Not sure about IP thermostats, but if you're willing to get a USB stick to interface with the network, a thermostat that speaks Z-Wave might be your best bet. This one seems highly rated on Amazon[0], but there are lots to choose from.

From there I'd recommend using something like Home Assistant[1] - it's an open source project that provides a UI for a ton of different smart home gear. I use it at home to control some smart lights and automate things based on whether I'm home or not. (There's also OpenHAB, but in my experience configuring that is much more involved.)

[0] https://www.amazon.com/GoControl-Thermostat-Z-Wave-Battery-P... [1] https://home-assistant.io/

The problem is there isn't really much of a market for something like that. Sure, you can have detailed documentation on such a product that explains how it's functioning, why it might drop out when your ip changes, etc., but who is going to understand that? People like you and me, yes, but we're a pretty small market.

In order for there to be a reasonable audience for a product, you have to target the least knowledgeable person - or at least someone pretty far south of the median. Those people aren't going to understand what's happening, and they're going to get upset each time their ip resets and they can't connect to their thermostat anymore. They're going to take the frustration to the thermostat company, they're not going to understand what the service rep tells them, and they're going to return the product, leave a bad review, and buy a nest.

No one wants to be in the position of making and selling such a product. That's why you pretty much need to make your own with an esp8266. And that's why so many of these IoT devices end up being far too bloated.

> The problem is there isn't really much of a market for something like that. Sure, you can have detailed documentation on such a product that explains how it's functioning, why it might drop out when your ip changes, etc., but who is going to understand that? People like you and me, yes, but we're a pretty small market.

I understand why this is not available as a standalone product. The window is probably too small to warrant it. I don’t understand why no company appears to offer it as an option to a product that otherwise “just works.”

Sure, it would require additional investment on the part of the manufacturer to include an “expert-mode” toggle that lets the user make their own decisions, but a market exists for people who want that. Money is being left on the table.

>> Money is being left on the table.

It sure is -- I'd buy that immediately.

The nice part about it is, if their "cloud" services actually added value, I'd be happy to sign up for that too.

I just want a direct access and a fall-back if their cloud goes offline (and I read about that happening in the reviews of every single one).

And sure, we're a subset of the market, but I'd expect a sufficiently large subset to make it worthwhile.

If it was a really good unit with high-quality sensors, WiFi connectors and other components, I'd pay more than for a Nest.

If what you said were true, there'd be no routers which claimed to support DD-WRT out of the box as a marketing point.

The market for people who want to control their thermostat directly is exactly the same. And the best part? The manufacturer can support that AND have their 3rd-party cloud shebang going on, these are not mutually exclusive.

> The problem is there isn't really much of a market for something like that.

These days, the internet gives you access to several billion people. Even a very niche market can translate into a large customer base and a nice, profitable business.

Is there a way to bundle a dynamic DNS offering that would make the changing of an IP address invisible to the home user. It's 2018, so I'm wondering if setting this up has become easier than it has in the past. Just trying to think of something to keep the idea viable.

If you check out Home Assistant's climate control category, you can look for devices with IoT class "local polling" and "local push".

There are a number of options there.

Scroll to the bottom of this page for the listing:

https://home-assistant.io/components/climate/

looks quite interesting - thanks!

It's relatively easy to build one yourself. I did it, it works great.

I used 2 ESP8266's - one in the temperature sensing and control box in my office, and one controlling a relay inside the boiler. When it's too hot the relay is opened to switch the boiler off, when it's too cold the relay is closed to switch the boiler on.

The question is, do you trust your own handiwork to the extent you're not worried it'll malfunction and burn your house down, or freeze your pipes during winter? Personally, I'm reluctant to have electronics I built myself connected to mains and running unsupervised for longer periods of time.

Well, some parts of it are a lot more likely to fail-deadly than others. A commercial power supply that outputs 12V or whatever your servo wants basically can't burn anything.

As for bugs in the control code freezing your pipes? Well, just give the code as much care as you give to driving and you should be fine. ;)

(The serious solution to that is to have a dead simple failsafe controller that switches between on, off and auto, where auto is controlled by your crazy inventions.)

For projects like that, you don't connect anything to the mains at all. Specifically:

- At least in US, thermostats control is low-voltage (24 V AC), and come from low-power transformer which will just stop working on overload (ask me how I know :) ), so there is no electrical fire danger

- You never plug anything into the AC outlet directly -- you always use AC/DC adapters (and not the ones bought on ebay for minimum price, too)

- To prevent freeze hazard, you also leave original thermostat hooked up, but set it to really low temperature (10C/50F).

- Overheating house is possible, and you definitely need to consider this in your software (for example, require periodic control messages to keep your boiler on). However, I do not think it is an actual fire hazard -- likely, there are hardware safeties in your heater to prevent that. Still, if you are worried about this, you can install a second thermostat in series with your relay and set it to max temperature you ever want.

That's a very interesting point, I built my own system, but I don't trust my own handiwork enough, so I went the middle route: I got a Siemens RDF302 thermostat (which is a dumb thermostat with a Modbus RTU interface for remote control) and then I use a raspberry pi to change the setpoint temperature according to a remote signal or schedule. The actual decision to turn on/off the boiler however, is still left to a "real" thermostat made by a professional company, and a malfunction of the automated system is likely to result in a failure to change the setpoint, not in infinite heating.

Why don't you just turn the damn thing up when you get home?

I agree. It takes about 10 min to warm up the house, no problem.

Remote access is so I can turn it down remotely because I forgot to when I left :-)

That depends on the kind of heating you have on the house. Air vent heating takes 10 minutes. Radiators take some 40 minutes. Underfloor heating takes about 90 minutes.

Remote control does make sense for some, not all cases.

I'm actually looking for one for my shop, which has a longer heat-up time

For home, I agree that the programmable ones are ok, although they fall down when we travel for more than a week -- we have to either run it an unnecessary day, or arrive to an extra-cold house, so if there was a good one, it'd still be handy there.