1-Wire Weather Station Mods

This page describes two modifications I made to the Dallas Semiconductor 1-WireŽ Weather Station V3.0 produced by the AAG company. These two mods improve the resolution of the windspeed counter, and the electrical connection of the 1-wire communication line.

Photo 1 shows the original factory board with the two windspeed reed switches mounted in parallel near the RJ11 jacks at the left side of the photo. These switches register the passing of the two magnets mounted to the shaft of the wind speed vane. With this configuration the board can recognize each 1/2-revolution of the wind vane.


Photo 1: The board prior to modification. The windspeed reed switches are at the left edge of the photo.

I wanted to improve the windspeed resolution so I could measure the variance (gustiness) of the wind in addition to it's speed. This requires taking more frequent samples than you normally would if you were only interested in the average speed. With the original board's 1/2-rev resolution, the length of the sample period required to accurately measure the wind speed allows intervening speed deltas to slip by undetected. If I could improve the resolution, the software could sample more often.

Note that for the purpose of measuring variance, the ability of the station to accurately measure the true wind speed isn't important, it's the difference in windspeed between one sample and the next that's of value. For those interested, the true windspeed accuracy was determined to be +/-2% above 10MPH. That's pretty impressive given the cost of the Dallas/AAG station.

Mod 1 -- More Reed Switches

Photo 2 shows the board with additional switches mounted at the 1 and 5 o'clock positions. The new switches allow the DS2423 windspeed counter to recognize a 1/6th-revolution of the wind vane.

For the switch at the 1 o'clock position I borrowed one of the existing windspeed reed switches (since it's redundant in it's original position) and soldered it to the pads already there.

The other empty slot can't be used since the magnets would then trigger both new switches at the same time and nothing would be gained. The correct position is about 5 o'clock but that doesn't have existing solder pads. Fortuately the ground line here is beefy enough to secure one leg of the switch. The other leg is held in place by drilling two holes about 3 mm apart and running the wire down through the board and up again and then soldering it to the switch.

The holes are drilled close to the edge of the board so the magnets won't hit the wire where it's exposed on the under side of the board, but not so close that the board won't fit back into it's channel in the station housing.

I believe the switches used on the V3.0 board are the Coto Technology RI-27 Series. I don't know where they can be purchased, I scavenged mine from an older board I have.


Photo 2: The board with reed switches mounted at the 1 and 5 o'clock positions.

The new wires are soldered to both of the original reed pads so that the new switches will continue to increment both of the counters. (The original board setup has each switch increment a different counter.)

I used standard bread-board wire so that it's thickness would keep it from flopping around and getting in the way of the wind direction rotor which runs on the topside of the board. In retrospect, I could have used thinner wire and glued both it and the reed switch down. Then I wouldn't need to drill holes in the board - the glue would secure the switch in place.

Software Mods

I modified my weather station software to recognize that each revolution now produces six counts instead of just two. No other software changes are required. Both counters are still incremented - as with the original board - so the counts can be read from either memory bank.

Testing

My main concern with the new switch arrangement was that at high wind speeds the DS2423 chip might not be able to distinguish the more frequent switch closures, i.e. one switch might start to close before the previous one had released the line. As it turned out this isn't a problem.

Using a variable speed hand drill I was able to spin the wind vane up to 200MPH without losing any counts. At higher speeds the chip did start to lose count, but obviously this won't be a problem for most installations.


Photo 3: Testing the original and new boards at high "wind speeds". A small java program recorded counts for each board.

Mod 2 -- A Better Connection

The RJ11 connectors on the board are great for quick 1-wire hookups, however a number of station owners have reported that the electrical connection degrades fairly quickly when the station is mounted outside. Generally the best solution to this problem is to solder the wires directly to the board. That option isn't convenient for me since I sometimes need to assemble the stations outdoors, at their final location.

I noticed that a screw terminal with pins on 0.15" centers fits perfectly into the holes for pins 1 and 5 of the RJ11 connector. From pin 1 I had to jumper to the 1-wire data pin, which is RJ11 pin 3.

A two pin terminal is sufficient for the data and ground lines. I saw that DigiKey sells a 3-pin terminal (part #277-1274-ND) with pins on 0.05" centers that will plug into the 1-3-5 or 2-4-6 holes. That would take care of powering VDD also.


Photo 4: The underside of the board with the terminal screw block replacing one of the original RJ11 connectors.

Comments can be addressed to boardmods@baywind.org