How to design a power regulator circuit that is lightning resistant

william chan

Hi All,

Assuming lightning comes from the power adapter, what would be the best low cost method to protect the rest of the circuit board?
Could it be

a. Add a zener to the incoming power?
b. Use a mosfet to short the incoming power? ( when overvoltage is detected )

LoopyByteloose

Well with my router, I found that the wall wart itself acted as an excellent lightning arrestor.

Of course, after it was struck by incoming lightning, it needed to be replaced. But the router did not suffer any ill effects, and it has quite a few sensitive chips.

I have replaced the wall wart with a a tiny 5VDC, 3amp switcher which has more sophisticated protection circuitry and since then I have not had any problems with lightning causing anything.

The power lines and substations have quite a lot of lightning protection equipment to isolate lightning stikes, including fuses on just about every transformer down the line.

+++++++++++++++++
So I am thinking that you may already have adequate protection unless you are building your own conversion from mains to 5VDC or 3.3VDC.

+++++++++++++++++
There is quite a bit of discussion on the internet about people that needlessly replace their routers after a lightning strike. Nothing wrong with the router, just a fried power adapter.

kwinn

Depends on how much protection you want. An LC Pi filter and TVS with fairly substantial inductors does a pretty good job, but nothing is guaranteed protection from a lightning strike.

Peter Jakacki

william chan wrote: »

Hi All,

Assuming lightning comes from the power adapter, what would be the best low cost method to protect the rest of the circuit board?
Could it be

a. Add a zener to the incoming power?
b. Use a mosfet to short the incoming power? ( when overvoltage is detected )

To help protect against lightning I use the combination of gas-tubes + ferrite beads + poyfuses + TVS diodes. Having very high voltage on the power rail is not a problem, provided it's also on the other and there are no other paths to earth. Ah, but that is always the problem, so your gas tubes take the bulk of the surge but they need a path to a proper earth, not just common or "ground"

Zeners are way too slow and a mosfet acting as a crowbar even if it could be trigged fast enough is likely to be destroyed and also take out the power supply anyway.

william chan

What about Metal Oxide Varistors?

kwinn

william chan wrote: »

What about Metal Oxide Varistors?

They will help some, but what Peter has listed is better. Add heavy duty inductors and capacitors to that and it,s about the best you can do. Nothing we can make is guaranteed protection against a lightning strike. Best we can do only increases the odds of survival.

Tracy Allen

There is an electromagnetic front that moves out from the strike, through conductive paths like the power or telephone lines, through the earth, or as a EM wave that can be picked up by any suitable antenna. Those waves move at different speeds and create transient local differences in potential. Say for example, you have not only a power adapter but also a hard wired telephone modem, or an ethernet cable, or a USB cable to your computer, or RS485 to the next building, or a cable to an anemometer up on your roof or to a sensor measuring the moisture in your garden. The potential difference swing at different rates and appear as transient potentials across your circuit. The purpose of the protection is to funnel currents off to one common point, usually earth ground, and limit the remaining potentials to a safe and physically non-damaging level, or even to a level that does not perturb the performance of the circuit.

If your device is connected to only one of the paths, such as a single power adapter, it is unlikely that it will be damaged. There is not likely to be much of a surge in the power supply voltage itself, but there may be a surge in voltage with respect to other devices in the vicinity. Is the power adapter grounded? If so, maybe you can extend the safety ground to a cage around your circuit board.

LoopyByteloose

I'd go with what Peter J suggests. He lives in Oz and seems to be using Propellers with RS485 over long distance wires. That opens up the Propeller to have more than just the mains as an entry path for lightning. Every long wire network attached to your project becomes a potential antenna and entry.

One can do a lot to make such a set up more lightning resistant; but nothing is really even going to be 100%. So have sacrificial components that can take a hit and be easily replaced seems to be the way I would go.

Many of the items that Peter J. recommends require spending $$$ and a shopping spree. So, I tend to go with just having more generic components configured in a way that they can die and be swapped out.

MOVs were long touted as the necessary device, but the reality it that they only will take one strike unless they are configured with additional fusing. The big problem is that people have no way of checking the MOV that has taken a hit. And if it is fused, they may not be able to check a blown fuse if the lightning took out an indicator LED as well.

Fiber Optic networks may help quite a bit. Taiwan shifted over to underground fiber optic for its communications backbone. It is all those wires on poles that are bringing you trouble.

Gas tubes seem at the core of any design that will survive more than one strike.

william chan

In summary, a TVS alone would be sufficient or it needs to be paired with an Inductor to be effective?

LoopyByteloose

Well, I think the prudent approach is to have one device take the majority of the strike's energy, and then a series of other devices either shut off the system (fuses) or take away any energy that got through.

I simply don't bother. But I do see schematics that have this approach. If the device is on a pole in the middle of nowhere and requires sending out a service man that has to travel for a day or more, the Full Monty approach is wise.

If you are worried about a device used within an hour's drive from your office, maybe not.

Anything is better than nothing, but understanding that most devices are not self-healing and only take one strike is important. Tons of users have had equipment die because of the the 2nd strike.. after the MOV was destroyed by the first one.

kwinn

william chan wrote: »

In summary, a TVS alone would be sufficient or it needs to be paired with an Inductor to be effective?

Each component you add to the circuit adds another level of protection so it really depends on how much protection you need. Personally I would not consider a TVS alone adequate. One thing I have noticed over the last few years is that as the number of systems with switching regulators has increased so have the number of service calls that require only a reset or cycling the power off/on. IMO that is because the switching power supplies have very small inductors and capacitors that provide much less noise filtering than an iron cored transformer and large electrolytic capacitor. Power line glitches get through switchers more easily and tend to scramble the logic states.

I am not suggesting a return to linear regulators however, switchers have too many advantages. What is needed is a bit more power line filtering to keep the noise and glitches from reaching the switchers.

bill190

In the past, I read the following and find it interesting the pass through voltage is so high for a 120 volt AC device!

["The standard let-through voltage for 120 V AC devices is 330 volts"]...

http://en.wikipedia.org/wiki/Surge_protector

Then other than that, zener diodes or MOVS... From hot to neutral, from neutral to ground, and from hot to ground. And fuses prior to that.

LoopyByteloose

To do this right, I suppose one has to have something like three or more lines of defense.

You start out with an estimate of the maximum voltage, and the maximum wattage you want to deal with. Above the voltage limit, the lightning starts jumping and arcing outside the circuit. And the wattage is what you have to dissipate in excess of what the device actually consumes.

In step along the way progressively dumps these excesses, or disconnect the device from harmful inputs. So that is a lot of speculative fussing to get the right circuit. And you never know when a bolt of lightning will be too close to home and still fry your device.

So I just prefer the idea of a sacrificial wall wart. The devices I see in Taiwan have such tiny copper traces that a bolt of lightning is likely to fry the traces right off the board and nothing gets through. I wish I could verify there was some intentional design toward protecting the rest of the hardware, but I doubt that is the case... no documentation and labels are all in Chinese except for key electrical aspects.

Tracy Allen

As for the regulator IC itself, you can choose a parts meant to withstand abuse. They will advertise that they are meant for the automotive environment or something akin and that they can withstand load dumps, reversed batteries and other such insults. Some regulators don't have that. Small surges over the power line will always be more common than large surges, so the regulator itself is the inner layer of protection.

Kwinn, Interesting observation about switcher powered equipment needing more resets.