lightning protection for landscape lighting system

[RJSV]

My Lord,  those last numbers indicate that,  for 100 usec. 50,000 volts drives, through 1 ohm, 50,000 amps !  ??
   So, power being the V x I,  that's 2.5 billion watts.
   Or = 2.5 gigawatts during that brief 100 usec.

No wonder these specialists talk like philosophers,  (lol) !

[johansen]

My Lord,  those last numbers indicate that,  for 100 usec. 50,000 volts drives, through 1 ohm, 50,000 amps !  ??
   So, power being the V x I,  that's 2.5 billion watts.
   Or = 2.5 gigawatts during that brief 100 usec.

No wonder these specialists talk like philosophers,  (lol) !

that's just the power dissipated in the ground resistance alone.

the rare positive lightning bolts, 10 times more powerful with currents on the order of half a million amps have been recorded.

unfortunately the Fieldlines forum is gone now but not many years ago i saw photos of a lightning strike which had vaporized a buried conductor.. which meant the dude didn't need to dig a new trench to replace the line. it had been dug for him by the vaporized copper expanding plasma. --The average lightning bolt will not melt a 4awg wire.

I did notice utilities are no longer wrapping the 6" by 3 foot long copper sheet around the bottom of the pole, they are now just strapping it to the side and bending it over the bottom in an "L" -which makes more sense. its half a microhenry less inductance..

[RJSV]

...I ran a trial number, for a hypothetical case,  where that 100 usec. of power, at 2.5 GW,  gets spread out, in time, over 1 hour.  Of course there is plenty more time, for the heat to dissipate into a lot more volume of surrounding soil,  course.
   I just wanted to see what the numbers show.  Equivalent to 1 hour at close to 670 watts.  That would likely do set any standing water to boil,  maybe even with normal heat dissipating into soil,  over the hypothetical 'one hour'.  But of course the lightning strike HEAT creation in reality does not have that dissipation process, over an hour's test, as my speculation details.   Still, it gives an insight, into that 2.5 GW dump, into surroundings,...with a grain of salt of course.

Rick

[PCB.Wiz]

Indeed, I believe the recent event (which caused me to move this project to the front burner) probably did occur via the mains power.   Because, the damage was to the Arduino in the control controller and to the LT3081s in the circuit boards there.   The strings of luminaires were completely un-harmed.

What is the cable voltage and drop across the LT3081 and LED strings ?
You may not have much headroom for TVS parts to work.

The LT3081 is a slightly unusual choice, for LED drive, but I guess they are already in place.

[RustyShackleford]

What is the cable voltage and drop across the LT3081 and LED strings ?
You may not have much headroom for TVS parts to work.

Short answer: 30 volts.  One side of the cable is getting the 30vdc from DC power supply (subject to a fuse and a power pfet to turn off/on) and the other is pulled down by the LT3081, to whatever voltage pulls the desired current through however many LEDs are in series.

Yes, LT3081 might be an odd choice, but my whole system is odd, with the luminaires being in series.

And yes, not much headroom.  I'd probably use the TVS with 30.8 stand-off voltage, for which the clamping voltage is 49.9, which exceeds the 40v absolute maximum for the LT3081.   However, probably only for a VERY short time.   Parts are rarely characterized for such brief stresses, but I imagine there's a good chance it can handle it.

[floobydust]

I still don't get what happened to the Arduino, what did the autopsy reveal? Is the 30V kicking up due to lightning transient, or the 5V or the control lines etc.?

[PCB.Wiz]

Short answer: 30 volts.  One side of the cable is getting the 30vdc from DC power supply (subject to a fuse and a power pfet to turn off/on) and the other is pulled down by the LT3081, to whatever voltage pulls the desired current through however many LEDs are in series.

How much current is needed per dropper point ?

And yes, not much headroom.  I'd probably use the TVS with 30.8 stand-off voltage, for which the clamping voltage is 49.9, which exceeds the 40v absolute maximum for the LT3081.   However, probably only for a VERY short time.   Parts are rarely characterized for such brief stresses, but I imagine there's a good chance it can handle it.

'imagine' is into real Hail Mary territory
With not much headroom, you may be better using high power zeners, which have more precise specs on knee turnover.
Series impedance helps, but load current can limit series R scope

[RustyShackleford]

Short answer: 30 volts.  One side of the cable is getting the 30vdc from DC power supply (subject to a fuse and a power pfet to turn off/on) and the other is pulled down by the LT3081, to whatever voltage pulls the desired current through however many LEDs are in series.

How much current is needed per dropper point ?

Not quite sure what you mean by dropper point.   There are a number of strings of LEDs, each driven. by a current-configured LT3081.  The maximum current I allow is 360ma, which of course goes through every LED.

And yes, not much headroom.  I'd probably use the TVS with 30.8 stand-off voltage, for which the clamping voltage is 49.9, which exceeds the 40v absolute maximum for the LT3081.   However, probably only for a VERY short time.   Parts are rarely characterized for such brief stresses, but I imagine there's a good chance it can handle it.

'imagine' is into real Hail Mary territory

Haha yes.  Can't remember some ESD-related site where I read that absolute maximums for very brief events can be higher, but are rarely explicitly characterized.   And tend to be governed by junction temperature limits.

With not much headroom, you may be better using high power zeners, which have more precise specs on knee turnover.
Series impedance helps, but load current can limit series R scope

Looking at these zeners you mention.    I'm looking at the 1N53 series, rated at only 5 watts (the highest power dissipation that Mouser shows for zeners), and I believe these curves (see attachment below) do indeed show the sharp knees.

I can't afford any series R, because cuts into how many LEDs in series.   But maybe some series inductance, placed in between the zener and a TVS ?

[RustyShackleford]

I still don't get what happened to the Arduino, what did the autopsy reveal? Is the 30V kicking up due to lightning transient, or the 5V or the control lines etc.?

Search me.   Wish I knew.   When I connect to the USB port, the Arduino IDE doesn't see it.    Didn't investigate further than that, except to note the onboard 3.3v regulator was not producing output and getting very hot.  So I assume it's the 5v.   The Arduino never sees the 30v at all.   The 5v comes from a DC-to-DC converter driven from a 12vdc switcher that's there to support "legacy" (12v) landscape fixtures.

[RustyShackleford]

Well, I let this drop last summer.   Mistake.   It's been a tough summer here, lots of lightning storms.   Three separate instances of damage.  I've gained a lot of practice re-working my driver boards to install new LT3081s.   I've also lost one Arduino and one SMPS.   I'm still not clear if the damage is coming over the AC mains or through the outside wires.   

But I decided it was time to do something. So I tried ... AI.   I'm pretty impressed with what it came up with, in conjunction with my questions and prompts.   Where the outdoor wires come into the utility room (where the control box lives), there will be this "surge protection" circuit ...



And then on the control board, very near the LT3081, three bi-directional TVS diodes (one from each LED line to GND and one between the LED lines).   I wondered at using bidirectional ones, and it came up with ...


A unidirectional TVS will instantly turn on in forward bias at just 0.7V.This creates a direct, massive short circuit that invites thousands of amperes of lightning surge current straight onto your driver board's delicate quiet ground plane. This sudden current loop will easily warp ground references, corrupt your 0-350mV control loops, or completely vaporize your 1 oz copper ground traces.A bidirectional TVS stays completely open (blocking current) in both directions up to its breakdown threshold (33.3V). It acts as a strict firewall, isolating your driver's quiet ground from the outdoor lines during a negative-going transient. This forces that massive energy to stay on the outer surge card where the heavy-duty GDT and earth rod can safely dump it into the soil.

Seems plausible.  This is assuming, as it also insisted upon, that the earth ground on the "surge protection" board be completely separate from the local GND on the driver board.

On the AC input to the two SMPS's that power the thing, there'll be a very similar circuit.

[Kjelt]

Dont forget that the pcb layout will be critical with lightning.
It will take the shortest way to earth so take care of enough distance and isolation and enough copper mass on the tracks it will not evaporate.

[RustyShackleford]

Dont forget that the pcb layout will be critical with lightning.
It will take the shortest way to earth so take care of enough distance and isolation and enough copper mass on the tracks it will not evaporate.

Indeed, and the AI is talking about this as well.   Especially an issue since I want to use Elecrow's super-cheap 100mm-square 1oz copper PCB fab (less than $20, delivered, for 10 boards); I'll probably put two of my eight channels on each board.

Suggesting 1/4" wide traces for the LED+/LED- traces up until the inductor.  And even wider for the earth ground, plus another thing I'd never thought of, block solder mask over the whole trace, then flood a bunch of solder on there when assembling.

Also, slots all the way through the board, between the 3-pins on the GDT device, to prevent arc'ing.

[Kjelt]

plus another thing I'd never thought of, block solder mask over the whole trace, then flood a bunch of solder on there when assembling.

If it is not producing for 3rd parties and only for yourself, do as I did, solder a 2,5mm2 copper wire ontop of the trace.
See my pre-previous post the fuse will protect from further damage, the sparkgap should absorb the power.

[MarkF]

... and a couple of 8ft ground rods (plus two more, one on the ground-mount solar panels, and one by the shipping container workshop).  I guess I need to add something more where the AC enters my controller - some configuration of the afore-mentioned devices.   Perhaps as simple as a higher-voltage selection of the TVS I proposed using on the DC conductors: https://www.vishay.com/docs/88369/p6ke.pdf

I believe that is the exact opposite of what you want.
When lightning strikes, it creates a voltage potential across the ground.  With multiple ground rods, you are creating a hugh current between them causing havac everywhere.

You want one (and only one) connection between you service panel and a grounding rod (or multiple rods close to each other bonded together).  You want everything inside the house bonded together to keep their potential the same (electric, phone, copper water lines, etc.).  The electrical code specifies rod spacing distances.


Side note:
  My parents house was struck years ago.  It went down only one side of the house.
There was an outlet where the faceplate and receptacle looked normal from straight on but when viewed from the side it was blown out from the wall 1/2 inch and the whole back was gone.  Just the metal pieces of the receptacle with wires attached hanging loose.

[RustyShackleford]

... and a couple of 8ft ground rods (plus two more, one on the ground-mount solar panels, and one by the shipping container workshop).  I guess I need to add something more where the AC enters my controller - some configuration of the afore-mentioned devices.   Perhaps as simple as a higher-voltage selection of the TVS I proposed using on the DC conductors: https://www.vishay.com/docs/88369/p6ke.pdf

I believe that is the exact opposite of what you want.
When lightning strikes, it creates a voltage potential across the ground.  With multiple ground rods, you are creating a hugh current between them causing havac everywhere.

You want one (and only one) connection between you service panel and a grounding rod (or multiple rods close to each other bonded together).  You want everything inside the house bonded together to keep their potential the same (electric, phone, copper water lines, etc.).  The electrical code specifies rod spacing distances.

Hmm, it's always been my impression that it's good to have add'l ground rods at separate structures (such as my ground-mount solar and shipping container).   Of course the grounds will all be connected, the one at the solar by 10awg and the shipping container by 8awg.   And of course, ground and neutral bonded ONLY at the main load center.

[MarkF]

Hmm, it's always been my impression that it's good to have add'l ground rods at separate structures (such as my ground-mount solar and shipping container).   Of course the grounds will all be connected, the one at the solar by 10awg and the shipping container by 8awg.   And of course, ground and neutral bonded ONLY at the main load center.

That has been the standard for a long time.  And large machinary still states to ground the chassis at the unit in their installization guide.  I wish I could remember where I read a discussion to this policy and that it should not be done.

Someone mentioned streets light earlier.  I believe they still ground the pole.  They were also part of that discussion.


But, think about.  A million volt strike and the wire resistance of 8AWG wire gets you how much current.  You don't want to provide that path.  You want the chassis all tied to a central point but not between two ground rods where a strike will induce a large voltage potential.


I once believed as you.  But now have stopped driving ground rods.

[MarkF]

Think of it this way:

Your home is a floating island.
Do you really want to put down two electrodes across a gigantic battery.
Cardiac arrest shock therapy for your house.

[RustyShackleford]

Do you really want to put down two electrodes across a gigantic battery.

Well, if for no other reason than to avoid driving another ground rod, I'm going to ground my "surge protecton" circuits simply by running a piece of 6awg the roughly 12 ft to where the main load center ground rods are.

That still leaves the ones by the solar panels and the shipping container ...

[RustyShackleford]

But, think about.  A million volt strike and the wire resistance of 8AWG wire gets you how much current.  You don't want to provide that path.  You want the chassis all tied to a central point but not between two ground rods where a strike will induce a large voltage potential.

Although for the duration of the lightning strike’s discharge a large current would flow through the feeder's ground wire (because of the voltage difference between the house’s grounding rod and the outbuilding's), the current would only last a few milliseconds and therefore the ground wire would not be harmed.

[RustyShackleford]

Check this out, particularly the one answer:

https://diy.stackexchange.com/questions/248041/why-do-detached-structures-feeder-panels-require-grounding-rods/248064#248064

[MarkF]

I get confused everytime I go into this stuff.
Now I'm not sure I'm telling you the right thing.

Mike Holt put out many videos discussing the electrical code.

Study this video in detail and make your own conclusions:

  https://youtu.be/mpgAVE4UwFw?t=1400


I believe there is a video for subpanels and remote buildings but I can't find it.