Hello LF-enthusiasts,
some of you obviously use the DSP56002EVM.
Therefore, let me again mention a simple method to
synchronize an LF-receiver to an atomic clock.
I do it the following way:
The 24576 kHz of the CODEC oscillator on the EVM board
is the mother oscillator of all frequency-relevant
parts of the receiver. The signal is picked up by a
small capacitor from the board and amplified by a
gain block and then divided by digital CMOS circuitry
as follows:
24576 / 64 / 3 = 128 kHz
These 128 kHz (well LC-filtered) is used to downconvert
the 136 kHz band to about 8 kHz, which is the input to
the left channel of my EVM board.
The digital 128 kHz are divided further:
128 / 2 = 64 kHz
This signal (also LC-filtered) is used to downconvert
DCF77 from 77.5 kHz to 13.5 kHz which is the right
input to the DSP board. Only one receiver is written
in assembly language. It is switched between left and
right input. The DSP program generates cos/sin-pairs
of 13.5 kHz and 7.7...9.8 kHz for downconversion into
the baseband. After the DSP-broadband noise blanker
(correlation of both inputs 136 and 77 kHz) a
multirate filter follows and a quadrature demodulator.
The rotation of the DCF77 signal in the baseband
exactly says at what the 24576 kHz were in error.
So the program can correct this error in software.
In place of DCF77 also MSF and others (50 kHz)
could be used by simple change of the LC-input
filters of the receiver and the downconversion
frequency generated by DDS in the program.
The advantage of such a synchronization is to
allow very high resolution over long periods.
A spectrum of a transatlantic signal could be
made over a whole night at a resolution of
0.01 Hz or even better. The transmitted signal
must have the same quality, off course. The
signal to noise ratio against a resolution of
1 Hz then is better by a factor of 10.
My simple receiver is used therefore the other
way also. At the moment an OPAmp produces only
1 Volt at 50 Ohm at any frequency within the
136 kHz band (no PA, no Antenna).
It should be pointed out that only convertion
frequencies (generated by DDS) and the sampling
frequency are relevant to the receiving or
transmitting frequency, and not the DSP clock
and not the clock of the PC which does the FFT.
Because of heavy QRL, this LF project is growing
slowly.
73 de Klaus, DJ5HG
|
