Power supply EMP protection

The documents about lightning surges I have say there are significant common mode and differential parts. Additionally, my experiments in LTspice indicate that the currents can be in the 100A ballpark, which would make a CMC rather huge. The high voltage across the windings would also require TIW, so the dimensions inflate again.

An air core 500uH/3A single inductor made of 1mm TIW would be about the size of a matchbox. Can you make a reliable CMC of comparable size?

I like this idea. Even a modest amount of L, but replicated sufficiently many times should slow down the lightning dI/dt to a survivable level. Thanks, Win!

Best regards, Piotr

Reply to
Piotr Wyderski
Loading thread data ...

Just don't arc over the inductor. I had a lot of electrical metering gear installed in Florida, the Thunderstorm State, and things sometimes got seriously fried inside the metal boxes. The spark gap things are a good start.

In Mississippi, the hazard was more rednecks shooting at the boxes.

We don't get much lightning here. The power lines don't even need the high ground wire between poles.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

You won't be able to withstand direct or very near hits. But pulsed ground spikes of tens of kV might be handled with a distributed system. A nearby ground may be forced to handle high currents, but your system needs to accept voltage spikes without incurring high currents flows.

--
 Thanks, 
    - Win
Reply to
Winfield Hill

Of course there are both common mode as well as differential spikes.

An incandesce lamp hanging from the ceiling only needs differential production. It doesn't care if the lamp holder and preferably the whole house suffer a common mode ground bounce of several kilovolts.

The lamp "communicates" with the external world through the glass by sending light (and heat) to the environment. This radiation is not affected by the common mode voltage differences even at megavolts.

An electronic device wanting to communicate with the external world with unknown grounding arrangements, the best option is fiber optics, but galvanic isolation surviving a kilovolt or more is often sufficient.

Reply to
upsidedown

Your voltage drop margin is not that big, since you are going to require some voltage for the (low drop) voltage regulator.

I would suggest a SELV isolation transformer with preferably primary and secondary in different compartments. This will have several kilovolt isolation between primary and secondary and the primary to secondary capacitance is low (perhaps only 10 pF), so the common mode voltage coupling between primary and secondary will be small.

A spark gap on the primary side should help protecting the isolation transformer and some lower grade protection on he secondary side will do the rest. Defending against the direct induction into the 100 m twisted pair may need some mild protection on the load side. At 48 Vac, the current is only about 0.5 A

With 48 Vac SELV feed you most likely would need three separated and in practice isolated power supplies to make 3.3 Vdc from 48 Vac.

If the other devices are powered from the same power supply feed and there _verifiable_ no other path to ground, the CAN/RS-485 links may survive without isolation, provided that the power and data lines are bundled together, so that no large area loops are created between power and data.

Reply to
upsidedown

Prepare to be pleasantly surprised:

formatting link

802.3at-2009[6] (PoE+ or PoE plus) up to 25.5 W (Type 2) 802.3bt[11] 55 W (Type 3) and up to 90-100 W (Type 4)
Reply to
Przemek Klosowski

While the CAT standards specify how well each individual pairs should be twisted, but do these specify how individual _pairs_ are twisted together ?

After all, power in PoE is connected between the center points (via CT transformers) of two pairs. Any unbalance of the two pairs will increase the interference connected to the DC power. A separate well twisted pair for power might be a better solution compared to not so good twisting of the (four) twisted pairs around each other.

Reply to
upsidedown

The twist rate will have *no* effect whatsoever. What's more likely to cause such DC phantom powering to interfere with the differential signals are intermittent connections in the various connector assemblies (as well as any permanent joints in the cabling - badly made IDCs, not forgetting intermittent breaks in the single core wires making up a CATn cable due to damage or manufacturing defects).

Since such DC wetting currents tend to break down the initial thin oxide insulating film at such poor contact sites which can block the millivolt to hundreds of millivolt ac signals, this issue of 'induced noise' by such phantom powering is not regarded as a 'problem'.

Indeed, it is often regarded as an 'asset' since it provides an early warning, along with easy traceability of the location of faults that would otherwise produce difficult to interpret and track down symptoms of signal corruption (bad connections will be a problem in any case, regardless of whether or not phantom DC powering is involved).

--
Johnny B Good
Reply to
Johnny B Good

Taking this to the extremes, think about there are two separated well twisted pairs, so full-duplex Ethernet works fine, but these pairs run in parallel without any twist.

Much of lightning EMP power is at 1 MHz (wavelength 300 m) or below, so quite a lot will be induced into 100 m nontwisted wires.

These are known phenomenons close to kW to MW transmitting antennas working at LF/MF/HF/VHF.

Lightning EMP can cause similar field strengths.

Reply to
upsidedown

There is no magic about lightning protection, just solid good engineering practice.

Unfortunately this requires that lightning protection design starts literally from the ground when the house is built.

Trying to later on add some gimmickry sold by some snake old vendors to an existing building, might not produce the expected level of protection, unless you really understand the current paths involved.

Reply to
upsidedown

Hun? There's no common mode connection of POE with the signal, and no differential-mode either, because the transformers cannot put out any representation of the cable-side center-tap voltages into the data-carrying computer-side windings.

If there were some capacitive ooupling, the 10 MHz to 125 MHz signalling of data is pretty far removed from any power-supply frequency, and the 'common'mode' rejection of the receiver makes even that hard to make an interference out of.

Reply to
whit3rd

I am not claiming that any DC/LF voltages are going through the transformer. However, if you intend to use DC voltages present in a PoE cable, you must make connections to the transformer primary center taps to extract the DC power.

If there are some differential voltages between the pairs, it will show up as voltages at the DC/DC converter input. The DC/DC converter must be strong enough or some protection against DC/DC converter input will be needed.

Reply to
upsidedown

There is a little bit of magic. Lightning rods on buildings don't quite work the way most science articles tend to describe them.

The building had quite well designed lightning protection but unfortunately thieves had stolen the thick copper lightning conductor from the exterior walls. It was one of those big industrial sheds with a sheet steel formed roof on brick walls so approximating a capacitor.

Strangely despite having very tall supergrid mains pylons nearby the bolt hit the apex of our roof and then went internally finding the easiest path to earth. It wasn't hard to find the damage.

--
Regards, 
Martin Brown
Reply to
Martin Brown

Surely CCA lightning conductors would make more sense all round. Stamp them CCA & the scrotes will eventually learn.

NT

Reply to
tabbypurr

I doubt it. They steal copper earthing bars from substations, sometimes collecting Darwin awards in the process. John

Reply to
jrwalliker

but not CCA, as it has near zero scrap value.

NT

Reply to
tabbypurr

Even better, make it hardened, or use the stuff the power line companies use -- stranded, aluminum with cladding if possible, with a couple strands being hardened steel. You know, for... support. It's just accidental that the stuff is brutal on wire cutters...

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
 Click to see the full signature
Reply to
Tim Williams

Some power companies will bring copper wire directly to the front doorstep: making it too easy for the copper thieves.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

Non ferrous scrap metal theft is only a serious problem during boom years - it has largely ceased during austerity. Various countermeasures have also been imposed on scrap metal dealers to try and control it.

--
Regards, 
Martin Brown
Reply to
Martin Brown

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.