• Introducing a Cast of many Characters

    FrazzledBadger03/02/2015 at 20:19 0 comments

    So a hardware list..

    The Braiiiiiins

    Firstly the processor, I'm a PIC man through and through, ain't no stinkin' Atmel on my boards.. Flirted with ARM, but found the lack of simple to set up and free IDEs frustrating..

    I wanted something cheap, low powered with a reasonable set of peripherals. I went for the PIC18LF25K22, big fan of the K range at the moment, low voltage parts with a nice little chunk of internal EEPROM, unlike the J range. This little fella has 16k of Program Memory, 1536 bytes of RAM, 256 bytes of EEPROM, 2 Master SPI/I2C ports, 2 EUSART ports, 2 Comparators, a ADC Module, a capacitive touch module, a DAC Module, a bunch of timers, PWM capability and an internal 16MHz oscillator. All for £1.89 in 1 off quantities on Farnell and in a tidy little QFN package. I do like QFN..

    It can also be run up to 64MHz, and down to 1.8V. I'm using the internal oscillator as this saves on money and board space. A crystal would pretty much be half the size of the processor..

    I've recently started using the TAG Connect programming cable : http://www.tag-connect.com/

    This is a great invention, it means no added cost adding a programming connector, and is a lot smaller than my previous method.


    The Magic Link

    The inspiration for this project came from the RF chip, the Silicon Labs Si4455 Transceiver. This marvel can be set to either 283 - 350MHz, 425 - 525MHz and 850 - 960 MHz. I'll be using it in the 850 to 960 range. Modulation is ether FSK/GFSK or OOK/ASK, up to a power of +13dBm, at a data rate up to 500kbps.

    Communication is SPI, and all the chip requires is a matching network, Crystal and antenna.

    All for a princely sum of £1.57, in a tiny QFN package again. Did I mention I like QFN?

    I'm using a 868MHz Johnson chip antenna, costing a measly 92p..


    The Muscle

    For a power supply, I needed something cheap, I'm used to bunging in a Linear Tech chip, but that would have blown my budget straight away. Chips of Kings indeed.

    It needed to be a boost convertor that can cope with very low startup voltages, current capabilities aren't so much of a worry as it should be a low power project.

    I went for the Microchip MCP16251T-I/CH, this is a little 6 pin SOT23, can deliver up to 650mA, can start up from 0.82V, and can run down to 0.35V once everything is ticking away.. Output range is between 1.8V and 5V. Oh and its 50p...50p!

    I'm going to measure the battery level as well in the processor, to do this I've run it through a TS5A3159 analogue switch, I discovered a foible years ago that if you have a processor powered down, but a voltage applied to a pin, it can wake up again. The analogue switch provides a true disconnect. I guess a N-Mosfet could also be used.

    I'm going to run the whole system on 1.8V.


    Feel no Evil, See no Evil

    For this design, I'm planning on using it as a dev platform, so I want to include a few sensors and access as many of the pins as I can. I'm including a temperature sensor and ldr light sensor. Spare processor pins will be routed to external headers so I can test other sensors including PIR, smoke, strain gauge etc. The temperature sensor is the Microchip MCP9844, this is accurate to +/- 1deg, is I2C and is in you guessed it, a QFN package..

    All for 79p.

    I'm just using a bog standard LDR as the bottom resistor in a divider pair.

    For the external headers, I'm going to use 2.54mm SIL Headers, these are nice and easy to solder to, and potentially can be breadboarded. To these, I'm routing a I2C port, a USART port, a SPI port, 3 Analogue In pins, and 3 GPIO pins. If needed the second USART port can be...

    Read more »

  • Not quite the beginning but a place to start...

    FrazzledBadger02/20/2015 at 21:41 0 comments

    This project is my first step on the road to Home Automation. I decided to start from the ground up and build some boards to test a low cost, low power rf communications network. I know there are plenty of projects and ideas out there to accomplish this, but I decided to roll my own for a number of reasons.

    Firstly, I'm the type of person who likes to build things from scratch, I feel it helps me understand how it works, I can spec it exactly how I want it, and it's also a good learning exercise.

    Secondly, I want the individual nodes to be as cheap as possible, I'm aiming to get each one under £10, including case, sensors etc.

    Thirdly, there's no worries about obsolecence as I can easily change anything on the project.

    Fourthly, I can design each node to the exact size and shape I want, i.e. very small.

    Each node will have a rf module, processor and boost converter on it, basically making it a self contained unit. I've designed a development board that has a temperature sensor, light sensor, battery monitoring and a good selection of breakout pins on it. This should be multipurpose enough for my first range of nodes, which will be window/door sensors, general light/temp condition sensors, pir sensors and window blind/curtain motors.

    I'm not sure what to use as a main hub, but I will probably go for a raspberry pi for ease of use. Future upgrades will be light/socket control, some neopixel lamps, some sort of clock and a fridge milk sensor to detect how much milk is left. I want it all network/internet controlled