laforge at gnumonks.org
Sun May 9 19:47:27 UTC 2010
On Sun, May 09, 2010 at 04:29:06PM +0200, Christian Vogel wrote:
> I've measured how the 26 MHz VCXO changes frequency depending on the
> AFC DAC output. See the attached plots and the data file.
thanks a lot!
> Tapping point is between C132 and C210 (RITA Xout).
> On my 20 MHz scope it's 300 mVpp, but I'm hitting the bandwidth
> limit of the scope, the frequency counter says -0.03 V ...+0.02 V
> after a x10 probe (500mVpp).
You should consider to use the 13MHz signal. it's buffered and a simple
divide-by-two of the 26MHz oscillator output.
Anyway, from your data I would say:
At DAC=500 we have 26 MHz
At interpolated DAC=2750 we have 500Hz more
So this means the voltage control slope is 500/2750 = 0.1818 Hz / LSB
Converting this into ppb: 0.1818 /.026 = 6.993 ppb / LSB
The TSM30-based algorithm seems to assume 4.9 ppb / LSB, and for reasons I
don't remember we currently seem to use 9.76ppb/LSB (AFC_SLOPE in layer1/afc.c)
Of course this value is highly dependent on the particular XO and its
peripheral parts, so I'm not surprised to see that much difference.
Based on this, the ideal value for AFC_SLOPE according to the measurements
should be 229. I haven't tested that, this is just my thinking:
0.006993 * 32768 = 229
Also, to give you some idea what this means in terms of
"Hz carrier frequency / LSB":
Center of GSM 900 between UL and DL: (960+880)/2 = 920 MHz
overall multiplication factor for mid-band GSM900: 35.38
resulting theoretical slope: 6.43 Hz / LSB
Center of GSM 1800 between UL and DL: (1710+1880)/2 = 1795 MHz
overall multiplication factor for mid-band GSM900: 69.04
resulting theoretical slope: 12.55 Hz / LSB
- Harald Welte <laforge at gnumonks.org> http://laforge.gnumonks.org/
"Privacy in residential applications is a desirable marketing option."
(ETSI EN 300 175-7 Ch. A6)
More information about the baseband-devel