A DCF77 Atomic Clock (Mainflingen/Frankfurt)
Like the MSF atomic clock project described elsewhere, the installation
in Mainflingen is a highly accurate atomic clock. The transmission
frequency is 77.5kHz and the output power of the Mainflingen clock
is considerably higher than the one in Anthorn. Hence this clock can be
received over most of Europe
The finished project must be connected to digital input port 4
of a Velleman
K8055 interface board. The antenna should be aligned with Frankfurt in
Germany though I have found that I get the best results pointing the
thing in an entirely different direction! Notice that in the photograph
below the small antenna supplied with the receiver has been replaced by a
much larger home-brew antenna. Without this swap I could get no clean
reception.
Circuit description
The heart of the project is a small circuit board which
is made by a firm called Symtrik. They list it as the SYM-RFT-77
module. The module is sold combined with a small tuned ferrite rod
antenna for under 8 pounds. The components are mounted on a
small wooden board. This keeps the PCB and the rather heavy
antenna safely together. The receiver module is in fact the same as used
for the MSF 60kHz clock, except for a different ceramic crystal and a
ferrite rod antenna tuned to 77.5kHz rather than 60kHz.
Version 2
I have always had tremendous problems with the reception of the Mainflingen signal, and I blamed my location - the far North of Northern Ireland. I now blame the receivers. Our local Lidls shop sold a radio controlled clock the other day (this is written on the 18th of April 2012). I bought one on spec because I wanted the 77.5kHz crystal and the clock was only £5. It turns out that it contained one of the best atomic clock receivers I have come across. The receiver is easily removed and is a self-contained unit (shown on the right). The output is rock steady. It can also easily be converted to a MSF receiver by swapping the crystal and using a 60kHz antenna. Not bad for a fiver!
I designed a small circuit to adapt the radio to the K8055. The two
diodes drop the 5V supply voltage to make it suitable for the receiver. 
It
operates at 3V, though for all I know it can cope with 5V as well. As I
like to have a flashing LED and the original - now useless - clock came
with a nice blue one, I fitted a 7404 logic chip as a driver. This gave
me plenty of NOT gates which I used to give a normal and inverted
output to both the LED and the output pin. Two links select your choice.
Anybody interested in the PCB layout, please drop me a line.
The receiver works a treat. It is stationed about 3 meters away from my computer and shows no inteference whatever. The signal comes in at a steady 100ms for 0 and 200ms for a one and is quite predictable. The output is connected to the digital input 4 of the Velleman board. You can test the reception with the help of my Slowscope program. A trace can be seen below.
If you have bad reception and you run your receiver via a mains
power supply try using batteries. This can make a huge difference.
The trace is on a scale of 250ms/div. As can be seen the input is non-inverted and the 1s at 6,10 and 18 are twice as wide as the 0s at 2, 21 and 25.
The Program
The DCF clock takes exactly one minute to transmit the date and time.
The timing diagram of three seconds is shown below. The
decoding scheme is different from the one used by the MSF clock, though there are similarities.
Every second carries one bit. To show a 0 the carrier amplitude is
reduced to 30% for 100ms and for a one for 200ms. Second 0 of the
sequence is high for the full second.
Some translations may be of use here. Starting anti clockwise from the top:
Kalenderjahr - year
Kalendermonat - month
Wochentag - day of the week. Monday being 1 and Sunday 7
Stunde - hour.
I leave it to the imagination of the reader to figure out what Minute stands for.
The outside numbers indicate the binary coded decimal place values of the
set bits, the inside numbers are second indicators.. An
example will make things clearer. All time values are transmitted via
bits 21 to 57. The data for the year, for example, are
represented by bits 50 to 57. This
article was written in 2011, so bit 51 and bit 54 will be high. This
is because bit 54 is worth 10 and bit 51 is worth 1. Add these two
numbers and you get - you've guessed it - 11. The 2000 is assumed.
Say the year is 2087. 87 = 80 + 4 + 2 +1, hence bits 57,52,51 and 50 would be set.
The program recognises the beginning of the cycle by
measuring the length of the high bit. This is always larger than 130ms at the
beginning of the sequence, which is a unique value.
The fourth line shows the pulse train for one minute. Cyan is the previous minute, white the current minute. The colour
changes time about so that the new bits are always visible.
The next line indicates the purpose of every bit. Take the hour
for instance. This frames the bits 011010. They come in in reverse
order, so this should read 010110 which is one 2, one 4 and one 10 = 16.
i.e. four in the afternoon European time.
The bottom
line displays the decoded calendar and time information, again in two
colours.
Below is a link to an
archive with the BBC
BASIC program file and also a self-contained .exe file of the
program below. Notice that the
output is in central European time which differs from Britain by 1 hour.
The program can cope with both an inverted and a non-inverted signal.
Please note that this latest version of the program uses the digital input 4 rather than an ADC input of the Velleman board.
K8055_DCF_4_0.zip
Here is a video of the screen output of the program
Here is a link to a video of the signal as displayed by my Slowscope program
And here is the program adapted for the K8061
K8061_DCF_atomic_2.zip
The fourth line shows the pulse train for one minute. Cyan is the previous minute, white the current minute. The colour changes time about so that the new bits are always visible.
The next line indicates the purpose of every bit. Take the hour for instance. This frames the bits 011010. They come in in reverse order, so this should read 010110 which is one 2, one 4 and one 10 = 16. i.e. four in the afternoon European time.
The bottom line displays the decoded calendar and time information, again in two colours.
Please note that this latest version of the program uses the digital input 4 rather than an ADC input of the Velleman board.
K8055_DCF_4_0.zip
Here is a video of the screen output of the program
Here is a link to a video of the signal as displayed by my Slowscope program
And here is the program adapted for the K8061
K8061_DCF_atomic_2.zip
