Hard to believe the amount of time I’ve spent on this thermostat and the backup unit. I’m still too deeply buried in R&D right now to fully develop another full article but here is some information to add to the previous blog entry on the subject. My Node-Red/ESP-GO thermostat is now in operation with two stat units, originally to handle any issues whereby one locks up occasionally, but with recent improvements that simply does not happen.
The 2.5w laser engraver turned out to be less of a blessing than it first seemed, thanks to it’s inability to handle clear Perspex (does a lovely job on coloured material) I initially had to go back to traditional DIY techniques for this project but see the update at the very end.
Some visuals – here is my Grafana logging – accessible externally thanks to PiVPN as is the stat itself.
Above is my Grafana info screen – monitoring and logging the two stats (main and aux which have now survived power cycling both for the stats and Node-Red without failure or needing intervention). I’m also monitoring (temperature in the hot water tank cupboard from the mains controller supplying on-off instructions to the air—source heating system.
The main (green) sensor is currently at 22c (air source takes ages to reach full set temperature) and that is the sensor in use (aux in yellow is a fall-back). Humidity is for reference only – ultimately I’ll factor that into the heat supply as it generally FEELS a little warmer when the air is dry.
Here below is the stat as it appears online (only via PIVPN, deliberately). The TOP green readout is high resolution but I’m only actually using integer resolution – why bother being SO accurate when the difference in temperatures is well over 1 degree from 1 metre above ground to 1.5m and from one side of the room to the other. There IS however a use for the more accurate reading, as turning the heat off when the stat reaches set temperature means it will only ever stay AT the set temperature for a few moments.
And yes, the temperature does need to rise quite early in the morning – on really cold mornings there is a lot of lag with air source heating before radiators get up to full temperature. As you can see, the Nod-Red flow is getting way too big for one screen. Most likely now unreadable even on expanding but hopefully it will give you some idea.
These stats are deliberately simplified allowing limited up-down temporary temperature change and monitoring of local temperature and humidity as well as SET temperature monitoring and heating status with WIFI signal RSSI thrown in for good measure.
The actual physical stats are starting to look ok, would be better if I had a decent engraver. Below, stat4 on test minus it’s up-down buttons on my bench, working on USB and with it’s plastic screen protector still in place. Stat3 is boxed, running on its own internal mains power supply and in the house, getting the spouse-test. 2 days without issues.
And here is one of the first “boxed” units… in my office which is always warmer than the house and therefore the worst place to test stats… this one now has buttons that light up when pressed. I just wish the buttons had an overhang.
April 2020 – I’ve just been putting the finishing touches on a new touch-switch version of the stat, still using ESP-GO. I had planned to do this with Tasmota but it seems that the ILI9341 installation in Tasmota isn’t ideal for the job. The ESP-GO implementation on the other hand is ideal for this application.