|
| |
|
SHORT RANGE COMMUNICATIONS
USING MAGNETIC PULSES
David A. Johnson, P.E.
|
|
|
Define Problem
Possible Solutions
Solution Evaluations
Design Ideas
Planned Experiments
|
|
|
PROBLEM:
How do I send sensor information from one side of a structure to the other side,
without penetrating the walls? Perhaps the structure is the concrete foundation of a home
or a high rise building. It might also be a standard wood frame building. The data I
imagine sending would not require a high data rate. Perhaps the data would be weather
information such as: temperature, humidity, wind speed and barometric pressure. Or maybe
it is sensor information for a lawn irrigation system or temperature data for a roof
mounted solar heating system. The technique might also be used to send useful information
such as gas, electricity or water consumption from one side of a building to the other.
Rough design requirements The signal needs to have a range of only three or four feet.
The signal needs to be able to penetrate concrete walls that may have steel reinforcing
rods. Since it is often difficult to tap into a standard power source outside, the
technique needs to be able to operate for long periods of time powered only by a small
battery. It may also be possible to extend the useful operating time with the use a small
solar array that would keep a battery charged. However, replacing a single battery once
each year may be acceptable.
TOP
|
|
|
Possible Solutions:
Among the possible methods to transfer the data are: radio transmitter, ultrasound,
earth ground signals and magnetic coupled coils.
TOP
|
|
|
Solution Evaluation:
Radio transmitter methods may require too much power. An ultrasound technique might
work for a rigid structure such as concrete but will most likely not work for wood frames
walls. Earth ground signal methods may be too unpredictable. The best candidate looks like
a magnetic coupled technique. The magnetic method would use two coils to establish a
communications link. One coil would form the transmitter and one coil would be used as the
receiver. The transmitter would be positioned on the outside of the structure and the
receive coil on the inside. The two coils would form a primary and secondary of a
transformer.
TOP
|
|
|
Design Ideas |
|
|
Communications Link Range?
The system I imagine would not have to have a range of more than about four feet. Such
a range should be enough to penetrate most structures. In cases where the signal only has
to penetrate a conventional 2 x 4 wood stud wall, a 12 inch range may work fine. The
maximum range will occur when the transmitter coil and the receiver coil are in parallel
with each other. But, a system could take advantage of a potentially longer range by
allowing for some coil misalignment.
The magnetic field strength of two loosely coupled coils drops off rapidly as the
distance between the transmitter and the receive coil is increased. The curve tracks a
"cube" function. So, if the distance between the two coils doubles, the signal
picked up by the receive coil will be only be one eighth as much.
TOP
|
|
|
Ring Frequency?
What frequency should I use? Perhaps I could use the US license-free band that exists
between 160KHz and 190KHz. Maybe shoot for a 175KHz center frequency. 125KHz is also an
international standard for radio frequency identification devices (RFID) so that too might
be a candidate. I think I will plan for a 175KHz frequency.
TOP
|
|
|
Transmitter Network?
The best way to produce a powerful magnetic signal is to drive a series resonant LC
circuit with one or more low impedance pulses. The pulses will induce a "ring"
frequency in the LC network. The highest magnetic field will occur when the coil current
can be maximized. One way to keep coil current high, is to keep the Q of the LC network
high. The practical limit of the Q will depend on the source impedance of the circuit
driving the LC network. Maybe a Q of about 10 could be the initial goal. With a Q of 10,
each pulse will induce a ring signal that will decay to near zero in about 10 cycles.
The transmitter coil needs to be as large as practical, but not too large. Perhaps a
coil diameter of about 6 inches would be a good place to start. Maybe I will start out
with a 6 inch coil for both the transmitter and the receiver. If that works out, I could
then consider reducing the size of one or both of the coils.
TOP
|
|
|
First Experimental Unit Power Source?
The power source of the transmitter should be a battery. Perhaps the first experiment
should use an inexpensive 9 volt battery as the power source. If I get sufficient
transmitter power, I could later reduce the voltage to 6 volts or even 3 volts. A single 3
volt lithium battery might provide the best long life power source if I can keep the
average battery current very low.
TOP
|
|
|
Network Driver?
OK, if I use a 175KHz ring frequency, the single drive pulse launched into the LC
network will need to be one half the period of 175KHz or 2.86uS. To produce the strongest
magnetic field, I need the highest practical coil current and the most number of turns
possible. If I use a 9 volt battery and a good pair of small power FETs, then maybe the
source current could be about 5 ohms. With a 9 volt battery, that would put the peak coil
current at about two amps. If I want a Q of 10, the reactance of the coil at 175KHz will
need to be about 10 x 5 or 50 ohms. Working backwards, if the reactance of the coil is to
be 50 ohms at 175KHz, then the inductance of the coil will then need to be about 45uH. To
tune the LC network to 175KHz, the series capacitance will then need to be about 0.018uF.
To insure I get the best peak current, the capacitor will need a low equivalent series
resistance (ESR) rating. A polycarbonate or a mica capacitor should be OK. With a peak
current of two amps, the peak to peak voltage across the capacitor would be about 2 x 2A x
50 ohms or about 200 volts. So, the capacitor will need to have a 250 volt rating.
TOP
|
|
|
Transmitter Coil Design?
I looked up the formula for calculating the coil inductance of a single layer coil.
Based on the equation, I will need about 11 turns of magnet wire for a 6 inch diameter
coil to get an inductance of 45uH. So, for my first experiment, I should plan for a 6 inch
coil with 11 turns. If necessary I could add or subtract one or two turns to get the
circuit to ring at a frequency of 175KHz.
TOP
|
|
|
Data Encoding Scheme?
To keep the transmitter power consumption low, I need a data encoding scheme that uses
the minimum number or magnetic field pulses. One such method is a pulse position scheme.
The method needs only N +1 pulses per data channel. To send one channel of data I need two
pulses. To send two channels, I need to send three pulses. The scheme works by using the
time between the first or previous pulse and the next pulse, to carry the information. To
convey the analog information, I will need a finite number of time windows. Each window
will need to last longer than the time for each pulse. Perhaps each window will need to be
at least three times longer than each LC circuit ring cycle. If I want an accuracy of one
part in 100 (1%) then I will need to provide up to 100 time windows. I also should include
some minimum time (zero level) between the reference pulse and the data pulse. Perhaps a
place to start is to plan for a minimum time of one half the maximum time.
If I set the ring frequency at 175KHz, then each ring cycle will last about 6uS. If I
want each time window to be three cycles then each time window would need to be about 20uS
long. If I want a total of 150 windows, then the maximum time between the first reference
pulse and the data pulse would be 150 X 20uS or 3mS. Using such a timing scheme it looks
like I could send up to 300 data channels per second with an accuracy of 1%. I could also
increase the accuracy to 0.1% if I allow for 1500 time windows per data pulse. Then, the
maximum data rate per second, or the maximum number of data channels send per second would
be 30. So, it looks like I would have plenty of options.
|
|
|
Pulse current based on desired battery current A fresh 9 volt alkaline battery will
give me about 0.5 amp-hrs of energy. If I want the battery to last a year, then the
average current will need to be less than 50uA. I got that figure by dividing 0.5 amps by
about 10,000, which is about the number of hours in a year.
OK, if 50uA average battery current is my limit, let's see how many pulses I can send
per second. That should define the data rate or the number of data channels sent per
second. If each pulse lasts about 3uS and if I send two of them each second, then the peak
current per pulse will need to be less than 8.3 amps to insure the average current is less
than 50uA. Since I already had planned on a peak current of less than 2 amps, such a
figure is reassuring. It means that I can either send about four more data channels or
increase the data rate by a factor of four and still keep the average battery current low.
If necessary, I also could launch more pulses per data burst. A few pulses for each burst
will make the ring signal last longer, which will help the receiver circuit. Also, for
many sensor applications, a data update rate of once every few seconds may be fine as
well. But, for my first experiment, I think I will launch a steam of pulses at perhaps a
rate of 50 per second. That will make it easier to see the signal on an oscilloscope and
to take some actual measurements.
TOP
|
|
|
Experiments Planned
The first order of business is to design and build a transmitter. Then, I need to
design and build a matching receiver.
For starters, I could use a simple parallel resonant LC network as the receiver
circuit. It would be nice if I could get a receive signal in excess of several volts peak
to peak at a range of 4 feet. Such a signal would minimize the needed amplifier gain and
would allow a receiver circuit to operate in a non-linear mode.
If I can generate signals for several volts peak to peak each time the transmitter
sends its pulse, I might be able to use a Schmitt trigger logic gate as my front end
signal processing circuit. Such a method would keep the power consumption very low.
The key in the transmitter is the driver circuit.
I think I will try a push-pull type driver that I have used before for driving low
impedance loads. The circuit consists of one N-ch FET and one P-ch FET. Both devices are
biased in the off state. The two transistors are linked to a signal source that generates
a single pulse. During the negative swing of the pulse the P-ch devices applies +9v to the
load. When the pulse swings positive, the P-ch devices is turned off and the N-ch device
is turned on. The transition time, when one device is turning off and the second is turn
on if very fast. So, the circuit is very efficient at delivering current pulses to a low
impedance load, such as a series resonant LC circuit.
Once the transmitter and the receiver circuits are built.
I then need to conduct some tests. The first tests will be setup in open air to
make sure the basic scheme works. Some basic measurements at 48 inches will give me an
idea the signal levels to expect. Then, I should repeat the tests through a wood wall and
through a concrete wall. I would expect to see some attenuation of the signal through
cocrete walls. Especially if the walls contained steel reenforcing rods.
TOP
|
|
|
|
|
|
