SoftFM/NOTES.txt

This file contains random notitions
-----------------------------------


Valid sample rates
------------------

Sample rates between 300001 Hz and 900000 Hz (inclusive) are not supported.
They cause an invalid configuration of the RTL chip.

rsamp_ratio = 28.8 MHz * 2**22 / sample_rate
If bit 27 and bit 28 of rsamp_ratio are different, the RTL chip malfunctions.


Behaviour of RTL and Elonics tuner
----------------------------------

The RTL chip has a configurable 32-tap FIR filter running at 28.8 MS/s.
RTL-SDR currently configures it for cutoff at 1.2 MHz (2.4 MS/s).

Casual test of ADC mismatch:
 * DC offset in order of 1 code step
 * I/Q gain mismatch in order of 4%
 * I/Q phase mismatch in order of 1% of sample interval

With tuner in auto-gain mode, device autonomously switches between gain
settings during a run. The LNA gain seems to switch between ~ 24 dB
and ~ 34 dB without intermediate steps.

With RTL in AGC mode, the level of the digital sample stream is normalized
to -6 dB FS. Unknown whether this is an analog or digital gain stage.

At first I suspected that AGC mode may be a cooperation between the RTL and
the Elonics tuner. I thought that the RTL would monitor the level and send
digital control signals to the Elonics to dynamically change the tuner IF gain.
However that is in fact NOT what happens. (Manually changing IF gain in AGC
mode causes a brief level spike, while manually rewriting the same IF gain in
AGC mode does not have any effect).
It seems more likely that AGC is a digital gain in the downsampling filter.


Default settings in librtlsdr
-----------------------------

Elonics LNA gain: when auto tuner gain: autonomous control with slow update
                  otherwise gain as configured via rtlsdr_set_tuner_gain
Elonics mixer gain: autonomous control disabled,
                    gain depending on rtlsdr_set_tuner_gain
Elonics IF linearity: optimize sensitivity (default), auto switch disabled
Elonics IF gain: +6, +0, +0, +0, +9, +9 (non-standard mode)
Elonics IF filters: matched to sample rate (note this may not be optimal)
RTL AGC mode off


Effect of IF signal filtering
-----------------------------

Carson bandwidth rule:
  IF_half_bandwidth = peak_freq_devation + modulating_freq

In case of broadcast FM, this is
  75 kHz + 53 kHz = 128 kHz  (worst case)
  19 kHz + 53 kHz =  72 kHz  (typical case)

Simulations of IF filtering show:
 * narrow IF filter reduces noise in the baseband
 * narrow IF filter causes gain roll-off for high modulating frequencies
 * narrow IF filter causes harmonic distortion at high modulating deviation

IF filter with 100 kHz half-bandwidth:
 * baseband gain >= -1 dB up to 75 kHz
 * less than 0.1% distortion of modulating signal at 19 kHz peak deviation
 * ~ 2% distortion of modulating signal at 75 kHz peak devation

IF filter with 75 kHz half-bandwidth:
 * baseband gain ~ -3 dB at 60 kHz, ~ -8 dB at 75 kHz
 * ~ 1% distortion of modulating signal at 19 kHz peak deviation

Optimal IF bandwidth is probably somewhere between 75 and 100 kHz, with
roll-off not too steep.
Weak stations benefit from a narrow IF filter to reduce noise.
Strong stations benefit from a wider IF filter to reduce harmonics.


Effect of settings on baseband SNR
----------------------------------

Note: all measurements 10 second duration
Note: all measurements have LO frequency set to station + 250 kHz


STATION  SRATE   LNA    IF GAIN   AGC   SOFT BW    IF LEVEL  GUARD/PILOT

radio3   1 MS/s  24 dB  default   off   150 kHz     0.19      -62.6 dB/Hz
radio3   1.5 MS  24 dB  default   off   150 kHz     0.19      -62.7 dB/Hz
radio3   2 MS/s  24 dB  default   off   150 kHz     0.18      -62.7 dB/Hz

radio3   2 MS/s  34 dB  default   off   150 kHz     0.46      -64.0 dB/Hz
radio3   2 MS/s  34 dB  default   off    80 kHz               -64.0 dB/Hz
radio3   2 MS/s  34 dB  default   off   150 kHz adccal        -64.0 dB/Hz

radio4   2 MS/s  24 dB  default   off   150 kHz     0.04      -41.1 dB/Hz

radio4   1 MS/s  34 dB  default   off   150 kHz     0.06      -43.3 dB/Hz
radio4   1 MS/s  34 dB  default   off    80 kHz               -51.2 dB/Hz

radio4   2 MS/s  34 dB  default   off   150 kHz     0.10      -42.4 dB/Hz
radio4   2 MS/s  34 dB  default   off    80 kHz               -48.2 dB/Hz
radio4   2 MS/s  34 dB  default   off   150 kHz adccal        -42.4 dB/Hz

Conclusion: Sample rate (1 MS/s to 2 MS/s) has little effect on quality.
Conclusion: LNA gain has little effect on quality.
Conclusion: Narrow IF filter improves quality of weak station.
Conclusion: ADC gain/offset calibration has no effect on quality.


STATION  SRATE   LNA    IF GAIN   AGC   SOFT BW    IF LEVEL  GUARD/PILOT

radio3   2 MS/s  34 dB 9,0,0,0,9,9   off 80 kHz     0.38      -63.2 dB/Hz
radio3   2 MS/s  34 dB 0,0,0,0,3,3   off 80 kHz     0.03      -61.5 dB/Hz
radio3   2 MS/s  34 dB 0,0,0,0,3,9   off 80 kHz     0.07      -61.5 dB/Hz
radio3   2 MS/s  34 dB 0,0,0,0,3,15  off 80 kHz     0.11      -60.5 dB/Hz
radio3   2 MS/s  34 dB 0,0,0,0,9,15  off 80 kHz     0.22      -61.0 dB/Hz
radio3   2 MS/s  34 dB 0,0,0,0,15,15 off 80 kHz     0.36      -61.4 dB/Hz
radio3   2 MS/s  34 dB 0,0,6,0,15,15 off 80 kHz     0.66 CLIP -63.5 dB/Hz
radio3   2 MS/s  34 dB 0,6,0,0,3,3   off 80 kHz     0.07      -62.9 dB/Hz      
radio3   2 MS/s  34 dB 9,3,0,0,3,3   off 80 kHz     0.14      -63.9 dB/Hz
radio3   2 MS/s  34 dB 9,9,0,0,3,3   off 80 kHz     0.26      -64.2 dB/Hz
radio3   2 MS/s  34 dB 9,9,3,2,3,3   off 80 kHz     0.45      -64.1 dB/Hz
radio3   2 MS/s  34 dB 9,9,6,0,3,3   off 80 kHz     0.49      -63.7 dB/Hz
radio3   2 MS/s  34 dB 9,9,9,0,6,3   off 80 kHz     0.77 CLIP -61.2 dB/Hz

radio4   2 MS/s  34 dB 9,0,0,0,9,9   off 80 kHz     0.09      -41.5 dB/Hz
radio4   2 MS/s  34 dB 0,0,0,0,3,3   off 80 kHz     0.01      -36.7 dB/Hz
radio4   2 MS/s  34 dB 0,0,0,0,3,9   off 80 kHz     0.02      -39.9 dB/Hz
radio4   2 MS/s  34 dB 0,0,0,0,3,15  off 80 kHz     0.03      -40.4 dB/Hz
radio4   2 MS/s  34 dB 0,0,0,0,9,15  off 80 kHz     0.06      -39.9 dB/Hz
radio4   2 MS/s  34 dB 0,0,0,0,15,15 off 80 kHz     0.09      -40.9 dB/Hz
radio4   2 MS/s  34 dB 0,0,6,0,15,15 off 80 kHz     0.17      -38.4 dB/Hz
radio4   2 MS/s  34 dB 0,6,0,0,3,3   off 80 kHz     0.02      -36.9 dB/Hz
radio4   2 MS/s  34 dB 9,3,0,0,3,3   off 80 kHz     0.03      -37.5 dB/Hz
radio4   2 MS/s  34 dB 9,9,0,0,3,3   off 80 kHz     0.07      -39.4 dB/Hz
radio4   2 MS/s  34 dB 9,9,3,2,3,3   off 80 kHz     0.11      -38.5 dB/Hz
radio4   2 MS/s  34 dB 9,9,6,0,3,3   off 80 kHz     0.12      -38.0 dB/Hz
radio4   2 MS/s  34 dB 9,9,9,0,6,3   off 80 kHz     0.22      -37.5 dB/Hz

Conclusion: IF gain has little effect on quality, although very low gain
            has an adverse effect. Librtlsdr defaults look good.


STATION  SRATE   LNA    IF GAIN   AGC   SOFT BW    IF LEVEL  GUARD/PILOT

radio3   2 MS/s  34 dB  default   off    80 kHz     0.36      -62.1 dB/Hz
radio3   2 MS/s  34 dB  default   ON     80 kHz     0.36      -61.6 dB/Hz

radio4   2 MS/s  34 dB  default   off    80 kHz     0.11      -38.5 dB/Hz
radio4   2 MS/s  34 dB  default   ON     80 kHz     0.36 CLIP -38.0 dB/Hz

Conclusion: AGC mode has little effect on quality.


Local radio stations
--------------------

radio2      92600000  (good)
radio3      96800000  (good)
radio4      94300000  (bad)
qmusic     100700000  (medium)
radio538   102100000  (medium)
radio10    103800000  (bad)
radio west  89300000  (medium)


Stereo pilot frequency
----------------------

Measuring 19 kHz pilot frequency vs Internet NTP (ADSL):

radio2 19 kHz pilot = 18999.79 Hz +- 0.04 Hz    (7 hours measurement)
radio3 19 kHz pilot = 18999.8  Hz +- 0.05 Hz    (5 hours measurement)

--