New hardware model! Hestia Zero - your feedback please

I like the idea a lot and the use case you describe perfectly matches my personal use case. The HestiaPi is just smart enough to run autonomously but my main automation is running on a server somewhere else on my network.

Some additional things I like about using a microcontroller:

  • it’s not as big of a deal to keep up with updates; as a security engineer I’m still uncomfortable having a machine on my network that shouldn’t be updated periodically
  • near instant start/reboot times
  • smaller size
  • less power

The software will, of course, need to be completely rewritten and depending on which ESP model you choose you might run up against storage limits if you have to import too many libraries (MQTT, sensor, etc). Definitely pay attention to that as you choose the model.

We have completed similar hardware combination (libraries for WiFi manager, OLED screen, BME280 and rotary encoder) in other projects and had only managed to reach something like 40% of storage memory. We don’t expect the actual logic for the thermostat to cover that much more space but most definitely, if we decide to go that way, we will make a few prototypes and maybe share the designs here as beta for comments and feedback. Covid situation is not particularly helpful here…

I have some history with ESP/Xbee based devices as a thermostat. I built several over a two year period from 2016-2018. These devices sound great on paper, but in a live setting I found them unusable for a couple of reasons:

  • sensor driver support was pretty limited
  • if you live in a dense area the baud rate past 10-20 feet is an issue due to interference on 900kHz and below bands. You can still transmit/receive data, but the delay can be several seconds.
  • quality oversees manufacturers of ESPs and the like isn’t very good. I’d get 6-9 months out of a device before it die. Buying offical Xbee or particle boards is pricey.
  • wasn’t easy to support complex sensors (I2C, SPI, blutooth, one wire). It’s doable, but a lot of work.

In the end, I switched to the RPi Zero W when it was released. Why deal with ESPs when you can use standard Linux tooling and ecosystem. If cost is the primary reason for this idea, I’d drop the touchscreen and find some off the shelf stuff instead.

Working with the BME280 was pretty easy to be honest and WiFi connectivity has never been an issue. Of course we understand this changes from house to house.
About their life time, we have maybe 10 of them running 24/7 off mains and none has died (yet).
The price was only the result of this change. The main drive was the conclusion that a RaspberryPi is over-engineering for some people’s setup.

I repeat this is not about replacing HestiaPi. This is about offering an additional model for people who only want an extension of their existing smart home setup.

May be an idea…
A thermostats need to be autonomous from other system, and reliable over time.
OpenHab is quite interesting sw but too heavy for this HW, a solution like HestiaPi Touch IOThermostat may be a better solution for the Hestia HW.

You may take this sw and improve it, from my point of view need a better UI interface, functionality are good, and offer as alternative to OpenHab.
An evolution may be integrating it with Apple Homekit, Google Alexa.
To lower the HestiaPI cost you my use a smaller touch display (with luminosity control may be interesting), a display need always

I love the idea of a light version!

I think Nest’s success is based on ease of use.
Especially for the Hardware part where even an old person can easily set the temperature by turning the wheel.
It would be interesting if a similar solution could be found.

I am planning on having three HestiaPi units in my house to replace the three traditional thermostats we originally had. I don’t have a OpenHAB server other than what is on the HestiaPi. I don’t think I’d be interested in the ESP based unit, but I could imagine it being popular among others.

I like:

  • The ability to operate in standalone mode
  • The lower cost

I dislike:

  • The ESP device could never be the center of a smart home, which means having mismatched thermostats
  • The touch screen is really nice (aside from being on all the time and making the hallway glow)
  • The ability to extend this would likely be very limited

I also like that there’s some possibility that it will be easier to use. The HestiaPi has a problem with lag when I’m trying to quickly adjust the temperature where it gets into a state where it just flips back and forth between 22.0C and 21.5C (or whatever transition that it gets stuck at) even after I stop touching it. That’s more of a negative about the HestiaPi than a positive thing about the ESP-based unit, but if it didn’t have this bug, it’d be easier to use.

Count me in for the ESP light model!

Personally, it just makes more sense to use a more robust Pi like the Raspberry Pi 4 for my main hub instead of making my thermostat with a Pi Zero the hub of my smart home.

It is not meant to operate as a hub, quite the opposite. Standalone or as an extension to an existing smarthome system running in another machine and talking MQTT.

get rid of openhab, its far too heavy for the purpose here. i realize it may be a lot of work to get the rules and things all correct but its more clean without it and just use logic to handle mqtt messages.

make it easier to use either built in mqtt for standalone use for people that dont have other home automation but also able to connect to another mqtt server. example i use home assistant and was really surprised at how hard it was to get things talking together.

battery backup would be cool, when ive lost power occasionally its kinda screwed everything up and the device needs a reboot to reconnect and be available again.

that’s the point here.

Yeap. That’s what is described to be done

Hestia Zero should have no important memory that will corrupt in the case of a power cut. Things like network credentials, timezone, last state and settings should be kept elsewhere that will not block the device from starting and working if they get corrupted. We are considering the storage options…

1 Like

This sounds exactly like what I’m after right now! Would be ace to have a thermostat that I can operate locally and also link up to my HomeAssistant instance to get all the smart functionality too. I can then have smart temperature sensors/TRVs in various rooms for zone control but keep everything on my local network.

1 Like

Have been working on this idea the last days and we noticed that in order to minimise components/complexity/costs/casing requirements we can use ESP32 that has touch sensor functionality integrated. As we are considering making the case out of FR4 material (what PCBs are made of) in some nice colour, we can integrate some touch pads on the outside or the inside of the front cover at almost no extra cost… Off to some prototypes…

This may be interesting for you:

SBS’ Cricket ESP32 Packs Espressif’s Popular SoC Into a Raspberry Pi Zero Form Factor

To test the idea simply.

Thank you but I’m not sure I see your point. Having both a Pi Zero and an ESP32 wouldn’t help during prototyping/development unless I am missing something.

This board have the same form factor of the PI, so may be possible to use it instead of PI on the same Hestia board.
My experience with Hestia is that PI warm too much and the temperature sensor do not read lower temperature than 20° about ( I have try to position the sensore in a lot position, and more). My solution was to use iothermostater sw adding an external esp32 board with temperature sensore, the same of Hestia, connected to Hestia using MQTT protocol to get the house temperature. So may be a good idea to substitute the PI with a esp32 board who warm less and put the sensor in the same box.

I have read 16’C (about 60’F) with no problem. Are you using the ribbon cable provided and placed the sensor inside that triangular with the sensor facing out? Has anyone else had issues with low temps?
Apart from that ESP32 can be much smaller which is something we are after as making the whole case smaller is a plus. The PCB will be a new design as many of the components can be different (PSU, relays, LCD)

1 Like

I observe consistently 2C higher when inside the bottom triangle with the sensor facing out compared to hanging beside the wall on the cable. This is compared to my commercial programmable thermostat on the same wall. They match when out, 2C higher when in. There was some discussion of this elsewhere on the forum too. This isn’t just when the pi is working hard, but all the time (for example at 2% cpu load and 36C CPU temp.) I’ve added a Samsung S4 as an “external sensor” in the same location, also 2C lower.

1 Like

This topic was automatically closed 91 days after the last reply. New replies are no longer allowed.