Skip to content

Latest commit

 

History

History
187 lines (123 loc) · 9.62 KB

README.md

File metadata and controls

187 lines (123 loc) · 9.62 KB

piook

Raspberry Pi On-Off Keying (OOK) Decoder for the ClimeMET Weather Station, model CM9088 (with 433MHz remote/outdoor module).

Overview

piook is a small program for the Raspberry Pi computer that will decode radio transmissions from a ClimeMET weather station remote module, in conjuntion with a cheap 433MHz radio receiver module wired to one the Raspberry Pi's GPIO pins.

CliMET Weather Station

The CliMET CM9088 Weather Station consists of two items; an indoor module with an LCD display, and a remote/outdoor module (part number CM7-TX for mounting on an outside wall for taking readings of the external temperature and relative humidity (RH).

The remote module periodically (approx. every minute) transmits a short burst of data to the main module for indoor display of the external readings. Radio transmission occurs at 433Mhz using on-off keying, a crude but simple data transmission scheme in which the transmitter is simply switched on and off very quickly with well defined off intervals defining the binary bits of the data sequence to transmit.

433MHz Radio Receivers

433MHz is a standard band for low power transmission; devices using on-off keying on this band are common e.g. wireless doorbells, thermostats and electricty meters typically use this band. As such cheap 433MHz receiver and transmitter modules are widely available.

I initially attempted this project using the widely available and very cheap XY-MK-5V receivers, which appear to be a simple regenerative or superregenerative type receiver; a primitive circuit with low sensitivity and selectivity, and indeed I was unable to get these to receive a clean signal beyond about 5 meters from the transmitter.

Web research lead to the RXB6, a superior superheterodyne type 433MHz receiver module.

Connecting to a Raspberry Pi

Both the XY-MK-5V and RXB6 have simple physical interfaces consisting of +5V and ground connectors for power, and a single data/signal connector that should typically by wired to the positive power rail via a pull up resistor to prevent a floating voltage on the data line; I used a 10k Ohm resistor for this.

The data line can then be connected to one of the Raspberry Pi's GPIO pins which we can interface to in software using the very handy wiringPi project.

Compiling piook

piook consists of just piook.c and associated header file, piook.h. The only non-standard dependency is wiringPi, for which installation instructions can be found at http://wiringpi.com/download-and-install/. Briefly though the steps are:

First check if wiringPi is already installed with:

gpio -v

If not then:

git clone git://git.drogon.net/wiringPi
cd ~/wiringPi
git pull origin
./build

The compiled library will be installed in the relevant system library folder, so from now on we can link to wiringPi with a -lwiringPi option passed to gcc or g++.

To compile piook run the following command in the folder containing piook.c:

g++ piook.c -lwiringPi -O3 -o piook

The -O3 option is optional, this is the highest compiler optimisation level.

Running piook (Usage)

Usage:

piook pinNumber outfile

pinNumber: GPIO pin number to listen on. Based on the wiringPi 'simplified' pin numbering scheme (see http://wiringpi.com/pins/)

outfile: filename to write data to.

Notes.

  • Must be called with root privileges.
  • piook will listen on the specified pin for valid OOK sequences being received by the attached radio module.
  • Valid sequences are decoded to a temperature (in Centigrade), and a relative humidity (RH%) value.
  • Temperature has range -204.7 to +204.7, and precision of 0.1.
  • Relative humidity has range 0-100, and precision of 1.0.
  • The decoded data is written to the output file in the format: temp,RH with a newline (\n) terminator.
  • Each received transmission overwrites the previous file, i.e. the file will always contain a single line containing the most recently received data.
  • Project URL: http://github.com/colgreen/piook

Data Modulation

Each transmission consists of a series of on-off transistions of the transmitter that are observed on the data pin of the radio receiver module as a series of voltage transistion steps. The steps encode binary bits; analysis of multiple transmissions yielded the following knowledge:

  • Each on-pulse has duration 1000 µs (1 millisecond).
  • There are two different durations of off-pulse, 500 and 1,500 µs
  • The short (500 µs) off-pulse represents a binary 1.
  • The long (1,500 µs) off-pulse represents a binary 0.

Message Format

Each transmission conveys 48 bits of data (6 bytes or 12 nibbles).

The message format is as follows:

byte# nibble# designation
0 0 Fixed preamble 0xF
0 1 Fixed preamble 0xF
1 2 Fixed preamble 0x8
1 3 Random code/ID (1st nibble)
2 4 Random code/ID (2nd nibble)
2 5 Temperature (1st nibble)
3 6 Temperature (2nd nibble)
3 7 Temperature (3rd nibble)
4 8 Humidity (1st nibble)
4 9 Humidity (2nd nibble)
5 10 Checksum (1st nibble)
5 11 Checksum (2nd nibble)

Notes.

  • The first three nibbles can be considered to be a fixed sequence/preamble that can be used to detect the start of a message; however, the first nibble tends to be maksed with noise, hence piook tests for the sequence 0xF8 instead of 0xFF8. I imagine additional leading bits are present to allow for just such lead-in-noise issues.

  • Nibbles 3 and 4 straddle a byte boundary and contain a unique code that is generated by the remote unit when it is powered on. I haven'to examine how this code changes to test if there is any p[attern, but I expect it's possible that it is randomly generated. The purpose of this code is to 'lock' the remote unit to the indoor receiver unit. The indoor unit will learn the first code it receives and will ignore any messages with a different code (e.g. from a neighbour who happens to have the same type of module). piook does not use this code.

  • Nibbles 5,6,7 convey a temperature reading. We can decode to an 11 bit integer by appending the least significant 3 bits of nibble 5 with all the bits of nibbles 6 and 7. This gives an integer value of between 0 and (2^11)-1 = 2027. Dividing this by ten gives the magnitude of the temperature. the next least significant bit of nibble 5 indicates the sign, 0 => +ve, 1 => -ve. Hence in principle this encoding scheme can encode temperatures from -204.7 to +204.7 degrees centigrade, with precision of 0.1 degrees C.

  • Nibbles 8 and 9 convey humidity. Just decode as an integer. Humidity has range 0 to 100, precision 1, hence values above 100 aren't ever used.

  • Nibbles 10 and 11 are a checksum. The checksum is a CRC8 variant. See the source code for precise details.

Theory of Operation

How this program works is actually rather dumb and inefficient, but it does work, read on...

The data line from the 433 MHz receiver module will receive random noise when no transmission is in progress, hence the voltage on the GPIO pin will just be noise. When a transmission begins the voltage will follow the pattern of the on-off pulses.

piook attaches an interrupt handler to the GPIO pin which executes upon each low-high or high-low voltage transition, hence when there is noise on the pin the interrupt handler is being constantly called, possibly thousands of times per second, certainly hundreds. The interrupt handler records the time since it was last called and this allows it to determine if the transition corresponds to a binary 0, 1, or noise. For 0's and 1's the bits are stored in a buffer until there are enough to pass to the decode subroutine.

Not that in principle the code is not thread safe since the interrupt handler may (I'm not 100% sure) be called during an already running instance, and there is no attempt at thread syncing in that eventuality.

Reverse Engineering the Data Modulation and Encoding

Message format was determined partly from internet searching and partly from reverse engineering the received signals. A raw signal can be recorded and manually examined by attaching the data pin of the receiver to an audio line in of a PC and using audio recording software (I used Audacity on Windows) to record the raw pulse trains. By comparing the pulse trains with the temperature and humidity readouts on the indoor modules we can gradually determine, firstly where the temp and RH data is located and with some effort, how those values are modulated and encoded. In order to determine how negative temperatures are encoded it was necessary to place the remote module in a freezer for a brief time. Similarly, the high end of the temperature range was tested by using an oven on a low heat.

Notable resources:

Colin, January 26th, 2017