An Inexpensive APRS Weather Station

Detailed project description

Note: This was the original description of the project when we first started. It was intended for Russ and I to make it pretty clear what we wanted to do as well as let other folk know what we were going to do. As the project has progressed, we have accomplished everything in this description and hopefully a bit more.

The T238+ is a follow-on project. The changes from the T238 to T238+ follow the T238 description below.

One-wire APRS Weather Station
Will Beals (NØXGA) and Russ Chadwick (KBØTVJ)

This document describes what the project is from a functional, physical, and interface point of view. It also describes the overall philosophy behind picking the project components, what our design goals are, and its intended place in the weather station product arena. Also included are discussions of issues yet to be resolved and known areas of risk.

Overall Philosophy
The basic idea behind this project is to get a weather station on the air without breaking the bank or resorting to lots of nasty custom mechanics for the sensors. A further requirement is that this project had to use parts that were easily available and not surplus equipment, a basic requirement for a kit. Commercial products are quick to get on the air, but are expensive, especially when all the needed goodies to support APRS are added. There are also several good construction projects available on the web that show you how to build your own weather station from scratch. These projects rely on a lot of precise mechanics to be built or require you do things like salvage parts from defunct VCRs. These are very good projects in thier own right, but do not lend themselves to a kit. Hence this project was born as an attempt at middle ground.

The biggest hurdle in coming up with this project is the outdoor sensor unit. Since building something custom didn't not meet our design requirements, the project remained elusive until we could find a good sensor unit. Finding just an outdoor sensor unit and not the internal display electronics is difficult. Russ did find such a unit being used as a promotion by Dallas Semiconductor that is (almost) exactly what we need. It is a self-contained outdoor unit that is intended to be hooked up to a PC.

Design Goals
1) A display for local viewing and setup.
2) No PC needed for APRS support, just TNC and radio.
3) Inexpensive (under $150 for the basic unit)
4) Off-the-shelf mechanics.
5) Accurate enough to provide useful information.
6) Minimum sensors for wind speed, wind direction, and temperature.
7) Expandable for other sensors (rain and humidity, pressure, etc.)
8) Able to be used in a remote location using battery/solar power.
9) As much digital circuitry as possible, especially over the cable

Product Fit
The intended place this project is to fit is in between the High-cost, Low-effort commercial weather stations and the Low-cost, High-effort custom projects based on surplus equipment.

The commercial equipment has several other limitations besides high up-front cost that makes them difficult to justify. The baseline weather stations typically do not support an RS-232 output and this add-on can be expensive. Even with the add-on, none of these stations are intended for APRS use, so some kind of translator is needed to take the native format and convert it to APRS-ese. In all the cases we are aware of at this time, this is a PC. We don't have a spare PC for this purpose and don't want to make room for one. Also, paying $10 or so per month in power to leave the thing on 24 hours is wasteful.

The full-custom weather stations built from the ground up are very tempting projects, for a while this type of project was being actively considered. However, building all the sensors from scratch doesn't make for an appealing project.

Functional Specifications
Temperature range: -30 to +120°F
Temperature accuracy: +/- 2°F

Wind speed range: 0-99MPH
Wind speed accuracy: <2MPH over range 0-24MPH
Wind speed accuracy: <1.4% over range 24+MPH

Wind direction resolution (short term): 22.5° (16 points)
Wind direction resolution (long term): 1.0° (averaged)

All three variables are sampled and displayed locally every 5 seconds.
Averaged data is transmitted every 5 minutes.)
Average temperature is the simple average of the 60 samples over the 5-minute period.
Average wind speed is simple average over the 5-minute period by counting total rotations.
Wind direction average is computed using consensus averaging to eliminate outliers (See FAQ).
The maximum wind speed sampled during the 5-minute period is reported as the gust speed.
Min/Max temperatures are remembered for the current 24-hour period and displayed locally.

Physical Specifications
The outdoor unit is a purchased item, so dimensions are already specified.
The cable between the sensor unit and display/processing unit will can be telephone cable for runs up to 100 feet or twisted pair for runs up to 1000 feet.
There is a 4-line 20-character Liquid Crystal Display for setup and current weather data display.
There are four main control buttons (Menu, Up, Down, Select)
There are two RS-232 connectors designed to use standard RS-232 cables.
a) 9-pin female connector for connection to a PC
b) 9-pin male connector for connection to a TNC

Interface Specifications
The interface to the outdoor is a two-wire interface consisting of a ground connection and the Dallas one-wire interface that carries both data and power on the other connection. The connection will use RJ-11 connectors using the center two conductors. This is a digital bus interface that allows many sensors to be hooked up to the bus without needing extra wires. This digital interface is claimed to be able to handle wire lengths of 200+ feet using telephone wire and 1000+ feet using twisted pair network cable. An additional wire is used to supply +5V in case some sensor needs it.

The CPU interface will be transmit-only interface where weather data will be transmitted in the same format as the averaged data. The intent for this interface is to allow users to hook this weather station up to programs such as WinAPRS that accept data from various commercial weather stations. The microcontroller has some limitations on baud rate and other serial parameters, but hopefully this will not be a problem.

The TNC interface will be a transmit-only interface where the weather station will send messages to the TNC for broadcast. This data will be sent at 4800 baud, 8 data bits, no parity. A single APRS line of data will be sent approximately every 5 minutes for broadcast by the TNC. The user will set up most of the TNC parameters and the weather station will execute a small number of commands to get the TNCs buffers flushed, turn echoing off, and enter an "unconnected" converstaion state.

NOTE: There is only one UART on the micro, the CPU and TNC cannot operate simultaneously. The only reason for two connectors is to handle the differences in the physical interface and still allow standard RS-232 cables. Data will be presented on both connectors simultaneously and it is up to the user to make sure they have the right cable plugged in the right socket.

The T238+

We started with the core of the T238 learned from you what its strengths and weaknesses were and came up with the T238+.

1) We've added an MX-614-based modem board. A primary goal of the T238 is a low cost weather station. Requiring an external TNC didn't help in the low cost department, so it is no longer required. While also wired for receive, the T238+ will initially be a transmit-only device. The modem board is a daughter card using the same pinout as the T238 expansion header.

2) We've removed the hardware needed for software development from the main board. This was an added expense for everyone to support very few folk. A header for SW development is still available, but will now require a breadboard with some additional components on it.

3) The board size has been reduced to the size of the LCD module and will be mounted directly to the back of the module. All components are still through-hole components. Buttons have been removed from the main board and are now on a separate board for mounting to a chassis. These changes are to make the board much easier to put in a chassis.

4) The "TNC" RS-232 connector has been removed. With the modem board, this is "mostly" redundant with the built-in TNC. If you really do want to hook up the T238+ to a regular TNC you can still do it, but you will now need a null-modem adapter. The "computer" RS-232 header remains and will be used for software updates as well as feeding raw WX data directly to a computer.

5) There will still be one software release for both the T238 and T238+. The software will determine at startup if the modem board is present or not and react accordingly. Feature upgrades (and bugs!) will continue to be applicable to both boards. The T238 will not be orphaned.

This page was last updated April 3, 2004.