A transducer to harvest electric power from radio waves
Or how to get electric power out of radio waves.
Years ago, I invented the "snake". It's just a wire, with a LED,
some elementary electronics and no battery. In most cities, when
holding the snake in the direction of a 100 MHz FM broadcast
emitter, it will harvest enough electric power from the radio waves
to make the LED light up, often even at a few kilometers distance
from the emitter. The intention was to make a simple gadget that can
be folded inside a letter and that can show the presence of radio
waves, with no doubt about their power because it is that power that
feeds the LED.
The schematics of the snake and building tips are available in this
I did try to adapt the snake to 900 MHz cell phone radio waves. I
build prototypes like this one:
The LED did light up, up to a few meters distance from weak
in-street cell phone towers. This was not of much use, since I
wanted to show the presence of those waves everywhere. And their
The problem lies in the fact that cell phone waves are 9 times
shorter than FM broadcast waves. The snake, being a half-wave
antenna, has a length of 1.5 meters. The LED needs 1.5 Volts to
light up. So, roughly, the snake for FM broadcast waves just needs
an electromagnetic field with a force of 1 V/m. The snake for cell
phone waves, being 9 times shorter, needs in the order of 9 V/m.
Instead of a LED, I tried to use a galvanometer. A 100 µA scale
galvanometer needs almost no electric tension to operate. But the
snake did not produce enough electric current...
Of course I tried many layouts, like Yagi-Uda reflectors and
directors or several snakes put in series. It was bulky and too
directional. I did get good results with 4 snake-like elements put
in series, that also were each other's parasitic element. The LED
would light up maybe 20 meters away from a cell phone tower.
Great... but still not useful.
Till I thought of this layout. The copper wire has a diameter of 0.1
millimeters and the LED is a standard 5 millimeters one.
It does harvest enough power to be connected to a 100 µA
galvanometer, which allowed to assemble this box:
In most open spaces in a city, like public spaces, docks and
bridges, the needle will rise at least a few lines. It should rise
on most rooftops and in most rooms that have windows above the
The needle will rise firmly if a big cell tower is in sight, even
maybe 200 meters away. And of course it will go berserk when close
to an in-street little cell tower. In the photo below, the antenna
is the white little rod just right of the galvanometer.
Places where the needle stays on zero will for sure be considered as
safer by electrosensitive persons. But I don't know if it can be
used by them to check for safe places. One main drawback is that
this is a resonator, dedicated to one frequency: the lower band for
cell phones in Belgium, around 900 MHz. This band, and the FM
broadcast band at 100 MHz, are the main overall sources of
electromagnetic pollution. But in order to be safe, an
electrosensitive person should rather use a broadband electronic
This is an inside view of the device. The box is that of an Eachine
H8 quadcopter turned inside out.
The central part of the transducer is a standard diode bridge yet
made of high frequency diodes. I used HSMS-285C-BLKG
ones. They are very sensitive. The capacitor in the middle is not
mandatory but it proved to be useful. I used a 100 pF high frequency
Each of the 8 antenna wires has a length of ½ λ
Hence for a frequency f
of 900 MHz, with a wavelength λ
of c / f
equal to about 33 centimeters, the length of each
wire would be 16.7 centimeters, which you want to correct because
the speed of a frontwave in a copper wire is slightly lower than c.
I used 16 centimeters wires. This page can help do calculations:
You must experiment and try out different lengths. In one
application for cell phone waves, the optimal length of the antenna
turned out to be about 2/3 of the calculated length! When using thin
wire, the window will be of about 10%. In other words, if for a
given frequency the optimal length of an antenna is 100 çirek then
it will dimly start working at a length of 90 çirek and stop working
above 110 çirek. It will work really well between 95 çirek and 105
çirek. If you need a broader window, use thick rods and some other
If you glue the wires on thick wood then they must be a little
longer, because the diamagnetism of the cellulose will increase the
frontwave speed of the wires.
My first trial was with the wires going straight up. This gives
quite good results. The middle part of each wire must be at a
distance of about ⅒ λ
of the center line of the
device. (I used a little more; about 3.5 centimeters, for the 900
MHz transducer. Hence the distance between two diagonally opposed
wires is 7 centimeters.) Each set of 4 wires will form a "cage" with
a square floor section. At the top and bottom, each set of 4 wires
is soldered together; they are electrically joined. Four wires, at
this given distance of each other, is an optimal layout to harvest
power, I believe because each wire is a parasitic element for the
The layout shown below is better. The end of the wires are turned
inwards. It doesn't matter that they are no more aligned with the
radio waves, because the ends of a dipole antenna are the least
sensitive to the radio waves (yet they are necessary for the
standing wave that develop inside the wire). The most important
advantage is that the middle parts of the wires are now closer. This
decreases the directivity and it allows the device to operate in a
narrower space, possibly making a better use of a zone where echos
of the radio waves add up. The fact that this also uses less space
is a bonus. The height of the transducer as a whole is 14
The main aim of this layout is to maximize the potential difference
across the diodes. When the standing waves in the upper wires
present their lowest potential to the diodes, the standing waves in
the lower wires present their highest potential, which gives the
most chance to rise above the forward voltage of the diodes.
The standing waves will increase inside the wires till the potential
becomes sufficient for some current to flow through the diodes and
load the capacitor.
If you want to build a parasitic element for the transducer as a
whole, assemble another same transducer yet with no diode bridge in
the middle (do not connect the bottom wires to the upper wires).
A more conventional looking layout would be like below. It should
not be called a "full wavelength dipole antenna" because the
resonances in the two wires are in phase. It rather is a "double
half wavelength dipole antenna with a high impedance". I did not
experiment with it.
If you want a simple device that can harvest energy from strong
radio waves, like close to a cell phone, use a dipole with its ends
connected to the diode bridge. That would be like the antenna above
but with one wire removed and replaced by the other end of the
remaining wire. Hence the green wire shown in the picture below
should have a length of ½ λ
, probably a
little less, like all the ones above. (Of course you can also use 4
wires like above, placed ⅒ λ
from the diode bridge
in the center. But then it's no more the simplest device and it
won't be handy close to a cell phone.)
By the way, the part of the wire that receives the radio waves is
the continuous line on the right. The two parts on the left, that
connect to the diodes, are useless for the radio waves but they are
mandatory for the wire to have the adequate length to resonate at
the desired frequency. And to connect to the diodes. (A regular
"folded dipole" is another type of antenna; with double the length
for a given wavelength.)
An example of inappropriate use of the formulas is when I computed
that a 0.5 millimeters wire should be 1.46 meters long to build such
a folded dipole for 100 MHz waves (½ λ
Later on an experiment yielded that the shortest length to get a LED
to faintly light up was 1.15 meters and the longest length was 1.45
meters. The optimal length, with the LED lighting up decisively, was
broadly around 1.30 meters.
are a strong and safe choice for powerful applications.
Whatever you connect to the diode bridge, be it a LED or a
galvanometer, if you want to get the most out of the device you will
need to cope with the impedance. The impedance of those transducers
is quite high, which means for example that when the diode bridge
produces a tension of 20 V, it will only be able to produce a
current of say 1 mA. If you connect this to one LED, the tension
will fall to 1.5 V but the current will still be 1 mA. So you are
wasting most of the available power. One simple solution is to use
several LED in series. Three LED will need a tension of about 4.5 V
and they will be traversed by still the same current of 1 mA,
roughly speaking. Hence you get three times more light, still
roughly speaking. A more sophisticated solution is to use a little
DC/DC converter, that will adapt the impedance. For example a
converter that can take any tension like 6 V or 20 V with a weak
current and produce a low tension of 3 V or 2 V with a higher
If the LED is meant to light up dimly, then my favorite choice is a
hyper red one, like the L-7113SEC-J3
If the LED is meant to light up very bright, then it will be a high
yield white one.
I built a transducer for 100 MHz radio waves, hence 9 times bigger.
Each of the 8 wires in the photo below is 135 centimeters long,
despite ½ λ
being 150 centimeters. The
optimal wire length for a closed dipole with the same wire is 130
centimeters, which would mean that when connecting an end of an
antenna wire to diodes it must loose 5 centimeters. The diameter of
the electric conductor is about 0.5 millimeters.
It manages to harvest enough power to make a buzzer sound. After
some experiments I converged to a buzzer with an impedance that
matches that of the transducer (about 2 kΩ) and that can sound dimly
with only 1 V. I also built something around it to separate the
front wave from the back wave and focus both. That would be the
transparent plastic structure on next picture.
These two videos show the result. The first one is on a dock of the
Meuse river, at about 800 meters distance from the emitter. Downtown
Liège gets maybe half that amount of radio pollution. The second
video is in the Parc de la Citadelle, much closer to the emitter.
Eric Brasseur - July 24
till July 8 2017