Natural Science Forum / Physics / Acoustics / August 2007

I think there is a large market for low power acoustic transducers
used to set up communication between controller and valves in an
irrigation system.

Currently this problem is usually solved by running two wires from the
controller to each valve.
A wire network needs to be laid along with the pipelines.  The wires
are fragile, and becuase they are buried, a break caused by a wayward
trowel or industrious mole necessitates re-laying the entire wire.

Another system that uses coded signals on a single pair of wires is
coming into use.  This makes it much easier to add a new valve to the
system, but has the same problem with fragile wires.

There are wireless controllers, but they tend to be very short range.

It occurs to me that using sound you could communicate to the valves
quite handily.

The data rate is very low.  Valves are typically on for 15 minutes 1-4
times per day, and off for the rest of the day. If a controller issued
only 1 set of valve commands per minute, it could control hundreds of
valves. Typically a control sequence would work as follows:

Controller:  Valve 27 report status.
Valve 27:  Closed
Controller:  Valve 27 open
Valve 27:   Opening
Controller:  Valve 27 report status
Valve 27:   Open.

3 byte packets would be sufficient:  10 bits for a valve ID, 6 bits
for a command/status
set. The rest for correction codes.

Because valves have a tiny battery and a latching solenoid they don't
have a lot of power for transmission. Typically they run off a
standard 9v battery for a season. So I think it best for the system to
be polled from the controller.

Current valves have no smarts.  Most of the time you are running a
solenoid with power from the controller.  Typically 24v at 0.5A.
Using a latching solenoid and a small circuit board, you can walk
around with a programmer and give the valve a schedule.

So far I've not found a system where the valve can report its own
status.

Multiple controllers could run on the same network, using different
frequencies.

It may be necessary to come up with a different form of pipe branching
fittings.  I can see that a normal T connection reflect a lot of
energy back.

So:  My question for the physics community:  Is this even remotely
workable?  I can see the follwoing problems:

1.  Echoes off of corners and connections.  What if the packet length
is small compared to length of pipe to the next connection.  E.g.
sound travels typically at 5000 ft/s in water.  If you constrain a
packet to 1 ms, it's only 5 feet long.  If the 5 feet downstream from
the transducer is clear, no echo can reach the controller bef ore the
last of the packet has arrived.  This would require a data rate of 24
bits per ms or 24,000 bits per second.

2.  Transmission losses.  In my irrigation system at home, it's about
2000 feet from well head to sprinkler tip.  In that run it goes from
1.5" pipe to 1" pipe to 5/8" pipe.  Choice of frequency is important.
Will a frequency that is long compared to pipe diameter propagate well
down the pipe?

3.  Need for repeaters.  Certain things, such as a water filter, may
interfere too greatly with the signal.  May need to make a repeater
that hears a signal, waits for the plumbing to stop ringing, and
repeats it.

4.  Cheapness of reliable transducers + associated circuitry. For this
to be workable, the additional cost at the valve has to be small
compared to the cost of laying wires.