I'm looking for a thermostat that I can control over my local network. Everything I'm finding is either networked together, not on my LAN, or only controllable via the vendor website.

Anyone have any experience with such a thing?

I look for the exact same product every winter. If I could fine something less than $200 I'd have it installed already.

Matthew

Mmm.. a "thermostat" is usually just a temperature sensor, and a relay for the furnace or air conditioner.

Both of those are simple devices, easily interfaced over USB to a computer.

At the Linux Symposium this year (mid-July), I'm giving a tutorial on exactly this kind of thing: how to control the real world over USB from a PC, for software geeks with little hardware knowledge.

See you there?

I probably will have notes, photos, and circuits for this stuff on my website at some point after the Symposium. It should be easy enough to pick up from that, if you can't make it in person.

Cheers

Right you are, of course. However, more and more thermostats are fancy and programmable and such. The one on my wall is currently failing, however, and likes to pretend it's been reset about once a week. This results in the AC not kicking in until it's about 85 degrees inside and, if JM hasn't noticed that it's getting warm before that point, constant running of the AC to get it back down during the hottest part of the day.

I end up having to re-program the thermostat just as often as she gets to experience its failings. Since it's time for an upgrade I figured I'd look for (what I consider) the next in the natural evolution of such devices.

$3 PIC + $1 sensor + $2 relay + 5V PSU = do it yourself "smart" thermostat, with serial connection for optional PC monitoring/control.

Or blow $100+ on a "real" one. But what fun is that?

Forgetting to program in a hysteresis and wearing out your AC in a few weeks: priceless.

I have one of these incorporated into my home automation network (along with a half dozen light switches - nothing fancy).

I can set it remotely, including from my blackberry - usually in the airport before I take off or just before I hop in the car to head home.

My only real complaint is that it loses time when the power goes out (which it does now and again here; damned trees).

-jk

http://www.proliphix.com/imt350.htm

Pricey, with some other models that are slightly less so.

Holy crap these things are overpriced. First off, why does it need a full color display? For that matter, it barely needs any display at all. And, then, all it's really doing is flipping some relays on and off. That's about $5 worth of parts. Maybe.

I was kind of poking fun at Mark, but, damn. Clearly some of us need to get into the business of making networkable thermostats, undercut everyone currently on the market by half, and still be making a 1000% profit.

In this type of business, the (product) costs are all in the software. BOM is many times negligible compared to the retail price of the product(s).

I'm pretty sure that the market for a network thermostat at this stage is very limited, which is another reason for a fat margin.

There are a lot of other avenues for low BOM-cost products with high margins and quite large market potential. Can't get past the need to spend time/money on software though, because at the end of the day, that's what's going to make or break the product (and differentiate it from the competition).

The software should be nigh trivial; certainly not worth hundreds of dollars.

I completely agree. Certainly the empeg can't be worth that much more then the Neo 35 just due to the software, and slight build quality differences.

Oh, wait, sorry. Had a 10 year flashback there :-)

Are you sure you're not in marketing?

There are possibly some significant hardware startup costs as well, but software is never trivial, especially when you're not copying something already out on the market.

Nothing today is the sum of its parts, not a car, not an elevator, not a WiFi thermostat. Yes, the EcoBee is priced quite high. It's obvious they're not targeting the average consumer at that price point.

I think, but could be wrong, the Proliphix ones are all maintained through their website. That could just be their cheaper models, though.

It can be had for slightly less here.

For what it adds, that price seems to be reasonable when compared to the rest of the market. Still a lot of money.

Or rather, the other way round: the price point is that high because the average consumer wouldn't buy one. It really doesn't take much software at all to be "worth hundreds of dollars" if only about six of you end up buying the product.

Peter

I didn't mean software in general; I meant software for that particular device. There's a huge difference between a device that processes multiple kinds of audio files, including some DSP, accepts RF signals from distant broadcasters, stores multiple gigabytes of data locally, provides an optimized interface to that data, etc., and one that flips a relay based on the input of one sensor.

I still think your underestimating the complexity of the software in that product then. It's not just sitting there watching one sensor to flip a relay (ie, replacing a simple mercury switch), but also sitting on a TCP/IP based network wirelessly, accepting touch input, allowing remote access, being able to self update, and so on. I could easily see the software costs for that product being over a million, including design and such. Sure, some of this comes along with an embedded OS, but it's still going to take a few well paid engineers and designers a bit of time to crank the needed software out. Really well paid ones if you want it highly optimized for an embedded system to use a low power CPU.

The device is clearly in the sales range of thousands at this point, so cost per unit on the software is going to be high.

The product isn't as simple as you're making it out to be. It's not the empeg, but today making an MP3 player is also not what it was back in the late 90's yet those can still be bought in the range of $20 to $20000.

I wouldn't be surprised if it cost upwards of a million to develop the EcoBee. That's being quite conservative too, IMO.

Don't forget the scheduling stuff.

They've got demos of the interface on their site. Looks like it's fairly capable. I'd much rather pay for something to do all of that than make one.

My opinion is that the majority of that is trivial, besides the UI, and the UI is unnecessary.

How much would it cost to hire you for this trivial amount of work? I'd love to compete in this (albeit currently small) market.

I'm not good at embedded stuff, but I imagine it would be something along the lines of 200 hours of work for someone competent, at, say, $100 an hour. I would hope that they're selling at least 1000 of them. That's $20 per unit for software costs.

Again, this is without the overcomplicated UI on the front. My opinion is that if you're selling to an admittedly niche market that's looking for a network controllable thermostat, there's no need to have any advanced UI on the unit itself. Maybe a simple override controller, but that's it.

If you don't have the fancy UI on the front, you're going to want a fancy UI on a web interface.

The niche market these products are selling to definitely want the nice UI. These things are going to get installed likely as part of a multi-thousand dollar home automation system. When someone's dumping that kind of coin they expect a certain level of polish.

I wouldn't tackle the task of competing unless I was sure I could architect something superior (which I'm confident I can) *AND* deliver it for a price where I could make decent money with decent volume. It's this execution bit where dreams meet hard reality and hard finances.

Thermostats aren't something I've seriously looked at, but I can see at least 1000 hours for a team of 3 to 5. I honestly think 200 hours is somewhat of a pipe dream for a marketable product. Which is both unfortunate, as I'd jump on it if it were, and fortunate, which means that not every other Tom Dick nor Harry will jump on it either.

I can't see having the thermostat being connected to the internet saving much energy over just a seven day programmable one. It would be easier to program probably but how often do you do that. What are you going to do tell it to not change the temp for another hour because you are late that wouldn't save much.

I'm sure the trendy website adds to the price too

Well, presumably you get more out of it than just scheduling. One would hope it would keep a log of when it kicked in and for how long and at which temperature it kicked in and at which temperature it turned off. Among other things.

There's a ton of things you would be able to do with a more advanced/versatile HVAC system controller - "thermostat" is really the wrong word for what such a product could be. Fan scheduling, not supported on any off-the-shelf 7-day unit I've seen, integrated HRV control, humidifier control & monitoring (including outdoor temperature), control of multiple systems from one UI, multiple monitoring/control locations, etc...

Less expensive still not really cheap.

Actually this one is only ~2x the cost of some expensive programmable thermostats.