Exact match. Not showing close matches.
'[EE] Addressable water detection sensors'
Long time lurker, first time poster. Maybe I'll send a brief
introduction at the bottom the message... For now:
I have an "aquarium room" in my house. One of my many hobbies is
developing devices to aid in the automation of the operation and
maintenance of these aquariums. Yes, I'm aware of the various
commercially available products for that purpose. Part of my fun if
making my own. People build boats and airplanes too, even though you
can buy them commercially.
In addition to such things like heating and cooling control, lighting
control, automatic water changes, etc., I'd like to monitor for leaks.
There are a number of points in the room I'd like to monitor. Several
places around the various tanks, under or around this filter or that,
under the water return circulation pump, at the top of the sump basket,
and so on. It's saltwater, by the way. In addition to simply sounding
an alarm, I might like to start a pump, stop a pump, send a message to
my blackberry, etc.
My first thought is to take some two conductor speaker cord or similar,
run it around to the various detection spots, and solder some two pin
header/terminal gold plated strip sections. Hot glue to seal the solder
joints and secure the "sensors" in place and away we go. A long, series
of parallel contact points, if saltwater gets across any of those sets
of two pin contacts, simply detect that, and take one or more of the
Wait... "one or more"? How do I know which action to take? Well, it
depends what's leaking, and how badly. If one sensor near a plumbing
joint intermittently shows a leak, that's one level of response. If
every sensor on the floor suddenly shows continuity, well, that's
something else. It seems that it would be useful to know exactly which
one(s) are actually wet.
Let's say there are 16 such sensors. Running a bundle of 16 pairs of
wires around seems... awkward. Certainly they could share a common
ground, so that's down to 17 wires at the controller, down to two at the
last sensor... 16 comparators into a multiplexer, and have the
controller scan them every pass through the event loop. Something like
that. Is that the best we can do? Maybe an analog multiplexer and one
I'm thinking perhaps something like a one-wire sensor (is there anything
appropriate?), bussed along the original idea's two conductor zip cord,
also polled for continuity, assuming the sensors are very simple
circuits, at least more simple than 16 comparators and a multiplexor...
Cost is an issue, a $5,000 solution will never happen, but it's a
one-off project, so I'm not trying to production quantity costs low.
If the addressable solution gets too crazy, and perhaps in the interim,
I'll just go with the quick and cheap parallel contact points solution,
and forget choosing an appropriate response.
By way of introduction: I'm a software developer. Java / J2EE web
applications for the last several years. Before that, a number of years
of n-tier C application design and development, mostly on Unix,
sometimes on PCs. That includes about a dozen years doing factory floor
automation, including a fair amount of embedded controller work, mostly
Z80 and 8051 based. Started initially as an electronics technician.
Also done a bit of EE design work. Long ago, designed and built a
number of projects, including a home weather station, including an
serial interface to a PC (or whatever) using the then very new MAX232
chip, LCD display of conditions, etc. Great fun. I think I enjoy
solving and implementing the solutions more than just having the
problems themselves solved. A large part of why I don't just buy a
solution to the aquarium controller. Well, that, and MY solutions do
precisely what I want, how I want it done, not settling for something
that comes close.
OK, that's a LONG email. Hope I didn't lose you all!
If you look at these
or at i2c offererings
You can find either multi-drop addressable digital I/O or ADCs.
You can set up a poll-able network using 4 to 6 wires. Each station
will be relatively cheep.
Check out http://www.i2cchip.com/ They are very easy to deal with.
Using an ADC you can measure resistance between the gold contacts.
Even easy is to put an alka-seltzer between two spring loaded
contacts. When the AS gets wet, fizz fizz and contact is made.
|Instead of developing a whole system from scratch, check out PLC's. They
program in ladder logic, and can have high level programming embedded in
the ladder. Look at http://www.tri-plc.com/home.htm. You can download
the Trilogi software free from the site. I use a
http://www.tri-plc.com/t100md888.htm to control a standby generator.
They have numerous inputs and outputs of various types and a keypad/LCD
display available. Mine sends an E-mail to my cell phone if the house
sump level gets high. But that's connected to a GEMS float reed switch.
For the 16 pairs running around, could use one or more smaller PLC's to
handle local I/O, and connect them RS-485. The software and hardware is
built in. but the 16 pairs is only (2) Cat-5 cables and that cheap. My
generator runs various sensors on Cat-5 some 40 feet distant no problem.
I think I would go with one CPU, and run the wire.
I have used the National LM1830 water detector chip. It uses the 2 wire
sensor like you indicated, and TTL out. That chip has been around a long
time, there probably is a more modern replacement, or could build a
circuit from scratch.
I'm just a hobbiest, probably some of the more experienced will have
some great ideas, but it sounds like keep the hardware simple is a priority.
Ron Burns wrote:
2009/5/21 Carl Denk <windstream.net>:cdenk
The smaller PICs are cheap enough to be considered as sensors so you
just need to work out how to power and run them on a 1-Wire bus. Not
too hard I think unless you want high speed.
As an off the wall approach - how about a TDR system that detects the
change in line impeance of a twisted pair? Sensorless. You should be
able to get down to less than a metre of resolution which is probably
enough. The TDR side of things would be a good challenge although
there are web examples of TDRs and homebuilt radar units that would
Alan B. Pearce
>Let's say there are 16 such sensors. Running a bundle of 16 pairs
>of wires around seems... awkward. Certainly they could share a common
>ground, so that's down to 17 wires at the controller, down to two at
>the last sensor... 16 comparators into a multiplexer, and have the
>controller scan them every pass through the event loop. Something
>like that. Is that the best we can do? Maybe an analog multiplexer
>and one comparator?
Right, for sensors, especially drip sensors, my initial reaction is to make
a PCB with fine tracks and fine spacing, without solder resist. With the
RoHS fanaticism that is currently around, one of the finishes you can get is
a gold flash, which would seem to me, to be a suitable method to protect the
copper. Then a small micro can sense moisture bridging the tracks by the
conductivity of a drop of water, and with salt water this shouldn't be a
A similar sensor could well be used as a flood sensor.
However another technology that could be worth looking at is capacitive
sensing. I don't know how well this would do for drip sensors, but would
work well for flood sensing.
As to how to have a heap of sensors spread around the place, I would look at
using a 10F200 series PIC at each sensor, as the sense driving element, and
communications to whatever host you use. It would be possible to use a
1-wire style protocol with these, but it could equally be a software
transmit only UART driving an open collector transistor to allow multiple
units on a single line (see the current thread under [EE] on RS485 comms
protocol). I would not worry about a two way protocol (although it wouldn't
be too hard to implement), but rather just check the line to see if a
message is being sent, and if idle send a message. The message would include
a sensor ID, so the host knows where the message comes from.
A 10F200 should be able to drive and sense a multifinger PCB as described
above, and Microchip also have an application note on using one for
capacitive touch sensing. The latter should be able to be converted to any
form of capacitive sensing.
As an aside, a suitable capacitive sensor could well be used as an immersion
sensor to check the level of water in a tank. This has been discussed on
this list before.
> With the RoHS fanaticism that is currently around, one of the finishes
> you can get is a gold flash
Gold flash would be better than lead anyway, as solder oxidises and
prevents good conduction. For a couple of special hand-made metal
components needed for projects I asked nicely and a jeweller had his
plater add them to a batch. Did a good job for no cost, even some
large ball bearings
2009/5/21 Jinx <clear.net.nz>:joecolquitt
>> With the RoHS fanaticism that is currently around, one of the finishes
>> you can get is a gold flash
> Gold flash would be better than lead anyway, as solder oxidises and
> prevents good conduction. For a couple of special hand-made metal
> components needed for projects I asked nicely and a jeweller had his
> plater add them to a batch. Did a good job for no cost, even some
> large ball bearings
John La Rooy
|On Thu, May 21, 2009 at 6:18 PM, Alan B. Pearce <stfc.ac.uk>wrote:Alan.B.Pearce
> >Let's say there are 16 such sensors. Running a bundle of 16 pairs
> >of wires around seems... awkward. Certainly they could share a common
> >ground, so that's down to 17 wires at the controller, down to two at
> >the last sensor... 16 comparators into a multiplexer, and have the
> >controller scan them every pass through the event loop. Something
> >like that. Is that the best we can do? Maybe an analog multiplexer
> >and one comparator?
> Right, for sensors, especially drip sensors, my initial reaction is to make
> a PCB with fine tracks and fine spacing, without solder resist. With the
> RoHS fanaticism that is currently around, one of the finishes you can get
> a gold flash, which would seem to me, to be a suitable method to protect
> copper. Then a small micro can sense moisture bridging the tracks by the
> conductivity of a drop of water, and with salt water this shouldn't be a
> Could you make a bunch of those by cutting up the edge connector from an
old isa/pci card?
> Why would you want golden ball bearings - some sort of religous
My devotees have to get their own accessories. At some slight mark
One project was a free-floating joystick. A ball bearing moved in
a cross-shaped enclosure. The bottom of this enclosure was a cross-
shaped piece of copper. This was the common terminal. At the end
of each arm was a copper terminal. The ball bearing shorted the
common to a terminal, depending on which way the joystick was
tilted. The other project was a ball bearing clock. Minute balls were
powder coated black, the hour balls were gold
Alan B. Pearce
>> Right, for sensors, especially drip sensors, my initial reaction
>> is to make a PCB with fine tracks and fine spacing,
> Could you make a bunch of those by cutting up the edge connector
>from an old isa/pci card?
That would certainly be a good experimental starting point, but I was
thinking in terms of a 10-12 thou finger, 10-12 thou space. But the spacing
on an old ISA (or even better an EISA) card may well work out OK.
But the other trick with doing ones own card is all the electronics required
can be mounted on it as well.
At least what used to be an issues, was DC currents plating out of
materials, the LM1830 and other water sensing designs I have seen lately
use AC currents to minimize the plating action.
Alan B. Pearce wrote:
On Thu, May 21, 2009 at 3:22 PM, Carl Denk <windstream.net> wrote: cdenk
> At least what used to be an issues, was DC currents plating out of
> materials, the LM1830 and other water sensing designs I have seen lately
> use AC currents to minimize the plating action.
This is exactly the problem with gypsum blocks, used to measure
moisture content of soil in gardening. Solution is to use AC currents,
but it should not be difficult to build a circuit around PIC so very
short measuring pulse could be used.
Measure the time needed to charge some capacitor, and give opposite
polarity pulse after that to remove any remaining charge.
Alan B. Pearce
>At least what used to be an issues, was DC currents plating out
>of materials, the LM1830 and other water sensing designs I have
>seen lately use AC currents to minimize the plating action.
This is part of the reason I saw for using a 10F200 series device at each
sensor - the AC drive and detection, along with the communication.
I haven't looked at the National part, so don't know if it may also fulfill
all the requirements.
That's what I was saying there are other ways, but the LM1830 is cheap,
just a few components (small caps and resistors) needed, schematic is
easy, no design necessary. Can drive an LED or sonalert (sp) alarm with
The LM1830 does use an AC current, which back then (say 25 years ago)
was a big deal.
Alan B. Pearce wrote:
More... (looser matching)
- Last day of these posts
- In 2009
, 2010 only
- New search...