1-Wire Surge Protection

Background

A number of people running weather stations have had problems with nearby lightning strikes -- myself included. Typically the failure is of the DS9097U. In some cases, temperature sensors are also lost. I beleive that it also can cause problems with the raingauge. There is more information on the sensors that I have.

What happens during a lightning event?

The lightning strike causes a large magnetic field to be created -- this induces significant voltages in long pieces of wire. These voltages can exceed 1kv per meter of wire (if the strike is within 100 meters). In a real network, some of these voltages cancel out, but there are still large voltages and energies to be handled. The big voltages will be 'common mode' -- i.e. the same voltage will be induced on each conductor. However, depending on the twisting in the cable that you use, you can also get a much smaller differential voltage induced.

The differential voltages are enough to kill sensors that are attached to the ends of long runs of cable. The common mode voltage doesn't impact them at all.

The common mode voltage plays havoc with the DS9097U and the computer's serial port. Typically what appears to happen is that the ESD protection diode (DS9503) sacrifices itself to prevent the energy from frying the computer completely.

What doesn't work

I had my 1-wire network plugged into a standard telephone surge protector. It didn't protect the DS9097U. However, a telephone circuit can have up to 200V on it in normal use, and this is more than enough to fry the ESD diode.

What I'm doing now

I have two circuits -- one for protecting sensors, and one for protecting the DS9097U. The inline circuit looks like this.
The idea is that one of the diodes (both 1N4004) prevents GND from going above DATA. The 12V Zener protects the DATA from going more than 12V above GND. The other diode is reduce the capacitance of the circuit to an acceptable level. The 47Ohm resistor acts as a current limiter during events -- it limits the current to about 300mA.

I chose a 12V zener as that is the lowest voltage which has a small leakage current (5uA). All the parts are supposed to be able to handle in excess of 30A, and so they are unlikely to die in service.

Happily this circuit can be layed out in a straight line.
I just put a bit of heatshrink tubing around the components and shrunk it. It made a nice package.

Suppressor for DS9097U


This is a picture of the main surge suppressor that I built. I should have taken the picture before I encased it in heatshrink tubing. Also, the ground wire really should come out of the other end of the tubing, but I was following my diagram! Note that in order for this circuit to be effective, you must attach the ground wire to the chassis of the computer -- or something with a very good ground.

The parts are:

The actual circuit looks like:

The physical layout is:

Yes -- I know that I haven't been consistent with using a dot to indicate wire joins. If two wires are crossing, and there is no dot, then the wires do not touch. I'm too tired (as I write this) to fire up my circuit design package to draw pretty diagrams.

The diodes and TVS part are large and I had to turn the temperature up on my soldering iron to get the solder to flow properly.

You might ask, why choose devices that can handle 60 Amps -- you can't possible get that much down the cable from the outside? The answer is that I bet you can (for a short enough time). The lightning pulse induces current for less than a millisecond. The cable has a resistance of under 0.1 ohm/meter. Thus 60 amps translates into 360 Watts/meter, or 0.36 Joules/meter/millisecond. This is not going to heat the cable up noticeably, and so I guess that it will survive! We just need to make sure that the surge suppressor survives -- hence the apparent overkill.

Principle of Operation

It works in very much the same way as the first suppression device. The difference is that the chassis ground wire must be no higher than the network ground wire, and the network data wire is no more than 10 volts or so above the chassis ground. The reason for having two resistors is to protect the DS9097 against a large common mode positive surge -- both network ground and network data will go up to about 14 volts or so, and the 22 ohm resistor will limit the current into the DS9097 to under 1 Amp. This can be handled by the ESD protection diode.

Does it work?

Yes! No lightning has come anywhere near since I added this circuit! Seriously, I don't have any way of testing this stuff apart from with a multi-meter. I know that what I have built corresponds with what I have drawn. I got the circuits from a booklet put out by one of the TVS companies, and the circuits look right.

What are other people doing?

You can check the mailing list. If anybody wants to give me a link to what they are doing, I'll add it to this page.



Comments: Philip Gladstone
Last Updated: October 8, 2001