If I only have 0.35 VA at hand, is my project doomed to fail?
I have some old temperature control systems for my radiators called iqtherm. In short they have a Triac module providing an interface towards an old control unit. The interface towards the control unit consists of 12 V isolated AC power from a BV 202 0167 (400/12V, 0,35 VA) that controls the Triac via an optocoupler (MOC 3063).
I have successfully built a new control unit based on ESP8266 but externally powered by an USB socket.
The radiators are based on 2-phase 400V without neutral so i haven’t found any of-the-shelf cheap AC/DC isolated power supplies and I’m trying to stay away from doing modifications on the mains power
You will need to determine how efficiently you can convert 12V ac power to either 5 or 3.3V DC.
You are properly better off driving your module with 3.3V DC generated by a DC/DC converter assuming you can find one that’s clean enough because if you go via 5V DC you’ll loose a lot of power in the LDO between your USB connector and your ESP8266.
My guess is that in a best case scenario you’ll have no more than ~100 [email protected]. This should be enough for RX for the ESP8266 but not for TX at typical TX level so you’ll have to experiment with how low you can set the TX Power and whether this brings current low enough and whether that TX power is enough to maintain a link with your access point.
Of course - if you have additional circuit on your ESP module the current that this circuit uses has to be accounted for as well.
ESP will consume up to 250mA at 3.3V so that’s little over .8 watts. Your supply can give you under ideal 100% efficiency conditions only about .35 watts, so no, probably not — the supply plus buck converter needed will falter on you. Sorry.
If you can put the ESP to run for short times but spend most of its time asleep, and you have plenty of physical space, you might be able to manage with a normal bridge rectifier circuit on the 12VAC but with an overkill amount of smoothing capacitance to run the DC-DC converter whilst it’s awake then recharge whilst it’s sleeping.
You’ll need to limit the inrush current into the capacitors with series resistors on the transformer side.
Guess the wifi scan/connect/transmit will cause to much power draw. Might work around it with insane amounts of capacitors or by placing a poor man’s UPS (power bank that can be charged and discharged at the same time).
Another idea is to disable wifi completely (just recently introduced in esphome if I remember) and try something more power conservative to transmit the data necessary like esp-now
ESPNow same power consumption as Wi-Fi just for shorter periods, so it will still get self-K.O.'d from the power demands. Same w/ BTHome, same chip synthesizing the signal, same antenna, and same power budget.
OP just tap into that sweet sweet 400V and attach a transformer in parallel. Please pay an electrician to do this.
Do you have any mileage on this topic? While it looks like esp-now still has one (compared to many with WPAx-WIFI) high TX spike at over 400mA with the default tx of 20dB the overall power consumption should be drastically decreased including much lower and much less (one? ) spikes.
In general esp-now is rather comparable to zigbee (same frequency) and has rather little in common with WPAx-WIFI
This scenario could make @erapade setup maybe fit to use with some extra capacitors (or batteries) so that this one spike can be supplied without crashing the esp. Also some quality power supplies actually often supply more then they claim - specially for short amounts of time - so it might just work
Thanks for all the answers where some has been giving hope and some haven’t. Yes it’s the communication that will cause the problem but I think I will give it a try at least.
But I didn’t fully understand if ESP-Now also requires a GW/Bridge or if HA now in some magic way also can communicate on the ESP-Now protocol.
For the ac/dc part I have been locking at this design for a while but as it’s not isolated and I would like my temperature sensor to be wired and remote from the radiator I have abandoned it.
But now I might consider it again and instead hook it up to the isolated 0.35 VA transformer. I will hopefully only have the main power losses in the rectifier bridge but we will see how much power I will get out of it.
Any other suggestions of cheep ~12V AC to 3.3 V DC regulator designs suitable for an ESP. As the transformer is already isolated the AC/DC regulator doesn’t need to be isolated
I also see an alternative for my design and that would be to give up powering the ESP on an isolated power source and instead powering it with the none isolated design above directly from the 400V mains and keeping it inside the Triac unit and instead relay on the Xiaomi BT sensors but I rather stay away from the mains.
Anyone having experiences from powering an ESP using this simple AC/DC design above?
Why not - They actually can be powered with a 0.35W psu I started to hard wire mine because changing one-use coin cell every year s*cks and is a lot of trash
Note that BTHome and ESPNow are connectionless and the device cannot know (and therefore cannot retry) if the message did make it to it’s destination. Furthermore, you will not be able to control your device from HA using the API if you do not continuously run Wi-Fi reception on your control node. These are great protocols but probably not suitable for thermostat or boiler control. You do not want your “turn boiler off” message to be a SUGGESTION, rather it must be a command.
Thanks for all the insight.
I did a load test of the transformer and it acts as a power source with in internal resistant of approximate 148 ohm that saturates at ~0.5 W at a pure resistive load, and this happens when the voltage has decreased to ~8 Volt
So it looks like I could get enough power for having WiFi on but not transmitting. With a large capacitor at 4700uF I would be able to deliver 3.3V @ 400mA during the 40 ms upstart phase. But… It would take more than 30 seconds to load the capacitor with energy so I would need a delayed start as well but that could maybe be a WiFi only delay done with SW.
Yes I would be able to decrease the transmitting power and then this looks better.
Let’s say I manage to build an efficient AC/DC that will not overload the power source (since it will saturate at 60 mA) and that will support a start-up scenario as well. Then it will depend on the transmitting versus none transmitting ratio.
And I forgot to say, the optocoupler I shall control will require another 5-8 mA.
This is still a funny project and I can see a bit of a chance that it will succeed but my previous electronic skills are fading away
) spikes. 
I started to hard wire mine because changing one-use coin cell every year s*cks and is a lot of trash 
