The amplifier has three stages, each equipped with a usual NPN small
signal transistor (you can use any available type):
The first stage is a voltage amplifier with a resonant LC circuit
for 77.5 kHz in its collector. To reduce load influences from the
next stage on the resonant circuit, the capacitor of the LC is divided
into a large and a small capacitor and so divided by 4.5 with that
The second stage is a similar voltage amplifier but with a slightly
smaller inductivity and larger capacitors. As the next stage is not
interfering the LC resonant circuit, no voltage division is made here.
The third stage is an emitter follower without any amplification
and drives the AM rectifier stage, it just decouples the high
resonance resistance from the low impedance of the diodes.
The gain of the first two stages is extremely high, due to the very high
resistance of the LC circuit in the collector at resonance (approx.
161 kΩ in stage 1 and 70 kΩ in stage 2). And this
high gain is frequency-specific. Each stage amplifies the signal by
I also tried an additional stage of a similar design. With that stage
I had to reduce the gain because of self-oscillation. I tried the diode
attenuator as well as reducing the emitter capacitors and increasing
the emitter resistors: it is all the same, the applicable gain of
stage 3 is of no use, so it does not make any sense to add it.
The diode attenuator on the first and second stage input works as
follows. Increasing the current through the diodes reduces their
resistance, which is R = UDiode / IDiode. At the
highest current here, IDiode is (5 - 2 * 0.6) / 1 k =
3,8 mA, so the diode resistance is RD = 0.6 / 3.8 =
158 Ω. As both diodes are parallel to ground (the upper one
direct, the lower one via the capacitor) the resistance of the two diodes
is 79 Ω.
With a capacitor of 1 nF, its capacitive reactance ZC
is ZC = 1 / 2 / Π / 77500 / 1E-9 = 2,053 Ω.
The capacitor and the two diodes make up a resistor divider that
attenuates the signal to the 0.0385 fold, or with a factor of 26.
Both attenuators reduce the gain of the amplifier by the 676 fold.
In stage 1 the diode attenuator following adds 1 nF to the C12
capacitor. That reduces the resonance frequency of the LC combination
slightly. This effect is calculated in the Libre-Office spreadsheet
here. The resonance
frequency shifts from 77.8 down to 77.15 kHz, still is within
the bandwidth of DCF77 and has no negative consequences.
The same calculation sheet calculates the influence of the emitter-base
capacitor of the transistor, in this case the stage 3 capacity. This
has positive consequences as it reduces the resonance frequency down
from 77.95 to 77.82 kHz. Because the tolerances of the parts used
have a much higher influence, this is rather of an academic nature.
This is the transistor amplifier on the breadboard. Make sure that
the cross antenna is at least in a distance of 15 to 20 cm of
the inductivities to avoid feedback and self-oscillation.
The signal of DCF77 is highly amplified, the gain can be reduced by
applying current through the diode attenuators. I use a trim resistor
which drives the base of a PNP transistor with 1 kΩ on its
emitter (with the collector on ground) to get enough diode current.
This stage is also required when driving the AGC with a pulse-width
modulated signal from an AVR. This allows a fine tuning of the gain.
This here is the rectifier for the amplitude-modulated RF signal. Two
Germanium or Schottky diodes rectify and double the DC made from the
RF, with two capacitors of 470 nF. The resistor of 33 kΩ
unloads the capacitors during the amplitude drop of the DCF77 signal,
with a half-life time of approximately 10 ms. The following RC
with R=10kΩ and C=470nF reduces humming of the 77.5kHz signal,
and a clean signal, to be fed into an ADC stage of an AVR.
This is the produced DC voltage for different AC voltages. The
rectifier does not work below 0.4 Vpp input voltage due
to the diodes. But it provides enough DC voltage for normal
RF or IF signals.
This is the unload curve of the RC combination with C=470nF and
R=33kΩ. With t = 0.69*R*C = 0.01 s it is steep enough
to detect the 100 resp. 200 ms long amplitude drops when
DCF77 transmits a zero or a one.
In red the delayed drop on the R=10kΩ/C=470nF filter can be
seen. It is slightly delayed, but drops down with a similar speed
like the voltage on the input.
Calculation of those curves was performed with the OpenOffice
spreadsheet here. The sheet
SimRectifier simulates for a frequency of 77.5 Hz
and for the diverse parts of the rectifier and for a selectable
resolution. Fields that require an input are with a green
background color. It simulates
the voltages on the two rectifier capacitors (in columns C
and D and their sum in column E,
the drop in amplitude with a selectable level, starting
at a selectable time, and
the voltage on the RC filter output.
From that simulation the ripple of the voltage on the rectifier
capacitors was also calculated. It is below 0.25 mV and
remains below one digit of a 10 bit ADC. Only if the capacitor
values down to one tenth or the reduction of the resistor down
by a factor of 20 yields one ADC digit. This would increase
the amplitude drop speed, but would also decrease the voltage
Even though the rectifier RC does not produce humming I added
the 10k/470nF RC filter. In practice humming was larger than
simulated here, so this filter was necessary for a clean signal.
The rectifier hardware and the calculation tool can also be used
for smaller frequencies, e. g. for the 32.768 kHz IF of a
superhet. It doesn't work with an IF of 455 kHz or
10.7 MHz, though.
The amplitude drop when receiving zeroes and ones and during the
2-second long missing drop when the minute is over leads to a
reduction of the long-term average of the signal, when averaged
over a time period of longer than one minute (as done in an AVR
or in a long-term RC filter). The following parameters were used
in this simulation:
Signal DC of
2 V without amplitude drop,
0,4 V with amplitude drop,
averaging by an RC filter of
with a time constant of t = 8,5 s.
Such a RC combination would be chosen if the DCF77 signal would
be used to adjust the gain of the RF or IF amplifier, e. g. in
a TCA440 or for a diode attenuator. In those cases the reception
of a zero or a one shall not lead to a relevant gain adjustment.
One can see that the average voltage is at 1.85 V (approx.
93% of the 2 V on the input) and differs only by +/-10 mV
during a zero or a one. During a minute change, the level change
is slightly higher and around +/-25 mV.
From that one can see that long-term averaging (via software or
with an RC filter) is an appropriate method.
2.3 Automatic regulation
For the automatic regulation of the frequency (AFC) and of
the gain (AGC) as well as for the complete decoding of the
DCF77 signal, including a serial interface for transmitting
the data to another controller that displays date and time
received, a controller of the type ATtiny45 has been developed.
This reads the rectified voltage, analyzes the voltages,
derives controls and adjusts AFC and AGC via two PWM channels
and, by detection and checking of voltage drops, decodes the
zeroes and ones received from DCF77.
The description of that TN45 controller can be found
In order to determine the filter properties of the two LC
resonance circuits in the collector of the amplifier stages
a generator was designed and built that allows to measure
those filters around 77.5 kHz. It produces sine waves
with adjustable frequencies between 70 and 80 kHz.
The amplitude of the oscillator is 4 Vpp at an
operating voltage of 5 V.
Those are the filter curves with different coupling
capacitors. The maximum of resonance with a 330pF
capacitor is not at 79 kHz (as calculated) but
by 5 kHz lower at 74 kHz. This is, on one
hand, due to the coupling capacitor (when fully in
parallel 70.2 kHz), but is also due to straying
effective values of L (5%) and C (10%).
The curve is rather broad and not very steep. It covers
+/-2.5 kHz for the amplitude drop down to half
When decreasing the coupling capacitor to 68 pF
(with a ZC of approximately 30 kΩ)
the resonance frequency increases. Reactance of the LC
is larger than 11 kΩ at resonance.
Due to the high bandwidth of the LC circuit it does not
make much sense to adjust the two LC circuits for 77.5 kHz.
Compared to a simple resistor in the collector, an un-adjusted
LC circuit is an immense advantage. Especially RF far away
of the 77.5 kHz (short wave, 90 kHz power supplies,
etc.) are not amplified.
Those who need it more narrow, because their power supply,
energy saving lamp or old valve TV transmits at 80 kHz,
can use a superhet with a more narrow filter, as also described
on this web page.