Lighning protection

Yes, that is a good rule.

I did build a system to transmit bidirectional RS-232 between buildings quite some years ago. I had a separately-shielded twisted pair cable run between buildings, about 20 pair, I think. Then, I made up little boards that had a pair of op-amp circuits on them, one a differential transmitter, the other a differential receiver. The single-ended levels were RS-232 levels, the differential signals were something like 1 V differential, so the slew rate was less of a problem. This worked great at 9600 baud. it also withstood several close lightning strikes that blew out various gear in the buildings, and completely blew up a locally-built optoisolated system for the same purpose. I assume that system was popped from the RS-232 side.

The resistances used in my differential converters plus the larger voltage range of the op-amps must have been part of why my design survived.

Your rule #1 only works if all the gear in the room is bonded to the same ground. Lots of places have several power panels feeding different equipment in one room (such as terminals on one feed, and computers on a different feed.)

Some CEs I knew long ago had a data center in downtown St. Louis that was fed from THREE different substations. The problems they had there were beyond belief. People sometimes reported sparks jumping between equipment and power panels, people being knocked on their rear ends when touching two pieces of nearby gear at the same time, and LOTS of crashes, where 5 computer systems from different vendors would crash at the same instant. One of these substatiosn was fed by a frequency converter station from Ontario Hydro, fed 25 Hz power to the substation that converted it to 60 Hz with giant roatary converters that were half a city block wide.

Jon

It's a violation of US national electrical code to have 'different' grounds, of course; there's only one exception, for secondary power systems (like, isolated power in hospital operating rooms). Even labs with busbar secondary ground systems, have bonding to the power ground.

I hope rule #1 will always work. Formerly, I hoped and trusted...

Then the first thing to do is to include a 4-foot passivated ground rod in the instrument's kit, to be pounded into the ground at the duct port entry to the receiver's building location. This needs to couple into the wire bundle's ground/drain/earth conductor using braided grounding wire or strap. Supposedly the same arrangement can be expected at the TX end, using the same kit. This is actually the best location for any air-gapped suppressors on your bill of material, if access to data or power conductors is physically possible. Same principal as a domestic residence's antenna lead-in suppressor.

Transient suppression works on the assumption that there's a route to redirect the current involved, that does not involve the sensitive circuitry, or other sensitive leads exiting the equipment container. The bulk energy entaileded is expected to dissipate in the tailored transient path, with any residual local voltage disturbances (shorter duration peak, lower amplitude tail) being handled by lower-powered limiters.

Most circuits that you see published already expect that this basic protection level is present. Limiters at the service entrance levels can be expected to degrade with use and to require regular inspection/ replacement.

The northern districts of the Indian Subcontinent aren't the worst place for lightning, being on par with the American gulf states, but an average of 1 strike a week per square kilometer isn't something to ignore.

RL

I should have said bonded to the same PANEL. Yes, all the grounds typically sprout from one common point. But, when lightning hits nearby, the magnetic field can be enormous and easily penetrates the whole building. If you have a big loop, where one feed comes around the left side of the room, and one comes in the right side, maybe coming down through several panels along a hallway, you can have a loop that encloses a huge area. The induced voltage is proportional to the area of the loop. Now, you can have hundreds or thousands of Volts between the grounds on two outlets in the same room. This is, of course, not the best practice in wiring up a building, but is quite common. That will cause stuff to let out the magic smoke.

Jon

It would be cool if you could collect and transmit the light from the burning cotton with a fiber. Something big and tacky like Toslink.

Interesting idea--pipe the signal all the way to detection modules installed in the panel, and so avoid the whole mess.

There's a bit of an etendue issue, though, and a lot of the signal is in the IR where plastic fibre isn't any good. My first official act on this project was to spend about a week and a half doing photon budgets for various scenarios. We wound up with a small InGaAs diode as the most cost-effective choice, interestingly.

Cheers

Phil Hobbs

Which very likely would instantly be stolen and sold for the metal value. In rural Bangladesh, folks are super poor. (You can tell a first-world country because the poor people are fat. That's far from true elsewhere.) And of course people steal copper wiring and plumbing even in Canada and the US.

Part of the fun of this project is that I can't make a lot of assumptions of that sort. The customer has been in the business for quite awhile, and so can provide good guidance about what has and hasn't worked in the past, but this system's architecture is quite different from what they've done before.

Cheers

Phil Hobbs

I don't suppose the ducts are willing to be straight, to give you a long view. Stuffing wide-angle light into a tiny fiber is inefficient, so you'd need a lot of light for that to work.

I wonder what burning cotton "sounds" like, namely converting the light to audio.

How about variations of ISM radios (whether it be WiFi, Bluetooth, or some sort of mesh) available for cheap, these days? It's probably cheaper and more reliable than wires, especially for low bandwidth uses.

On a sunny day (Thu, 28 Jan 2016 10:34:32 -0500) it happened Phil Hobbs wrote in :

I have seen that situation, and we decided to go optical fiber. that worked, and had not speed limit due to capacitance,

On a sunny day (Thu, 28 Jan 2016 20:26:49 +0200) it happened snipped-for-privacy@downunder.com wrote in :

^^^^^^^^^^^^^^^^^^^^^^^^^^^ Exactly.

You got it. :)

joe

I need timing information so we can make sure the diverter door opens at the right time. A glass multimode fibre bundle of at least a millimetre might work, but that's no longer so cheap, and needs more than a wrench and screwdriver to install.

Cheers

Phil Hobbs

Ok, here's a crazy idea or two. Have the sensor and detection logic on the machines. Power from whatever source is available. When fire is detected, sound a sonalert. The sonalert fires into plastic water pipes, and the sound is detected at the other end in the control room.

Somewhat more down to earth, a scheme I thought up for a remote rain gauge that would be lightning-resistant, would be your sensor/detector flashes a laser into a fiber that is detected at the control room. By sending a LOGIC signal on the fiber instead of the raw sensor input, you can't possibly not have enough signal. I'm guessing the sensor conditioning/detection threshold stuff isn't really complex.

If you want to make sure the sensors don't lose power, you could have the logic signal be laser on = OK, and maybe laser off = failure and laser flashing = fire detected.

Jon

threshold stuff isn't really complex.

It's a picoamp linear/log TIA and a couple of LM358s. The photon budget an d low cost are the most interesting parts.

Not that simple. It isn't just a fire alarm, it finds the first signs of fi re (a spark) and kicks open a solenoid-powered divertor door to dump the bu rning material someplace safe. The panel will poll each concentrator box pe riodically. Fibre isn't a good solution here because its installation and maintenance requires a level of cleanliness and skill that's unlikely to b e available on site. It's also expensive, at least if you get the armoured kind, which we'd certainly need.

Cheers

Phil Hobbs