Forum Rule: Always post complete source code & details to reproduce any issue!
Results 1 to 4 of 4

Thread: GPS Frequency Output surprise

  1. #1
    Senior Member
    Join Date
    Mar 2015
    Location
    UK
    Posts
    260

    GPS Frequency Output surprise

    The ability to lock a frequency signal to accurate GPS timing is very useful for a number of Teensy projects. It may come as a bit of a surprise therefore to learn that the frequency signal from a cheap GPS module such as a NEO6 or NEO7 board is not quite as dependable as you may think.

    The TIMING accuracy of such a GPS module is excellent, especially for the small price. But FREQUENCY output from these boards is not exactly the same thing as TIMING - as the following example demonstrates.

    A NEO7 module can be configured to give a square wave frequency integer output anywhere between 1 Hz and 10 MHz. You can use free U-center software to do this easily on a host PC. Now Fourier Analysis tells us that harmonics from the fundamental frequency should be available, and should be strictly related to that fundamental as multiples of that frequency all the way up the radio spectrum to 1 GHz and beyond. This relationship is extremely useful for activity such as filter design, or calibration of standard test gear, where you are interested in frequencies much higher than 10 MHz.

    But have a look at this next picture. It came as a big surprise to me when I first saw it.

    Click image for larger version. 

Name:	GPS 8MHz 28th Harmonic 02.jpg 
Views:	38 
Size:	63.3 KB 
ID:	16388

    The three spectrum displays shown were recorded using SDR software. I initially calibrated the software using an independant technique of comparison against the carrier of a local digital DAB radio station. Such stations have their frequency outputs also locked to GPS, so I knew the calibration was good around 222 MHz.

    In the top display, you see the harmonics from a GPS module set to 8 MHz under satellite "locked" conditions. The 28th harmonic of this signal ought to have been a clear and distinct peak set exactly at 224.000 MHz. There is such a peak in the display but it is far from being clear and distinct. Indeed, of the many peaks visible, it is not even the peak with the biggest amplitude. Nor is the display symmetrical about the 28th harmonic frequency. (8 MHz was chosen for the GPS output because this is said to have the lowest jitter).

    The middle spectrum shown is the result of disconnecting the GPS board completely from the PC. I wanted to be sure that the these previous many peaks were not caused by some other source.

    The bottom display shows the output from an 10 MHz OCXO which is connected to a 74HCT390 divide by 10 chip to give a 1 MHz fundamental signal. This display is now showing the 224th harmonic, which is precisely located on 224.000 MHz.

    I had expected to see lots of "mush" in the spectrum of the GPS board, but I thought it intuitive that the major harmonic multiples ought to have stood out clearly in the spectrum of a frequency source which is "locked" into the global navigation system. The picture shows my intuition to be wrong.

    I not fully sure of what is going on here. I wondered if others might have an expanded explanation.

    The TIMING of each 8 MHz square wave cycle is not inaccurate at all. Indeed, when I show the OCXO signal and the GPS board output both together on a scope, they appear rock steady over a long period.

  2. #2
    Senior Member
    Join Date
    Jul 2014
    Posts
    2,221
    Maybe you should try an odd multiple of the fundamental frequency (Fourier series of square wave)

  3. #3
    Senior Member
    Join Date
    Mar 2015
    Location
    UK
    Posts
    260
    Excellent idea... here is the 27th harmonic (one down).

    Click image for larger version. 

Name:	GPS 8MHz 27th Harmonic 01.jpg 
Views:	23 
Size:	25.8 KB 
ID:	16389

    A little more informative. The symmetry is still not centred on the 27th exact frequency. And the display is not static, the base noise level is bouncing around but that depends to an extent on the SDR behaviour in creating the display.

    Curious that there are 26 peaks in a 1 KHz spectral distance.

    On the scope, there is 20nS approx of jitter on the GPS 8 MHz (OCXO 1MHz used as trigger). Is this some form of PSK that is happening? But odd that it is not centred on the expected integer harmonic.

    Just out of curiosity, here is the 29th harmonic...

    Click image for larger version. 

Name:	GPS 8MHz 29th Harmonic 01.jpg 
Views:	24 
Size:	25.0 KB 
ID:	16390

    Looks like the symmetry has shifted ???
    Last edited by TelephoneBill; 04-10-2019 at 06:15 PM. Reason: its 1 KHz distance - not 100 KHz

  4. #4
    Senior Member
    Join Date
    Jul 2014
    Posts
    2,221
    what happens if you put a week LP filter in the line (to get rid of possible overshoots)?

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •