View Full Version : Building a sensor to locate a buried wire fence

07-11-2006, 08:03 PM
I am doing the old 'Build a robot to cut the grass' bit and while moving the platform abound it not a problem (no cutting motor yet) I have been perplexed about all the different suggestions of how to keep it on the grass. I have read about colour recognition (grass = green), GPS, Laser fencing etc. but I seem to have settled on the idea of a buried wire fence. having done a bit of research I have found someone who has done a similar thing but have no joy in contacting them. Below is a link to their website where they have circuit diagrams of the i2c receiver and a transmitter which sends a FR signal round a perimeter cable.
Receiver http://www.betuwe.net/~mowbot/UDO5recvRcv.PDF (http://www.betuwe.net/~mowbot/UDO5recvRcv.PDF)
Transmitter http://www.betuwe.net/~mowbot/UDO5transSch.PDF (http://www.betuwe.net/~mowbot/UDO5transSch.PDF)
I like the idea of this unit being i2c as it makes the interface a lot easier but my concern is if I get PCBs made for these using the designs I have been quoted GBP 50 per design (for 4 & 5) little boards respectively. This seems a lot to lay out on an experiment.
My other problem is that they use a tuned coil to pick up the signal, can anyone advise how I can make a coil which will pick up the correct signal.

07-11-2006, 08:54 PM
Google "Cable locators" (or something similar)....

The cable/phone guy uses such things to locate buried cables....there's also smaller handheld devices to locate wires in a bundle.
Essentially it puts a 1kHz Warble tone on the line....you take an inductive pickup sensor and sweep across the suspected area and a little speaker will start playing your tone to you. I'm sure there are numerous sites with a circuit example of an inductive pickup....and as far as building one, it's just a matter of winding a coil and attaching the appropriate conponents for the tone frequency you're after (you might want to use higher freq's to keep animals away...dont' know if that really works!).

You may even be able to use a Hall effect sensor....be sure it's one that you have access to it's full output...not one that latches when it sees something above a threshold.



"Inside each and every one of us is our one, true authentic swing. Something we was born with. Something that's ours and ours alone. Something that can't be learned... something that's got to be remembered."

07-11-2006, 10:00 PM
Thanks for he reply.

I was initially looking at a Hall Effect Sensor but I got confused by the amplification system required so that the signal is large enough to be meaningful to the processor. I had looked at page 3 of the following http://www.robotshop.se/micro/pdf/rob_sch.pdf (http://www.robotshop.se/micro/pdf/rob_sch.pdf)·and was ok with the principal but I could not find the info on the capacitors required or the coil construction. I can bolt stuff together but have minimal understanding of how the components interact to do the job.· I then found the RF solution which looked good but still had a few questions.
If ay one can help me understand the hall sensor circuit I would appreciate it.

Loopy Byteloose
07-11-2006, 10:33 PM

http://www1.int.conrad.com/scripts/wgate/zcop_in/~flN0YXRlPTI3NjgxNTc5NzU=?~template=PCAT_AREA_S_BR OWSE&glb_user_js=Y&shop=A_B2C_IN&p_init_ipc=X&~cookies=1

Try the above links for the Conrad DCF77 coil.· That appears to be the 'fine print' on your drawing.

"If you want more fiber, eat the package.· Not enough?· Eat the manual."········

···················· Tropical regards,····· G. Herzog [·黃鶴 ]·in Taiwan

Post Edited (Kramer) : 7/11/2006 3:46:45 PM GMT

07-12-2006, 12:24 AM
Hi Billericay;

The "electric fence" concept is used by the folks who make the shock collar your dog wears to keep it in your yard. It works on the principle of an (approximate) 10 KHz current pulse fed through the fence wire, and the electronic pickup in the collar detects this as the dog nears, about 5 feet. The reliability and distance repeatability for your application may not be good enough. The amount of "pickup", hence sensitivity, varies gratly with orientation of the sensor.

I think your application is possible, but it will need much refinement and sensing algorithms. You may be frustrated before you get it to work reliably. Clearly you don't want a powered and running lawnmower to escape from your fenced area.

To do this reliably is not an easy task. Probably the best way (most reliable and repeatable) to tackle it is with a range finding laser, locating and tiggering beacons that emit an RF signal when they are located. Pretty costly, lots of software, but will give you better than 1 inch precision with fast acquisition. My experience with them has yielded one tenth of an inch in 30 feet using slow acquisition (2 seconds) which averages multiple readings.


Peter (pjv)

07-12-2006, 04:38 PM
Thanks for all your input. Some additional information for your peace of mind regarding the design and operation of my ‘mower’.
* Mercury tilt switch fitted on main power line from battery to ‘Kill the Beast’ if it is picked up.
* Keypad to input run code to stop unauthorized use.
* Drive motors are run by the Devantech MD23 which require a move command every 2 seconds otherer wise it will stop. Therefore is electronics fail the maximum time it can continue to run is 2 seconds.
* Using the RF loop for boundary marking also leaves a RF signal over the garden while the transmitter is powered. So by having a third receiver coil continually checking for a signal will allow the failure of the transmitter to be detected and therefore initiate the ‘STOP’ routine.
* Bumpers on the front – bit like whiskers will allow it to bump and reverse away from obstacles.
I have not seriously considered the use of a speech synthesizer to give a warning but there are options to have an ultra-sonic surveillance zone around the mower to trigger an audible warning.
My interest stems from my Mothers robot mower, MOWBOT. http://www.mowbot.co.uk (http://www.mowbot.co.uk/) ·This does all the things I am trying to recreate but for less than the GBP 1,300 she paid for hers. The buried wire fence if the bit that I am in need of help with. I have looked at the site mentioned above and read their diary but they have not answered my request for help in understanding how their sensors work.

Loopy Byteloose
07-12-2006, 06:36 PM
The mercury switch should do a complete kill [not a reset].
The mower may capsize at sometime and arrive at a position that allows the switch to reconnect. The motor would be in a runaway condition and expose people and animals to harm.

Also, striking a hard object may produce a transient off, then off. If you hit a sprinkler head, a baseball, a dog bone, or a rock; you don't want the mower to continue to wander as there is debri in the lawn that may hurt someone. I have seen all these items launched 30 feet in various directions [with a rotary mower]. Optionally to can use a 'weed eater' type safety blade. It is a piece of nylon string and doesn't have the ability to push heavy objects.

The antenna I referred you to is the 'complete' version with mounts and housing. There is a $10euro version with just the ferrite rod and coil. Since is has 4 inputs, I suspect it is a bit sophisticated and you should just buy one rather than try to build.

"If you want more fiber, eat the package.· Not enough?· Eat the manual."········

···················· Tropical regards,····· G. Herzog [·黃鶴 ]·in Taiwan

07-12-2006, 07:04 PM
Thanks, when the mower is rolling with boundary sensors the program will be

If the sensors detect different things then they will either reverse, turn and go forward again or STOP.
It the power is cut then reconnected the routine will not get past the ACCESS CODE subroutine so the mower will remail inert.

i vas thinking of monitoring the blade speed and if it drops rapidly (ie. hits something) then stoping the cut. the balde will be low to the ground and there will be protective bumpers to pick up obsticles, or even to push them out of the way if they are light like a ball.

I ordered two DCF receiver boards last night and aim to have the rest of the sensor ready in my bread board when they arrive.

07-12-2006, 10:40 PM
Hi Billericay;

Well, as you seem determined and have some proper ideas about safety, I hope you are ready for a fair bit of development work. That is, if your intent is to have this thing operate autonimously, and have it cut grass to within an inch or so from the buried wire fence. I have done significant experimentation with these systems, and can tell you it's possible, but not easy.

Obviously, the tighter the distance control, the harder it is. I would say that a variance in detection repeatibility of something better than 2 feet would be not too difficult, whereas 2 inches would be next to impossible.

Your chances of success with burying two independent wire loops, and sensing an amplituse as well as phase difference will yield much more precise results. But then you also would need to bury them precisely, say separated by one foot with a precision of better than one inch.

For myself, I would find the results inadequate if the positioning could not be controlled to within one or two inches. Perhaps your need is such that you wish the autonimous part just to do the bulk of the work, and you'll clean up behind it yourself later. You did not specify what you are expecting.

Anyhow, good luck!


Peter (pjv)

07-13-2006, 06:19 AM
Hey Billericay_Boy check this forum out.


Post Edited (bennettdan) : 7/13/2006 1:01:14 PM GMT

07-13-2006, 06:52 PM
Thanks Dan for the link but it directs me back to this thread. ? ?

07-13-2006, 08:02 PM
Hey Billericay-Boy
Sorry about the link I changed it it connects to a thread in the robotics that I gave a curcuit that I got out of an old robotics book check it out. very

07-13-2006, 09:00 PM
Dan, i think i found it. it was a posting by Molle last september. sounds very promising. i have lots of ideas and lots of circuit diagrams to try. Now to buy all the bits and try to get one to work. I have posted on Molle's discussion to see how they got on with their mower.


07-14-2006, 08:02 PM
How do you plan to use hall-effect sensors if you do I think they have to be closer than what you can get

07-14-2006, 08:44 PM
I was looking at using a curciut i found to amplify the detected and this will then go through a chip and output via i2c.

The other idea i have been looking at yesterday is on the Lego site http://www.philohome.com/sensors/filoguide.htm worth a look.

Beau Schwabe (Parallax)
07-14-2006, 10:20 PM
Billericay-Boy (http://forums.parallax.com/member.php?u=46473),


The idea that you mention here looks like it would work ok, but there are a few things that I see about it that you might consider.

The loop oscillator using the 555 is ok, but I would probably want to use something a little more frequency stable if it were me.

On the·receiver end, I would make C1 and C4 adjustable.· What you are doing here is forming a radio tuner in conjunction
with the coils L1 and L2.· If you have component tolerances that are different, it will affect the balance between the two
sensors.· Also, as mentioned above, the stability of the frequency might/could also have an effect here.

The output of this circuit is not clear in the schematic.... Your measurements should be taken from the Emitter of Q1.

Careful to keep the leads from L1 and L2 as well as R1,R6,C1, and C4 close to the IC or you will·create an antenna.· This however
"might" be a desired effect for longer distances, but for close range sensitivity you will not want to do this.

Beau Schwabe (mailto:bschwabe@parallax.com)

IC Layout Engineer
Parallax, Inc.

07-15-2006, 03:20 AM
The sensor interface in Philo's design is made for an active sensor of the LEGO Mindstorms RCX. An active sensor recieves power from the RCX but also reads the voltage. It's a multiplexed I/O, with a timing scheme of: The sensor recieves power for ~3ms, then the return voltage is read for 0.1ms etc. You could recreate this with a Stamp if you wanted and then build the whole circuit for the wire guidance sensor.

You've got to play the game.
You can't win.
You can't break even, except on a very cold day.
It doesn't get that cold.
~Laws of Thermodynamics~

07-15-2006, 03:48 AM
Hi Billericay;

The circuitry suggested in the "Lego" example is conceptually what has been discussed. The major difference is, (and it is a HUGE difference) is that the Lego example is a wire follower rather than wire proximity. A wire follower can sense the differential signal in pick-ups, whereas in a proximity scheme with random orientation, this is not possible.

I imagine you want only to bury wires around "keep-out" areas, rather than the precice path for the mower to follow. This is immensely more complicated than a wire-following technique. For instance, the orientation of the coils to the buried wire, and the depth of the wire from one point to another point will change the received signal, hence a perceived change in distance from the wire.

As I stated before, if your application is to just do a rough cut, say within a variation of 1 foot or so of the wire, and then trim rest by hand, then this simple scheme will work.

If, however, you want the system to trim to an accuracy consistency of 1 or so inch, then you WILL require a more complex scheme. That can be done of course, but be prepared for a lot of work.

In fact, the unit I experimented with eventually had the coils direrectly fed to an SX; no amplifiers were used, just lots of tricky algorithms.


Peter (pjv)