5
$\begingroup$

a) Given that I don't have a frequency standard or other known transmitter available, what is the easiest way to calibrate the NooElec NESDR SMArt which is using a TCXO rated at 0.5 PPM.? I.e. to determine the PPM calibration value?

b) Also, is this value frequency dependent? (How so?)

c) Are there any European frequency standard transmitters I could use?

$\endgroup$
6
$\begingroup$

Given that I don't have a frequency standard or other known transmitter available, what is the easiest way to calibrate the NooElec NESDR SMArt which is using a TCXO rated at 0.5 PPM.? I.e. to determine the PPM calibration value?

Find any continuous broadcast signal that has a carrier and you know the nominal frequency of, and adjust the PPM value until its peak in the spectrum display is centered on that frequency value.

The higher the frequency of the signal you are calibrating to, the more your local oscillator's error is magnified, so the more precise this eyeball procedure can be. (If you have a large error, it may be useful to start at a low frequency before moving up, so you don't have trouble finding the expected signal in band.)

Of course, it's only as good as the transmitter, which you don't know, but you can easily get it good enough for accurate tuning (as opposed to frequency measurement).

Also, is this value frequency dependent? (How so?)

No. The physical oscillator generates a signal at a fixed frequency, and the tuned frequency is derived from that in a way which more or less involves counting — there is no systematic “analog” error in it.

$\endgroup$
  • $\begingroup$ The gps signal data includes a 0.1 PPM REFERENCE on the second interface. $\endgroup$ – Rowan Hawkins Mar 24 '17 at 18:46
  • $\begingroup$ @RowanHawkins That sounds like material for another answer if you also explain how to easily compare it the with RTL-SDR clock. But it doesn't suggest an improvement to this answer. $\endgroup$ – Kevin Reid AG6YO Mar 24 '17 at 18:53
3
$\begingroup$

I want to add some solutions to Kevins answer. Apparently I'm lucky to own the SMArt device as it is really stable with a PPM between 0.5 - 1.0 at all bands I've looked at. (Just as advertised.) So how do you calibrate your less stable RTL-SDR?

Basically there are many different ways out there. The easiest is indeed what Kevin Reid posted, whose method is shown nicely in this and this videos. However, a much more accurate way, is to calibrate against a GSM station whose known ARFCN frequencies are exact. This is what the ADS-B people usually do with older SDR sticks, before using dump1090. The best explanations I found, are the ones provided by [1] and [2].

The public available code to do this is called Kalibrate was originally made by Joshua Lackey from here. Since then there have been many iterations and alternatives, but at the end of the day the original is the best, as others are either device dependent, broken or dependent on libraries that are too new or outdated. There are now dozens of variations of this.

Finally I just used the old Windows version as the latest Linux one didn't compile. It worked great and made sense. This is a two step process.

  1. Find the GSM channel (ARFCN) with the highest power.
  2. Use this channel to calibrate to.

    # We use an arbitrary gain of 30 dBm and a starting PPM of ~1,
    # and throughout most of Europe we only have EGSM for 2G.

    $ ./kal.exe -g 35 -e 1 -s EGSM -v

    E-GSM-900:
            chan: 6 (936.2MHz + 481Hz)      power: 297161.97
            chan: 13 (937.6MHz + 427Hz)     power: 80136.18
            chan: 45 (944.0MHz + 173Hz)     power: 72538.40
            chan: 74 (949.8MHz + 264Hz)     power: 1357267.92
            chan: 81 (951.2MHz + 157Hz)     power: 91010.07
            chan: 104 (955.8MHz -  35Hz)    power: 426084.14
            chan: 108 (956.6MHz + 128Hz)    power: 61358.88
    Found 1 device(s):
      0:  ezcap USB 2.0 DVB-T/DAB/FM dongle

    Using device 0: ezcap USB 2.0 DVB-T/DAB/FM dongle
    Found Rafael Micro R820T tuner
    Exact sample rate is: 270833.002142 Hz
    Setting gain: 35.0 dB
    meh: Scanning for E-GSM-900 base stations.
    channel detect threshold: 28151.015811

We see that the ARFCN 74 is the strongest one, so we use it for the calibration.


    $ ./kal.exe -v -e 1 -c 74

    average         [min, max]      (range, stddev)
    + 235Hz         [228, 243]      (15, 4.489251)
    overruns: 0
    not found: 0
    average absolute error: 0.753 ppm
    Found 1 device(s):
      0:  ezcap USB 2.0 DVB-T/DAB/FM dongle

    Using device 0: ezcap USB 2.0 DVB-T/DAB/FM dongle
    Found Rafael Micro R820T tuner
    Exact sample rate is: 270833.002142 Hz
    meh: Calculating clock frequency offset.
    Using GSM-900 channel 74 (949.8MHz)
            offset   1: 230.82
            ...

And thus, for this test, we get a PPM of 0.753.

References:

[1] and [2]

For other implementations see:

$\endgroup$
  • $\begingroup$ Rather than cellular for which gsm is not world wide, the second interface on most non consumer gps heads provides a 0.1 PPM time signal. $\endgroup$ – Rowan Hawkins Mar 24 '17 at 18:48
  • $\begingroup$ @rowan do you mean a 10 MHz GPSDO, or are you mixing up Pulse Per Second with parts-per-million? A cheap GPS probably won't have 10 MHz, and the question is really how to find frequency error without buying any test gear. And I don't think all SDRs can even receive 10 MHz. $\endgroup$ – tomnexus Mar 26 '17 at 1:30
  • $\begingroup$ I was referencing the receiver generated PPS signal that is determined by differencing the time from all sats being received. (tekmon.eu/1-3-2-satellite-and-receiver-clock-errors) I can no longer edit the other comment $\endgroup$ – Rowan Hawkins Mar 26 '17 at 9:47
  • $\begingroup$ @RowanHawkins How would you do that? I also was thinking to use code similar of kalibrate (or other) to sync to some GPS. Any practical ideas how this could be done with this SDR? (Perhaps there are already solutions to do this?) $\endgroup$ – not2qubit Apr 7 '17 at 5:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.