Our home is a colonial built in the late seventies. The home came with a single thermostat on the ground floor. Quickly we realized that the upstairs was cold in the winter and hot in the summer while the downstairs remained at a comfortable temperature. A commercially available replacement added a remote temperature sensor but its schedule function leaves a lot to be desired and eats batteries to boot.
Now it would be great to buy two Nest thermostats but who wants to spend $500 on thermostats? So the plan was to create two thermostats that communicate with a central server via Wi-Fi. PIR activity sensors will help the system tell where we are. The thermostats will use a touch screen interface to communicate / set the current and set point temperatures, and which thermostat is providing the process temperature.
The ground floor unit will have a relay module "backpack" that will switch the existing control wires for the furnace and AC.
I stumbled upon the latest generation of TI sensor tags, based off the CC 2650. The tag has a temp, light intensity, and humidity sensors on it. It also has a 9 axis IMU but I won't be using it. It claims to get over a year of battery life at 1 Hz update rate off of the a CR2032 coin cell... Which helps in the number 2 wife requirement, no visible wires. :) So I'm thinking of picking up several of them to act as my sensing nodes. After some looking around for a USB-Zigbee dongle with actual support, I came to the conclusion I would probably be stuck with Bluetooth as my communication method between the nodes and the controller.
For the wall mounted control unit in my living room, I've got 4 wires coming up from the basement currently switching 24VAC power to controller the furnace, fan, and AC compressor. My thought is to temporarily re-purpose those wires to provide DC power to the controller and establish a bi-directional serial link to a relay slave. The relay slave would be located at the furnace and would switch the 24VAC. It would also send relay status information back to the controller. My intent was to monitor the relay status directly by using one of the poles on a DPDT relay, that way I know fairly certainly if a contact refuses to make or break. I'm not too worried about contact welding as it's only 24VAC, current limited by the transformer. Lastly I would put a buzzer on the relay slave to sound loud and proud if communication with the control node is broken for more than 30 seconds.
Being bogged down in grad school I find myself with a lot of time to think on this project and very little time to move the ball forward. :/
In that vein, the home was constructed with in duct vents to (attempt to) turn off the HVAC to parts of the house. Currently I have the unoccupied rooms always turned off while leaving the master bedroom and most of the downstairs on constantly. Wouldn't it be nice to be able to use servos to open and close the vents on command? That way when we retire to the upstairs for the evening we can simply close the vents downstairs and in the morning vice versa?
One of my goals was to make a system that I could easily remove upon moving day and leave the next homeowner with a standard "dumb" thermostat.
That objective forces me to switch 24VAC to turn on the fan, furnace and compressor. A Digikey propaganda email pointed me to some Panasonic solid state relays on a SMT package. Those clocked in around $7 each, further searching yielded:
and it really got me focused on using an ESP8266 as the brains of the operation. The touch screen module I want to use can communicate via SPI and I'm sure I can find a temperature / humidity sensor that can output SPI. That would significantly decrease the I/O pin count I would need especially if for the downstairs unit, I could use an SPI GPIO expander to drive the relays.