super-stable local oscillator and extraordinary algorithms for reducing the noise? I took a look at http://www.allanstime.com/AllanVariance/ and also at the interesting App note HP application note 1289: "The Science of Timekeeping he linked to there. Do you know other good tutorials on this stuff?
Sign In or Register to comment.