Some tricks for radio communication
I thought of following systems in order to build devices I needed.
Some of them are classical solutions.
1. The sign detector
A receiving antenna is a device that produces a weak
electric signal. That signal is intended to be fed to the receiving
The trick is to first input the antenna signal into a sign detector.
That is a device that outputs a binary signal: 0 if the antenna
signal is negative, 1 if it is positive. A comparator, an
operational amplifier with a high gain, a c-mos TTL gate or a
saturating receiving antenna amplifier may be used for that purpose.
The fact the sophisticated antenna signal becomes a simple binary
signal allows the rest of the receiving electronics to be mainly a
rudimentary 1-bit digital circuit.
Take for example a superheterodyne FM receiver where the
multiplicator is a precisely adjusted MOS-FET transistor used in his
resistive functioning area. Thanks to the sign detector MOS-FET and
passive components can be replaced by a simple X-OR gate. One input
of the gate receives the binary antenna signal, the other gate
receives a digital clock signal with a frequency slightly different
from the signal frequency. The X-OR gate output may be directly fed
into a rudimentary low frequency filter.
Imagine the use of a microcontroler to be an AM receiver. Thanks to
the signal detector only one digital microcontroler pin will be
necessary to input the signal. And only simple 1-bit instructions
will be used to increment and decrement the two receiving
accumulators, with no fear for saturation or counter loop.
This system may make you fear it implies a loss of information.
There will indeed be a loss of information for receptions with a
high signal/noise ratio. But there will be no loss for receptions
with a weak signal/noise ratio, and that's what matters.
The sign detector offers three main advantages:
- The receiving electronics becomes a lot more simple and
cheaper to build.
- The receiver is no more sensitive to variations in antenna
signal intensity. Indeed, only the sign of the signal matters,
not it's amplitude.
- In the case of an AM receiver, the receiver will directly
calculate the signal/noise ratio.
And four drawbacks:
2. The modulators and demodulators chain
- The sign detector may not be used for extremely high
frequencies, because there exists no digital devices capable of
operating at those frequencies.
- If an AM signal is received too well, stronger than the radio
noise, the sign detector will not be able to allow the
measurement of the AM signal amplitude. Unless an artificial
strong noise signal is intentionally added to the antenna signal
before the sign detector!
- Every electronic circuit generates some radio noise.
Especially if it is the circuit of a receiver it should be more
or less shielded. Otherwise that noise will be received by the
antenna and will lower the performances of the receiver. Digital
circuits are extremely noisy circuits and need therefore a more
I've once been responsible for the crash of a drone build by
friends because one circuit used a microcontroler. Although that
microcontroler was completely surrounded by grounded metal, the
high frequencies it produced have disturbed the radio link with
the pilot. The high frequencies traveled to the drone's radio
receiver through the Vcc and signal wires. The solutions have
been the following ones (depending on the application some
solutions may be pointless):
- The noisy electronic circuit and it's shielding should be
as little as possible.
- All wires should come out of the shielding through one
sole hole inside that shielding.
- To prevent high frequencies from escaping out of the
shielding through the wires, the following can be done or
- Common selfs and coils let the high frequencies
through, because of a capacitive or inductive effect
between input and output loops. That's thus not a good
solution. It is possible to use selfs yet they would
either be very long or use too thin wire. Maybe selfs
can be purchased intended for this purpose yet I could
not find any (or better said: I found a lot but none did
- Common optocouplers are not a good solution because of
the capacitive effect between input and output. High
frequencies cross an optocoupler with no problem. A
short fibre optics should be used or special
optocouplers with a very low capacitive effect.
- Dc/dc converters may be used for Vcc, provided there
is no capacitive effect between input and output. Such
devices too cannot be found in common stores.
- High impedance metalfilm resistors are a very good
solution for signal wires. Something like 100 k or
1 M. High frequencies are blockaded and weakened
heavily when they must pass through such resistor (any
signal will be), and virtually disappear at the output.
The resistors should be located near the output hole
inside the shielding. (Please note for security against
noise and glitches the c-mos inputs receiving a signal
that went through such a resistor should be latched to
ground through a little condensator.)
- Little common condensators, a few nanofarads, provided
they are suited for high frequencies, are a very good
solution: they should be used to latch the wires to the
ground. Best place is where the wire is coming out of
the shielding. For some applications the condensators
gave such a good result the shielding was useless.
(Because condensators are a short-cut for high
frequencies, the outputs that are latched to the ground
that way should first go through a resistor; in order
not to make the output gates heat up and use too much
- A sign detector for weak radio signals may be difficult to
- One solution is to use a high gain receiving antenna
amplifier circuit (possibly two amplifiers put in serial)
and cut out the output signal with a resistor and two diodes
(nothing looks more like a digital signal than a saturated
- The input should not have any sort of hysteresis.
- The circuit should not produce a stronger internal noise
than the noise received by the antenna. If you want to
verify the quantity of noise produced by the circuit itself,
try to measure the signal before it is converted into a
digital signal and see what difference you get when the
input is latched to ground. When the input is latched to
ground, the noise you measure is the one produced by the
- Some circuits have an unpredictable input level, or a
level uneasy to find. For example a HC gate may switch
around 2.49 Volts while another may switch around
2.50 Volts. That's of no importance for common digital
or half-analogic purposes, yet for a weak radio signal it is
a real problem. Provided the circuit is an inverting gate,
the solution is to latch the input to the output through a
high impedance resistor. That way, should the input for
example be too low, the output will be set to 5 Volts
and will load the input condensators until the input level
becomes high enough. (The resistor can be calculated in
order to filter away very efficiently strong low frequencies
like the 50 Hz or 60 Hz AC net current.)
- The current supply for such circuit should often be
perfectly regulated. Do thus not forget to put a good
condensator between Vcc and ground. Sometimes a good
solution is to build such circuit with their own dedicated
A common radio emitter contains one modulator
is a device that produces a high frequency signal whose amplitude
(or frequency, or phase shift) is influenced by the signal you want
to transmit. For example: full amplitude to transmit a binary 1
signal, zero amplitude to transmit a binary 0 signal.
A common radio receiver contains one demodulator
. That is a
device that inputs a high frequency signal and outputs a usable
signal. For example: a binary 1 is output if a high frequency signal
is detected, a binary 0 is output if no high frequency is detected
Suppose we send 1,000 binary bits per second. And we use a high
frequency of 1 Mhz. Then we need clocks with a precision better
than 1/1,000 th.
Suppose we decide to use a higher frequency; 100 MHz. That
frequency being 100 times higher, we are supposed to be able to send
the signal 10 times further. (This is not true in
fact: because there is less environmental noise at higher
frequencies we will have more than a 10 times increase. Unless
there are obstacles like a hill or the earth curvature, then low
frequencies, by diffracting around obstacles and bouncing back on
the ionosphere, can travel a lot further than high frequency
signals. But here we neglect these phenomenons.)
Yet this requires our clocks to have a precision of
Suppose we do not own clocks with such a precision. We only have
clocks with a precision of 1/1,000 th. Then, although we
transmit at 100 MHz, we will have no increase in transmission
A solution is to you use intermediate modulators and demodulators. (This is commonly done for signal encryption or
transmission of several signals on a unique radio frequency. An
example: the stereo difference signal, for standard FM receivers,
is encoded on a subfrequency of 30 kHz that is added to the
common mono signal. Another example: RC systems for modelists use
several modulators in parallel, called "ways", using frequencies
in the order of 10 kHz, to transmit commands to different
actuators inside the model plane, ship or car. The modulators
signals are added before being fed into the radio modulator.
Inside the model the received signal is given to all demodulators
and each one of them, having his own frequency, extracts the
subfrequency intended for him. But here we do it to increase the
We use our binary data signal at 1 kHz to modulate a binary
frequency at 1 MHz, then we use that modulated 1 MHz
signal to modulate a radio signal at 100 MHz.
The receiver will contain a demodulator with an output at 1 MHz
that will be fed into a second demodulator that will render the
1 kHz digital binary data signal.
That way, we will able to transmit further while using low precision
The distance increase will not be of 10, it will be less. There is a
factor two difference. Yet the result is still important.
A chain of more than two modulators and demodulators can be used.
They may be three, four, five... That way it is theoretically
possible to transmit a signal up to any distance, whatever the clock
It is possible, for example, to transmit one bit per hour with a
1 GHz radio signal thanks to two modulators using little quartz
crystals with a precision of 1/1,000,000 th (available in
common electronic components stores for $2). That's very interesting
for special purposes like space probes or measurement and activation
systems in very noisy environments. Would such a transmission system
have been programmed into the processor of the Pioneer 10 probe, we
would still be in contact with it while using a much cheaper antenna
system. Maybe the transmission rate would be of a few bytes per
year, yet that's enough to receive some key information.
Most funny with this system is the fact at big distances the first
demodulator only feeds a noise to the second demodulator. Yet the
second demodulator manages to detect the signal out of that noise!
As mentioned above, this system is not strictly equivalent to using
a single modulator and demodulator stage with high precision clocks.
There is a loss (comparable to the particle/wave packet detection in
quantum mechanics). Therefore the number of stages should be kept as
little as possible and best possible clocks should be used.
Another problem is the fact the radio signal will use a bandwidth a
lot broader than strictly necessary. This may be partially
compensated by the fact several transmissions may work at the same
time, using the same main radio frequency, but using different
subfrequencies. If one emitter emits at 100 MHz a signal
modulated by a signal at 1 MHz that is modulated by a signal at
1 kHz, it will not be heard by a receiver that receives also at
100 MHz but then demodulates at 2 MHz to produce also a
signal at 1 kHz. That way the bandwidth can be used
effectively, without too much wasting. It is simply another way of
sharing the bandwidth.
Other advantages are:
3. The fractal signal
- The possibility to build little and low-cost devices because
no very high precision clocks are needed.
- Telecommunication systems that will for example be a lot less
sensitive to the Doppler effect.
- The possibility to consume a continuous and very weak amount
of electric current, without the problems associated with the
emission of radio "flashes" that consume a lot of electricity an
a short time (this is another trick to emit far with low
The system described above implies information is
transmitted at a given rate; 1 kHz in the examples.
Suppose now you just want to transmit a "ON" signal. To call for
help, for example.
The problem is it is difficult to choose at what speed you want that
signal to be transmitted. Suppose you choose 1 second. Well that's
quick. Okay, but what if you are very far from the receiver? Your
signal will not be heard. Unless you transmit slower, say on one
hour. That way you can transmit 60 times further. But that's stupid
if you were close to the receiver. Why wait an hour before the
rescuers are warned?
When using only one modulation and demodulation stage, the solution
is simple: just let the receivers' accumulators accumulate
information on a one hour basis, but look at their content every
second. If a strong signal arrives it will make the accumulators
increment very fast. That will be noticeable and will allow the
system to react instantly.
But what if a demodulators chain is used ?
The solution is to allow each demodulator stage to have also the
property to detect the signal at his frequency, instead of only
feeding it to the next demodulator. That way, if the signal is very
strong, the first demodulator will instantly react. "I hear it!". If
it is weaker, the first demodulator will only hear a noise. That
noise is fed to the second demodulator stage that will detect the
signal and react, yet his reaction will be slower. If the signal is
too weak for that second demodulator too, the noise fed to the third
demodulator will perhaps make him react, yet still again a lot
slower than the second. And so on, you may use a guirland of
demodulators/detectors, allowing the signal to be detected in a few
milliseconds or in a year...
At each stage there must always be a signal at any time. For example
there must always be a frequency of x Hz or 2 x Hz.
The way the signal changes between the two frequencies is used to
feed the next demodulator. But the demodulator will react if any of
the two frequencies is detected.
When you plot such a modulated signal, you get a one dimensional
fractal pattern. Previous system also had a fractal look, yet here
the full identity of a fractal appears: the fractality can be
developed up to a virtually infinitely low frequency, and every
stage has the same purpose: make a demodulator react.
4. The noise link
In general the basic signal is a sinusoid. That sinusoid is
modulated in order to transmit information.
You may use anything else instead of a sinusoid. You may even use a
noise; a totally non-repetitive random signal.
The only condition is that the receiver must know that noise. It
must know any detail of it and know when it is emitted. Precise
clocks and time shift adjustment techniques must be used.
For secret transmissions, the advantage of using a noise is obvious.
It is totally unhearable for receivers who do not know what noise
exactly is emitted. There are just three drawbacks:
- Output power must be reasonably weak. Otherwise the emitter
can be heard and localized just because it is emitting a strong
- There must not be too much radio echo. Sine waves can deal
relatively well with echo because the random sum of sine waves
is always a sine wave. The noise wave, on the contrary, will
destroy itself if there is too much echo.
- The receivers electronics, in order to find the time shift of
the signal, then follow the time shifts in order to remain
locked, needs some calculation power and clock precision. (Yet
nothing too tremendous.) What's more, the random numbers
generators used by the emitter and the receiver must be of good
quality; non-repetitive and unpredictable, yet both producing
the same numbers. (They may use a combination of calculated
random numbers and shared lists of real random numbers.)
Best way to modulate the noise signal is to invert it. For example
it is emitted non inverted for digital 1 and inverted for digital 0.
The receiver just has to multiply the antenna signal with the noise
and sum the results to get big numbers that will indicate digital 1
if they are positive and digital 0 if they are negative. That way
the emitter emits constantly with the same power. (This method of
multiplicating the signals and summing the results is called a correlation
Again, the time shift is a serious problem. If the receiver is not
locked on the emitter, it must try out all possible time shifts
until the signal is found. Then the signal must be followed
carefully in order not to be lost.
One way to emit a noise-like signal while being closer to usual
techniques is to emit the sum of a wide range of ultraweak
sinusoids. The receiver will also contain a wide range of
demodulators, one for each frequency, yet all demodulators use the
same shared accumulators to sum their sine periods. Those single
accumulators, containing the sum of all sums of all demodulators,
will then be used to tell if there was an emission or not. (Such a
set of demodulators can also be used to determine if for example an
emitter is emitting a continuous series of Dirac wave patterns.)
5. Modulate several bits together
At first hand the method to transmit bits along a very noisy
channel is to modulate the status of each bit and send each bit the
one after the other. Yet there is a drawback. If say a hundred or a
thousand modulation periods are used for each bit then the
transmission yield will be quite far away from the maximum baudrate
allowed by the Shannon theorem. A better yield is achieved when a
lot more modulation periods can be used for each bit. Say ten times
up to hundred times more. Yet this would imply a slower baudrate.
the solution is to transmit several bits together, each with its own
modulation. If ten bits are transmitted together, then the number of
signal periods available for each bit will be multiplicated by about
Eric Brasseur - 1 January 1997