Lightning Protection

[RAL]

Using a relay like that is an interesting option, since it keeps the two sides more widely separated.   However, it doesn't ground the antenna side (or in your case the sprinkler side) at all.  I'm not enough of an expert on lightning to know for sure whether leaving it ungrounded is better or worse.  If it arcs, it should arc to the grounded contact of the relay, but I would think that providing a solid ground connection in the first place would be better.
 
With antennas, you should always ground the coax shield, but I have seen differing opinions about what to do with the center conductor.  The energy of the lightning strike needs to go somewhere, and if you don't provide a path, it will find a way of its own.  Some people recommend grounding the center conductor of an antenna coax, others say just route the wire back outside the house and leave the center conductor connected to nothing.  Leaving it connected to nothing doesn't seem like a good solution to me.

 

[NeverDie]

I would look for a relay that has the widest contact gap you can find.  There are some relays meant for use with solar panels and/or power systems that have gaps of 2.5mm or more.  I recall seeing some that were on the order of 6mm.
 
A 6mm gap will withstand a surge of 5kV or so.   

How big a gap would be required to get up to 100kV?  Somewhere (can't find it now) I saw a chart about lightning strikes, and I vaguely recollect that 95% of the time the voltage surge entering a house (by whatever means) was 80Kv or less, and only around 1% of the time was it 200kV or higher.  IIRC, 6kV or greater was actually a rather common event (although maybe "true by definition" as a consequence of how they defined the event in the first place), happening around 99% of the time.  However, at least it gives a sense as to the voltage distribution, and what realistic targets might be worthwhile while remaining economic..

 

[wkearney99]

Realistically though, at what point does this stop mattering? Lightning is more likely to have set the house ON FIRE perhaps well before all these theortical factors come into play. At some point it starts seeming like audiophile-like fetishization...

 

[RAL]

The dielectric breakdown voltage of air is about 33kV/cm between spherical electrodes.   But that can vary a lot with the humidity in the air as well as the shape and size of the electrodes.  Since relay contacts aren't spherical, it's difficult to say what it would be in that case.
 
One way to achieve a wide separation would be to use a pair of SPDT relays wired together.   But even if you use a single, wide-gap relay, if you look at the image you posted, you'll see that even though the left side and right side are widely separated, you still have a much smaller air gap on each side between the contacts.  The electric charge would have to jump the first air gap to the grounded common contact, and then if the ground isn't adequate to suppress the entire charge, it might jump the second air gap to the circuit.  I think the grounded common contact and the double air gap give you good protection, although I have no idea how to come up with an equivalent breakdown voltage for that.  Overall, my gut feeling is that it would be better with the outdoor side of things being grounded in the first place, which would make it even more difficult for an electric charge to jump the first air gap.

 

[RAL]

Realistically though, at what point does this stop mattering? Lightning is more likely to have set the house ON FIRE perhaps well before all these theortical factors come into play. At some point it starts seeming like audiophile-like fetishization...

 
I agree, it is possible to carry things too far.  Depending on where you live, the probability of a lightning strike can vary a lot, but if you do get hit, it sure can do a lot of damage.
 
The best way to protect the house is to keep the lightning from ever entering your house. Disconnecting wires with a relay mounted outside the house will give you better protection than leaving the wires connected and protected with just MOVs. 
 
A few years ago, a friend of mine had a tree in his yard get struck by lightning.  The tree was about 100 feet from the house, but the wire for his invisible dog fence happened to run across one of the tree's roots about 10 feet from the trunk of the tree.   The lightning jumped from the root to the fence wire and traveled through the wire to the fence controller in his garage, and from there, entered the AC wiring of the house.   It didn't set the house on fire, but it took out about half of the electronic items in his house, as well as the refrigerator and a couple of other appliances.
 
It's hard to say whether a relay isolation circuit would have completely eliminated the damage in the house.  I believe it would have at least reduced the damage.  
 
Adding the relays should be relatively cheap if you don't get too carried away.

 

[NeverDie]

Realistically though, at what point does this stop mattering? Lightning is more likely to have set the house ON FIRE perhaps well before all these theortical factors come into play. At some point it starts seeming like audiophile-like fetishization...

 
From what I've read, the number of house fires per year caused by lightning is actually pretty low.  Something like 14/year in the US.  i.e. the possibility is so remote that it really shouldn't be a concern for anyone. 
 
If I were living on the California coast, I wouldn't worry about lightning at all, as the amount of lightning there is paltry.  Perhaps that's like that where you live.  However, lightning barrages do seem like a fact of life where I currently live, which is Texas.  I've already experienced enough flash density here that I'm skeptical typical grounding and MOV's alone are going to cut the mustard.  i.e.it seems almost inevitable, not theoretical.  
 
Anyhow, as strange as it seems, nothing seems to outperform simply unplugging your gear during an electric storm.  Nothing rarefied about that.  The question then becomes: how to automate such a seemingly simple thing, for when I'm not home to do it?  This is, after all, an automation forum.  Relays are easy to understand and usually inexpensive, so if there's an easy way to automate the unplugging using some kind of relay, then why not?  Again, this is brass tacks, not something obscure or hard to grasp.  

 

[wkearney99]

 I believe it would have at least reduced the damage.  

 
So would not having the dog, thus no fence.  Seems stupid to have something with an underground wire not isolated.
 
At a certain point it's either drive yourself crazy trying to prevent it or just have a decent insurance policy.  Given the hassles involved with disconnecting everything (resetting clocks, settings or clean shut downs) it doesn't seem terribly realistic to get into utilizing hard disconnects.   Or do you plan on having a generator running?

 

[NeverDie]

RAL,
 
It looks as though there exist relays contained within a "high vacuum" to get a higher dielectric breakdown voltage.  The higher the vacuum, the higher the breakdown voltage.   Using the low-end of the range for "high vacuum" from  http://en.wikipedia.org/wiki/Dielectric_strength, then a 4mm gap might provide 80kV insulation.   Not sure how to adjust for it not being spherical electrodes, but it's encouraging.
 
So, it might actually  be feasible to get pretty high isolation using vacuum relays, or perhaps relays that are in some  medium other than air.
 
If you put two in series, do you get the same breakdown voltage, or double the breakdown voltage, or somewhere in-between?

 

[NeverDie]

So, for instance, relay RGH3-24D2524 from http://www.amstechnologies.com/fileadmin/amsmedia/downloads/4035_vacuumgasfilledrelays.pdf
 
can handle 70kV.  I'm not saying it's the best or most cost effective, as I haven't done any meaningful search yet.  Just an existence proof.

 

[NeverDie]

Or, more concretely, you can get a 40kV relay on ebay for maybe around $10.  This is more like what I was thinking about from the beginning.  It simply uses a long solenoid stroke to facilitate a really big air gap.
 

 
http://www.ebay.com/itm/High-Voltage-Relay-Ross-Engineering-40-kV-/111401060794?pt=LH_DefaultDomain_0&hash=item19f00549ba
 
There's a 120kv version of the same thing on ebay, but the asking price for that is $200.  That's why I ask whether 3 in series (3 x 40kV at $30 total) would work just as well?  I'm doubtful it works that way, but I honestly don't know.
 
You could easily imagine a motor driven contact closure that would deliver a truly gigantic air gap.  For the intended purpose (disconnecting based on advanced warning), the speed doesn't need to be fast.  Perhaps it would be cheaper than a solenoid version.  All you need is something to push-pull an insulated piston sufficient distance and a long-enough crossbar connector to bridge a big enough air gap.
 
Seems to me this could provide a clean disconnect that wouldn't depend on the quality of your grounding, or be circumvented by a lightning strike to nearby ground or a nearby tree.
 
So, with a 3 pole version, you could completely "unplug" a valuable piece of equipment (your home automation controller, perhaps?) and have it run wirelessly on battery power until after the electrical storm has passed.  

 

[RAL]

I really like the size of the gap on that relay.  But... it looks like it is a normally closed relay and the coil needs to be energized to open the contacts.  My preference would be for a relay that would open the contacts should the power to the coil fail.  Better yet would be a double throw version.
 
I can't claim any experience on how placing several in series would affect the overall breakdown voltage.  My guess is that the multiple gaps would be additive, or at least close to it. 
 
[Edit:]
 
I did some reading on multiple spark gaps.  Turns out they are commonly used in Tesla coils.   When a gap arcs over, it behaves like a short circuit, although there are some conduction loses.   I couldn't find any data on how much of an additive effect there is.  I'm left with the impression that there is some, but it might  be much lower than I initially thought.

 

[NeverDie]

I was just now noticing that weatherunderground can show you on a map where recent lightning strikes are, together with their latitude and longitude.  They age them into clusters, such as "within the last 20 minutes", "within the last 20-60 minutes", etc. It also tells you whether it was a + or - bolt, whether it was cloud-to-cloud or cloud-to-ground.  The latter is  interesting detail, but is it helpful/worthwhile to know?
 
Not sure where they get their data, or how accurate it is.  Assuming the info is timely and the accuracy is good, it would, in principle, be easy to automatically monitor the distance of the strikes from your exact location, and whether they're getting closer or further away.  I presume that's much better info than what I could get with a one-wire lightning detector.
 
Are there better sites for getting this info?
 
I'm looking at some of the recent lightning strike info, since it's coming my way even as I type this.  The weather radio went off, notifying about the approaching storm.  Does it always do that when there's lightning happening?  It didn't mention lightning in the warning.  This particular time it was warning about the hail potential.

 

[NeverDie]

Here it is more than 10 minutes later, after both seeing and hearing a large lightning strike that was loud enough to shake the windows, and it'[s still not visible on the WU website.
 
I'm not impressed.
 
Is there anything better?  Or do I have to join the club of people with the really fancy lightning detectors who are networked to triangulate in order to get really good information?

 

[NeverDie]

Hmmm...  Lightningmaps.org isn't showing the lightning strike I witnessed either, but their map is way cooler than WU.  It shows you within seconds the strikes it does detect, plus an animation of the shockwave emanating from the strike.  It also shows you in near real-time on the same map which stations detected it.   Wonder why they didn't detect it?  Lightning maps does show a lot more strikes than WU, though.  Just not the one I observed first hand.
 
I bet I know the problem: no local detectors.  There are only 3 detectors in Texas: one in San Antonio, one in Houston, and one in Dallas.  None in Austin.  So, I guess it's hit or miss as to whether they detect a stroke of lightning or not.