Results seem off & missing documentation
tapiov opened this issue · comments
Hi,
I have been running kalibrate-rtl in an effort to confirm my manual calibration results. Kalibrate-rtl returns crystal correction values around -24 ppm while using known beacons the manual calibration is in the range on 55-65 ppm.
I used LTE-Scanner and got ppm of 64.6 which agrees well with manual calibration above.
Another question, the domain listed for documentation, http://thre.at/kalibrate, is gone. Anybody have another source for documentation?
me too, I have two dongles
1st: kalibrate-rtl report -33 ppm, manual calibration is 54
2nd: kalibrate-rtl report -3 ppm, manual calibration is 88
I use local UHF HAM repeaters for manual calibration, never tried LTE-Scanner since its Linux only.
After some playing with kalibrate-rtl and using NOAA weather radio to manually calibrate what I learned was:
- For the US use GSM850, the higher frequencies (1,800 - 1,950 Mhz) are out of range of the R820T's but will work with the E4000's
- kalibrate-rtl can only automatically correct for small PPM valules (< 20-30 ppm)
- Given how large many of the RTL errors are, you need to supply a guess to put kalibrate-rtl in the right range. If you don't, it will give you the wrong answer.
- (I also think there maybe a but here, possibly involving a signed short, offset frequencies seemed to flip at different corrections from -32,500 to +32,500.)
- kalibrate-rtl finds mobile/cellular tower transmissions, but it doesn't really know what the correct frequency/channel number is, so it is making a guess base on closest expected frequencies. This is why if the error is too large, it will think it is hearing a different frequency and give you the wrong value.
- rtl_test -p, if you have a new enough version (keenerd) can give you a reasonably accurate starting point in many situations. I think this test isn't or wasn't reliable on Windows.
- In the US NOAA weather radio is known to be keep on frequency with a high degree of accuracy. Broadcast TV, such as the ATSC pilot tones are also carefully calibrated.
- Ham repeaters are generally not very well calibrated by comparison.
- The tuners in the RTL sticks have a tuning error that varies by frequency due to the design and software implementation. Some recent versions of librtlsdr take some steps to prevent this.
Hope this helps in the absence of docs.
For rough estimate of PPM (can find correction up to 1000 PPM) you can try https://github.com/viraptor/fm_tune which uses local FM stations.