smarter electric fences

I was thinking about starting to get animals this year (25 chickens, 3 lambs and 3 child pigs).  If I a supplier will sell me glass fabric, at least the chickens could be here.

To have pigs start to clean up woodlot that adjoins pasture, it looks like you probably have to start from the pasture and work your way in.

Reading about sheep has been the most interesting.  They are much more electrically insulated than cows and horses (I would expect pigs to be even less insulated for 2 reasons, they use their nose which is typically wet, and if they have hair it is probably less than cows or horses.

There are lots of articles on DIY electric fence.  I think they all devolve into using a coil from a car and/or using 555 timers.  There were interesting articles on monitoring electric fence.  Knowing that a particular conductor had continuity is something people were interested in.  Quality of the pulse was on some people's minds.  I think one person would like to record pulses (and maybe save the ones that involved shocking something).

Most of these fencers seem to send pulses at about 1 Hz.  Which in a day is 86400 pulses.  How many involve shocking something?  10?  20?

I gather pulses are about 5J.  A Joule is a 1 amp current at 1V difference in potential for 1 second.  So, 5J could be 0.5a for at 10V for 1 second.  Or 0.5a at 100V for 0.1s.  Or 0.5a at 1000V for 0.01s  People talk about 6kV (or more) and pulses are typically measured in mS (0.001 seconds), so this sort of gives you an idea where these numbers come from.

A thread in the adafruit forums, got down to things like tasers.  What can we learn before we send the big shock.  It might be, that a person needs to have "smart clips" holding the wire to the posts.  In this model, you don't have to restrict yourself to one energizer per fence, you could put up smaller energizers and have multiple units per fence.  Maybe the smart clip has a supercapacitor for energy storage.    And well guess that it can power itself for say 2 minutes before running out of power.  We still send pulses from the energizers at about 1 Hz, but they  might only be 20V or something.  Just enough to useful at halfway to the next energizer.

We might need to use special peak shapes, and perhaps we can analyze the pulse to extract how much loss we are getting from weeds contacting the fence.  At some point, send a message to the grass mower, to mow the grass.  But, we want to identify what side of the fence the contact is coming from (outside animal trying to come in, inside animal trying to get out), is it a cow, horse, pig, sheep, or something else?  Is it something peeing on the line?  And we want to identify where the contact is.

When a contact is made, the two energizers which bracket the contact are determined.  If a 6J pulse is called for, we send a 3J pulse from both energizers with delays chosen such that the two pulses arrive at the contact at the same time; delivering (nearly) 6J of pulse.  Or rather, if one energizer has to send past 3 smart clips and the other has to send past 5 smart clips, we send 3J + 3 times delta from the first and 3J + 4 times delta from the other,; knowing that all smart clips will remove delta J from any pulse for powering themselves.  Subject to not becoming over charged.  The smart clips try to identify the contact (so may send messages of cow, horse, sheep, pee) to the energizers.  So the energizers send an appropriate charge based on that information.  For smart clips outside of the contact bracket, they subtract off enough energy to fill their needs.

I think this interesting thread on adafruit was 2012.  If someone was so inclined, such a product could be on the market already.

Part of the reason I am interested, is that in the process of trying to grow a windbreak, I am going to have black locust, honey locust and Osage-orange nurse trees, that will get coppiced or pollarded during their usefulness as nurse trees to produce fence posts (BL and OO) and outdoor wood (HL).  And both BL and OO become very hard when dry; and possibly impossible to pound a nail r staple into.  Screwing in a smart clip might justify drilling and tapping holes for that.

That would be a really interesting electronics project. But I wouldn't want to trust a number of smart devices equal to the number of fence posts. Its just too many things that might stop working at any time.
A standard electric fence system just has to worry about the wire breaking, the insulators cracking and falling off in freeze/thaw cycles, and plants growing up and grounding out the wire. Most chargers will at least have a light that lets you know if the charger is still working, some tell you the voltage so you know if there is a problem to go find.

Gordon, that sounds like an awful lot of stuff to fail! I don't want any electronics exposed to the weather here, and given how much fence is needed, I doubt a 5 "smart clips" on every post would be affordable.

I have no idea if things could be reduced to a screwhead or not.  What I would want to avoid for myself, is anything produced by MBAs.  

I think you want the wire in contact with something like a scroll button on a mouse.  So that it can measure how much the wire scrolls positive or negative at a post.

If a person can staple into a post (such as treated SPF), I think a box might make the most sense.  If the wood is hard like Osage-orange or black locust; I think drilling and possibly even tapping makes the most sense.  Something that resembles a screw (with a big head or possibly a really fat washer under a screw head) might work better.

Having these boxes charge from the pulses emitted by the energizer (the scary ones and the smaller test ones) avoids a lot of problems.

On my farm, there are big chunks of the farm which do not get any direct sunlight on the winter solstice.  Having solar powered energizers won't work there.  The NE corner of my farm gets the most direct sunlight on the winter solstice, at about 4.5 hours.

I agree that just to replace existing performance this idea won't pay.  This idea does make a bunch of other things possible.
1. Variable sized pulses, chosen by need (sheep need more than cows).
2. Record every pulse delivered to the "animals".
3. Feedback telling the farmer how much parasitic loss to weeds (and other) is happening.
4. New data.  I suspect the fence line can tell us what the wind speed is.  I've no idea of the accuracy/precision of such a thing.  Can it give us information on precipitation?

For me, the big one is point 2.  Point 1 could also be of interest to many people.

Sebastian Köln wrote:Gordon, that sounds like an awful lot of stuff to fail! I don't want any electronics exposed to the weather here, and given how much fence is needed, I doubt a 5 "smart clips" on every post would be affordable.

If you have to pay someone (I am picking the labour rate from local car dealerships here) $160 per hour to build things, then none of this could be affordable.  And knowing the work some of these "technicians" perform (I would rather have a mechanic fix my car, than a technician), it won't work long.

Have you ever looked at an Arduino board?  Or a Raspberry Pi computer?  I think the original price point for Raspberry Pi (model 1) was $35 USD.  About the size of a credit card.

Two of my desktop/servers that I have here, have 750W power supplies.  I just bought a Pinebook Pro laptop which has a Rockchip 3399 CPU in it.  I also have a Raspberry Pi clone here, which has the same Rockchip 3399 CPU in it.  It has a 15W power supply (wall wart).  I first tried to install a 1 TB NVMe SSD in it which drew 6.8W of power, and the machine wouldn't boot.  Too much power.  I had to go buy a different SSD which only draws 4W of power; now it boots.

I am going to use that Rockchip 3399 RPi clone, as part of a GPS basestation on my farm so that I can do realtime kinematic corrections and get precisions down to about 1cm.  Not bad for something with the cross section of a credit card (it's about 2-3 inches thick, but most of that is air).  I think the local John Deere dealership sells add-ons to equipment (and then yearly subscriptions) for thousands (10s of thousands?  More?) of dollars.  Once I get my GPS base station up and running (probably about $400 CDN for parts), I may let local farmers use my data for free, or some minimal price.  If commercial operations look to access the data, it will be closer to John Deere's prices.

Lots of electronics gets "potted".  Once the circuit (which could include a printed circuit board) is finished, it is put in a mold and liquid epoxy resin is poured in to encapsulate all of the electronics.  This works for low power electronics (because epoxy is lousy at conducting heat).  But the epoxy encapsulation pretty much keeps all the weather away from the electronics.

We can make epoxy conduct heat better.  The two best heat conductors are diamond and graphite/graphene/buckminsterfullerene/carbon nanotubes.  Diamond doesn't conduct electricity, whereas all the other carbon based things in the second category do.  So, some kind of diamond addition to epoxy could probably let epoxy potting work for higher powered circuits to (if we get the price of diamond low enough).

There has been a zillion service stations and similar, who have buried steel tanks for gasoline, diesel, ....  And lots of those installs (especially the early ones) resulted in huge leaks of fuel into the ground and huge potential bills for someone to redevelop the land after the service station went away.  A local company in that business a few years ago dug up a glass/epoxy tank which had been buried for 50 (?) years.  It looked like new.  Glass/epoxy is still more expensive on the initial install; but lifetime costs is like comparing someone who gets migraines most days with a person who has never had a headache.

This isn't how electric fencing works.

The way that electric fencing works is that the fence is on ALL THE TIME so that when a young animal touches the fence, it gets a shock.  Depending on how smart the animal is, it learns not to touch the fence.

After that, the animal won't touch anything that looks like the fence.  It doesn't matter if it has a charge or not.  You can unplug it if you want.

If the animal accidentally touches the fence too many times and doesn't get a shock, depending on how smart it is, it will learn that the fence doesn't work any more.

If you want to make a "smart" fence, develop a herd that learns quickly not to touch the fence, and is risk-averse to testing the fence later on.

You're treating the animal like a cog in a system, and trying to engineer the system.  The animal has a brain, and can easily game whatever system you devise. You need to work on animal psychology, not electronics.

What is your fence going to do that's better than a normal fence?

to your points

1 why? this isn't necessary sure a sheep may need a higher shock so just use a higher shock it's not going to kill the cow.
2 why? I don't care how many times my animals touch the fence, and as others have said it's normally very few times. (we did have a horse once that LIKED the fence..)
3 this is already built in to many chargers and can be checked by a, grabbing the wire, 2 listening for the tic, and with various meters on sticks.
4 why rebuild a weather station?

the price of  connector for an electric fence is tiny, each of my posts has to have a minimum of 3 normally 4 there is a post every 4-5 meters, well over 200 posts. and that's only for 4 acres!

What is the need that you are thinking to fill? If it's your own enjoyment fine but I really don't think anyone is going to buy a "smart" fence.

I think this is going to be far harder than you think to accomplish.

1.  how are you going to protect your low voltage circuitry from the high voltage? MOV, Sidactors etc all have limited life expectancies and you will need at least 2 layers of them to provide enough clamping.  The other answer is to put a high frequency filter in and use it for clamping.  Problem here is cost.  You will need a fairly hefty inductor and high voltage capacitor.

2.  how are you going to sense animal type, direction and identify weeds?  Realize they may be wet or dry, dirty or clean meaning their electrical signal will always be changing as well as the environment around them.  Also realize that you are talking slow microprocessors so simply measuring the arrival time of 2 electrical signals is tough too.  Modern systems involving timing mostly use some sort of signal interferance system to measure the difference between signals  But if you are going to detect side of the wire you will need detection stations out a distance perpendicular to the wire because measuring along the wire unless your stations are extremely close together is going to be difficult.  

3. Likely you will find you need to run a ground wire too around to make your system work.  So be aware of induced voltage and its affects on fencer range.

If you are doing this for the fun of it more power to you.  If you are doing this for commercial reason how does any of this information make the system pay?