EMP-proofing my system with tvs diodes design idea

[Maast]

I've been back-of-the-envelope designing supplemental EMP and transients protection for my system using tvs diodes in conjunction with the Midnight lightning arresters (I can build them cheaper than just buying them, but they're inexpensive and the time savings is worth it). I recently saw video on Sol-Arc inverters being "EMP proof" and realized all they really did was include about $15 in tvs diodes to do it.

The Midnights MOVs clamping voltage is something like 470v, which is fine for common mode events like nearby lightning but do nothing for high voltage transients in the grid power lines.

A common mode event is where high voltage is induced in both the positive and negative lines - thus no voltage differential between components. To receive a UL rating among other things the component/system has to be able to withstand a 3000v common mode event without damage.

The design elements include:
MOVs over time wear out from high voltage transients, TVS diodes do not wear out.
MOVs can handle more sustained current in an event. TVS diodes do okay but not as well as MOVs
MOVs reaction time is 8-12ns, TVS diodes are 2-10ps
High alt nuke EMPs rise time is in the ps range, faster than MOVs can react.
Tvs diodes sustained current capacity is dependent on how hot they get
Lightning can strike anywhere in the world. I was living in Iceland and a lightning bolt hit the apartment building next to mine and toasted a modem in my computer and caused damage to the motherboard.

My thought was to encase an array of three to five 300v clamping 1.5kw tvs diodes in a .5"x2"x3" block of thermal transfer epoxy to act as a thermal mass and have one going between each power conductor to earth ground for the PV array, 48V lines, and the in/outgoing AC lines. Each one would cost about $20 in materials - most of that being the epoxy.

In an EMP event the idea is the TVS diodes would clamp just long enough (before saturation or burnout) for the MOVs to begin conducting and short the overvoltage to earth ground. In addition the tvs diodes would clamp any transients thus protecting the equipment and keeping the MOVs from degrading. My primary goal is transient clamping, the EMP protection is a highly desired "just in case".

Thoughts?

 

[Dzl]

Waay over my head, but it sounds very interesting, I will be following this thread out of interest/curiosity.

 

[Maast]

Unfortunately I don't have any way to actually test it out unless I build my own pulse tester. And I have WAAAAAY too many thing to do before I get around to doing that.

 

[Maast]

I also had the thought to dip the tvs diodes in paraffin wax before potting it in the epoxy to act as a immediately available phase change heat absorber. A clamping event would very quick and it'd be over before any temperature rise could be detected in the potted block.

With a CME the voltage would rise slow enough that the gridtie breakers would pop before overwhelming the tvs/arrester combo.

 

[DanF]

IMO, if life gets to the point where we need EMP protection.. I probably don't want to be alive anymore. But someone has to carry on life on this planet I guess..

 

[Maast]

IMO, if life gets to the point where we need EMP protection.. I probably don't want to be alive anymore. But someone has to carry on life on this planet I guess..

Not just nukes, even primarily not nukes - its high voltage transients and nearby lightning induced EMPs.

 

[Hedges]

Unfortunately I don't have any way to actually test it out unless I build my own pulse tester. And I have WAAAAAY too many thing to do before I get around to doing that.

I do have the pulser, so a good excuse to get it running. I think it is up to about 10kV. Don't know how fast, but probably nanosecond give or take.
It needs a DC source, which would just be rectified HyPot again for 7500 VDC.

My high voltage probes are only 100 MHz so won't measure that well.
When I did measure stuff like this at work (5kV 50 ps rise time) it was with Barth attenuators and a 12 GHz scope or something like that.
Those also loaded the source with 50 ohms.
I could probably use capacitive dividers to get the voltage down.
Measurements will be limited by 4 Gs/second 1 GHz scope.

 

[Maast]

I do have the pulser, so a good excuse to get it running. I think it is up to about 10kV. Don't know how fast, but probably nanosecond give or take.
It needs a DC source, which would just be rectified HyPot again for 7500 VDC.

My high voltage probes are only 100 MHz so won't measure that well.
When I did measure stuff like this at work (5kV 50 ps rise time) it was with Barth attenuators and a 12 GHz scope or something like that.
Those also loaded the source with 50 ohms.
I could probably use capacitive dividers to get the voltage down.
Measurements will be limited by 4 Gs/second 1 GHz scope.

If you get it rigged up I could put a tvs array together and send it your way. I'm in WA state.

 

[MattiFin]

Forget the epoxy. With short pulses it has absolutely zero effect.

 

[Maast]

Forget the epoxy. With short pulses it has absolutely zero effect.

True, but with a Carrington event type CME that could ramp up for several minutes before the input breakers pop. Since the tvs diodes clamping is at 300v they'd take the entire load until either they burnt out and passed the load onto the MOVs or the breakers went.

I need to look at the spec sheets of the interconnect breakers and the tvs diodes again and crunch some numbers on how long they'd need to hang on until breaker disconnect. 5 diodes may not be enough. At least they're cheap and I can add more for minimal cost.

Would molten wax act as a insulator to rapidly rising temperatures? Need to look at thermal conductivity of liquid paraffin wax.

Edit: delete the wax idea, it's a terrible thermal conductor.

 

[MattiFin]

True, but with a Carrington event type CME that could ramp up for several minutes before the input breakers pop. Since the tvs diodes clamping is at 300v they'd take the entire load until either they burnt out and passed the load onto the MOVs or the breakers went.

AFAIK geomagnetic storms are issue only for large-scale grids. Even a large DIY solar system has tiny wiring loop area compared to national scale grid.

https://www.swpc.noaa.gov/sites/default/files/images/u33/finalBoulderPresentation042611%20(1).pdf

20 volts per kilometer potential is not much for DIY installation.

5000nT/minute magnetic field change(Carrington event scale) is also next to nothing for small system.
0.000005 T change over 1 minute to 10m2 loop generates 8 microvolts
https://www.omnicalculator.com/physics/faraday

 

[Maast]

AFAIK geomagnetic storms are issue only for large-scale grids. Even a large DIY solar system has tiny wiring loop area compared to national scale grid.

https://www.swpc.noaa.gov/sites/default/files/images/u33/finalBoulderPresentation042611%20(1).pdf

20 volts per kilometer potential is not much for DIY installation.

5000nT/minute magnetic field change(Carrington event scale) is also next to nothing for small system.
0.000005 T change over 1 minute to 10m2 loop generates 8 microvolts
https://www.omnicalculator.com/physics/faraday

Yes but I'll be grid-tied so my installation will be exposed until the grid-tie breaker pops. Off grid it'd completely be a non-issue
Just did a bit of reading, lightning emp rise time is 1-10 microseconds
Nuclear EMP is 5 nanoseconds. So far I havent found anything that can react that fast. More reading to do.

 

[MattiFin]

Even with grid-tie your problem is pretty much limited to your nearest distribution transformer. This is usually not more than 1-2 kilometers from your house and in US it is often next to the house at street.

 

[ivote76]

There was a big EMP on earth in 1859. But there was nothing around for it to affect. 2012 a big one missed us by 9 days. Its only a matter of time before we get hit again but, this time, it will cause massive devastation. Its a good idea to keep back-up parts in a F cage or EMP proof your system somehow.

 

[Hedges]

These events are all a DC pulse, which carries various frequency components.

Problem with CME seems to be sustained induced V/m, which generates voltage in power lines that is applied across transformers. With DC, current rises and magnetic core becomes saturated, at which point it no longer resists AC from transmission line. The power grid then overheats the transformer.

DC PV input won't be harmed by small DC signal from CME. Anything with a transformer has the potential to have it saturated. Grid-tie transformer type inverters could experience that. Ordinarily, magnetic coupling to secondary limits primary current. With core saturated, winding resistance would determine current, and breaker would trip - hopefully fast enough to protect windings.

Off grid, inverter would be the source of power that would have to do heating damage. It couldn't maintain voltage into a saturated transformer. It could deliver enough power over time to overheat; consider 5kW 95% efficient, normally transistors and transformer dissipate 250W. If same battery current devoted to self-heating, would heat 20 times as fast. But I think drive circuits would give up for voltage or thermal reasons.

Seems to me protection against fast events causing voltage spikes is our primary concern. Those coming from AC grid could cause an accumulation of damage. I don't have data on prevalence, but the idea is promoted by vendors of surge suppressors.

I think saturation of transformers by DC from power grid is handled by circuit breakers - any opinions/analysis of that?

For CME, you will also have energetic particles causing electronic upsets. We know that causes bit flips in digital systems. Some systems have robust handling of undefined states and watchdog timers to they recover. No idea how well inverters are designed in that regard. So long as it only crashes the system but doesn't damage it, not a problem.

 

[Maast]

These events are all a DC pulse, which carries various frequency components.

Problem with CME seems to be sustained induced V/m, which generates voltage in power lines that is applied across transformers. With DC, current rises and magnetic core becomes saturated, at which point it no longer resists AC from transmission line. The power grid then overheats the transformer.

I think saturation of transformers by DC from power grid is handled by circuit breakers - any opinions/analysis of that?

Thats actually a really good point - will an AC breaker trip from a DC pulse coming from the grid?

Standard household breakers trip from overcurrent not so much from overvoltage. My first-pass thinking is the tvs/MOVs would act as a current 'sump' allowing enough current to pass through the main 200A breaker to trip. Without such a sump the overvoltage condition would pass through to the house and because household voltage is referred to a local earth ground (ground and neutral are bonded at the main breaker) it would be an actual overvoltage to the house and not a common mode event.

This means the tvs/MOV combination needs to be on the 'house' side of the breaker panel to provide the current flow for tripping the main breaker.

What happens with a CME (aka Carrington event) is the charged particles hit the planets' magnetosphere which shoves the field lines around which then in turn induces rising voltage in long wires until it plateaus and then the field lines move back after the charged particles have passed the Earth, unless you're in the far north or south I dont think the charged particles ever actually make it to atmosphere.

Its basically two halves of a single low frequency AC signal. How does a powerpole transformer react to that? Is it a overcurrent event that overheats the transformer or a overvoltage event that breaks down the transformers internal windings insulation. Both? I dont have enough knowledge of how power transformers operate, all my training is in component-level inductors. More reading.

 

[MattiFin]

Its basically two halves of a single low frequency AC signal. How does a powerpole transformer react to that? Is it a overcurrent event that overheats the transformer or a overvoltage event that breaks down the transformers internal windings insulation. Both? I dont have enough knowledge of how power transformers operate, all my training is in component-level inductors. More reading.

Not really either of them.
For powerpole transformer it is like DC and it will saturate the transformer core. Once the transformer core is saturated the transformer is going to start draw lot higher AC current.
Doesn't take that much DC current or voltage to saturate the transformer and that is part of the problem. 1 volt or 1 amp of DC can do it for rather large transformer. Not a problem for your equipment, DC loop is limited to between your house and nearest powerpole transformer.

So CME itself is not much of a problem unless you operate national grid. But it can cause various malfunctions, surges and overvoltages from grid to appear. These can be dealt in the usual way.

Protection from high-altitude nuclear blast would be completely different matter and crazy hard to implement.

 

[Hedges]

Thats actually a really good point - will an AC breaker trip from a DC pulse coming from the grid?

The presentation linked above includes a slide with graph representing an understanding of the issue (whether simulated, calculated, or contemplated, I'm not sure.)

What it shows is a bias on the current, so one side of AC puts core into saturation and current shoots up, but there is still a zero crossing.
The currents produced should cause I^2R heating of breaker sufficient to trip before damage to copper wiring, and arc should be extinguished at the zero crossing.

So my hope/expectation is that inverters and other transformer based equipment protected by suitable breakers will survive.
(Old-school wall-warts intended for currents much smaller than the 20A breaker would not.)

 

[MattiFin]

This should keep you busy for a while

https://www.cisa.gov/sites/default/files/publications/19_0307_CISA_EMP-Protection-Resilience-Guidelines.pdf

 

[Hedges]

IMO, if life gets to the point where we need EMP protection.. I probably don't want to be alive anymore. But someone has to carry on life on this planet I guess..

Consider that EMP and the horrors you imagine are two different events.
A nuke placed to cause massive destruction won't create significant EMP, and one placed for EMP will not cause any destruction.

Both could occur in MAD. But a single strike by a madman without a large arsenal might (if what we read is correct) take out the grid and any distributed, non-hardened systems. In that case preppers with survivable power systems for their well pumps and freezers would be well-served by transient protection.