I have finished building an HVAC system monitor, though it had been in the planning stages for months, I just needed time to get everything assembled. I had everything I needed except the relays, selecting the right ones was difficult, since I wanted a wider operational voltage and very low current draw.
It connects to the thermostat input wiring on the furnace, which in my case comes out of a multizone thermostat controller. The relays allow the Shelly i3 to read the control signals from the 24VAC wiring. I used a spare 24VDC passive POE injector to power the Shelly. The entire rig is DIN rail mounted and quite full, I will have to pull things off the ends to screw it into the wall, or wherever I decide it will go.
The Yel/Org/Red wires on yellow terminal blocks are the signal IO for the relays, the Blk/Gry/Pur wires on blue terminal blocks are the “common” wires, aka ground on DC side. The Phoenix Contact relays have a status LED so any kind of diagnostics should be simple, and they will operate at up to 1.4X nominal voltage, most of the other brands I looked at were only 1.1X.
The Shelly will run on AC, or 24VDC…, but not 24VAC, so I had to rig up the ethernet port for power, but that should be far cleaner than the transformer output of the furnace, and it can be plugged into a UPS.
So why would anybody want something like this?
First of all (the big one) is that I do not need any kind of smart thermostat to monitor the HVAC status, the dumb one on the wall works just fine, so why replace it?
It can allow estimated energy usage monitoring. I can easily find out the power draw of the AC unit and the gas usage of the furnace, and add a template sensor that takes the HVAC state to provide the energy usage.
The state combined with the change in rate of heating or cooling can provide early warning of component failure, it usually takes a few hours for a human to notice something like an AC start capacitor failure, but with the right programming I should be able to tell in under 30 minutes and be alerted by automation. I will also be able to get intelligence regarding the duty cycle of the system, and how many start/stops occur, right now I have no idea other than what I can infer from temperature readings.
I plan on collecting large amounts of data; the HVAC state, solar radiation output, internal temp of several rooms, attic temp, and exterior temp. I will then try to develop an algorithm that simulates the heat transfer rate of the house given the environmental conditions, to eventually predict when the AC or furnace should be activated in advance of the thermostat temp exceeding the set threshold, in order to optimize comfort in the rooms most likely to be occupied at the time.
During hot days there can be a several degree delta between occupied room temp and thermostat reading due to house geometry and the fact that the main floor is the upper floor. On cold nights bedrooms can become excessively warm, while the rest of the house is still cold enough that you need a sweater.
The long term goal is to ditch the thermostat and use high precision room temp monitoring instead, and have all the control hardware next to the furnace.
I have done some preliminary bench tests and it should function as expected. If all goes well I can get this installed tonight and start collecting heating data during next weeks cold spell