Nest - Data-free Badness

I have two Nests installed in my house for my upstairs and downstairs HVAC systems for some time now (since 2012-02-26), but I haven't written about it since I made my first post with my initial thoughts after installation.  I thought I'd share my current thoughts on the product.

What's Awesome (in no particular order):

• Super easy installation.  Best in class instructions.  Covered more in my original post.
• Ability to control my thermostat remotely via an iPhone app.  Cool stuff.  My kids (who live on the upper level) and I have a bit of a war with setting the temperature during the summer, but I always win.
• Auto-Away: it learns when you are not around and automatically cuts back on your energy usage.
• Monthly reports: Nest will email you a short summary of your energy usage and 
• Beautiful design: it looks good and its easy to use.

What's (still) Not Awesome

• I still can't get my data.  It turns out that I could have if I was willing to go through some gymnastics (see Gregory Booma's blog post - he provides a script to import the data into R, one of my favorite tools).  Unfortunately, Gregory has updated this as of 2013-12-03 saying that due to Nest introducing their API, the functionality described in his post is not available any longer.  I'm inclined to sign up as a developer, but would rather have intermittent access to all of my historical data, I don't want to have to set up a server to capture my data.  This was a $250 thermostat.  I think that they could give me the ability to download a CSV every so often.
• Given that I can't look at my data myself, the app and website still seem very underdeveloped.  You just can't look at much.  10 days of history is it.  That is pitiful.  See the graphic below.  Its crazy how much more my system ran last night (almost non-stop) when the temperature fell to 10ºF.  Sadly, that's a lot of propane.
• But what was running?  Was it just the fan running (I have the fan on the 15 minutes an hour schedule for greater comfort), or was the furnace chugging away burning propane?  Really guys, you couldn't figure out a way to represent this on the same graph?
• Why are graphics covered up to the point I have no idea what they are?
• Why can't I see a sparkline of the temperature in the house?  The humidity?  This seems stupidly obvious that I'd want to be able to see the history here.
• I can't connect the Nest to supplemental temperature sensors.  I'd love to have between six and 12 wifi (or Z-wave) sensors reporting to the Nest and be able to set up rules such as:
• Run heating (cooling) if any of the sensors gets below (above) a defined set point.
• Warning if the temperature gradient is higher than a defined level (e.g., 10ºF).  Help me trouble shooting by showing me the variation of all of the sensor over time.  I'd love to be able to tweak my registers in a way that limits variation and increases comfort.
• Allow me to activate register boosters (my term): basically, registers with supplemental fans to increase airflow to particular rooms or sections of the house. 

Will Z-wave Survive?

Introduction

I recently got an email from a friend asking for my thoughts:

Do you think zwave will die with the onslaught of wifi?
http://uncrate.com/stuff/wemo-light-switch/

The short answer is eventually, yes.  The long answer is below.

Features

The link above links to the Belkin page with more information about the WeMo Light Switch.  It looks like a pretty standard switch with a decorator style.  It promised to make your life easier by being able to:

• Turn on/off your lights via app
• Turn on/off your lights via a schedule (fixed times or sunrise/sunset)
• Doesn't lose the schedule during power outages
• Has no batteries to wear out (not sure how this works with the last one - a super capacitor?)
• No minimum power levels to allow it to work with all lighting (and ceiling fans)

It looks like an interesting device that could make people's lives easier.  As long as they are handy enough to replace a light switch.  I'm not sure that we are quite to onslaught for wifi-based home automation tools, but we are probably not that far off.  An interesting question to ponder and develop a framework around more generally.

Limitations

From the Belkin web site:

Please note that the WeMo Light Switch is not compatible with lights controlled by more than one switch or with metal face plates. Metal face plates interfere with Wi-Fi® signals and will not allow this product to work correctly in your home.

I think the first part is important, the second part most people can live with.  However, let's say this a different way:

• Neutral Wire Required
• Single pole (Not compatible with 3-way (multi-location control) switches)

Both are meaningful and more limitations of the product and not the technology.  I won't pursue these in my comparison below, because I think that both are easily overcome.

Specifics of Z-wave vs. Wifi

Z-wave has been around for some time, but is still a technology of which most people are not aware.  From what I can tell from the childishly designed Z-wave website, Z-wave has been around for about a decade.  It is described as:

... a next-generation wireless ecosystem that lets all your home electronics talk to each other, and to you, via remote control. It uses simple, reliable, low-power radio waves that easily travel through walls, floors and cabinets. Z-Wave control can be added to almost any electronic device in your house, even devices that you wouldn't ordinarily think of as "intelligent," such as appliances, window shades, thermostats and home lighting.

Generally speaking, Z-wave works with a master controller that sends commands to the devices over (a mesh-network of) radio waves.  The controller can have a good or bad interface for controlling the device.  My experience has been with the MiCasa Verde Vera (I have an older version than the 3), which has some frustrating limitations but is otherwise reasonably reliable and a well-thought out product.  My use has primarily focused on lighting control, for which I have about 12 different switches connected to it.  Most of them are on a dusk-to-dawn schedule for security purposes.  I also have a DIY garage door opener / sensor set up.  My pool controller (an Intermatic PE953) is theoretically connected to my Z-wave network, but won't respond to signals in its current location (the mesh is failing me!)

So what are the differences?

Characteristic	Z-wave	Wifi
Antenna/Controller Cost	Low	Relatively low and declining, given the wide-scale adoption and many embedded wifi applications
Bandwidth	Low	High (with growing potential)
Current Adoption	Limited to home automation applications (very low)	Widespread - most houses have wifi available
Reliability	High, low data rate and mesh networking approach should result in robust network.  However, depends on controller functioning.	High in places with network coverage (though many houses do not necessarily have complete coverage)
Control	Central - requires a dedicated server or box to run the system	Disbursed.  Each piece of equipment has to have the intelligence to control itself.
Available control items	Generally wide: - Switch controllers (2, 3, 4 and no neutral available widely today) - Plug-in modules - Thermostat control - Door locks - Sensors - Pool controller - Security (cams, etc.) not available given high bandwidth requirements	Wide in some categories: - Nest (and others) for HVAC - WeMo for lighting (limited to 2-way with neutral for now) - Others for plug-in modules - Door locks (small companies) - Security-based webcams (common) - Electricity usage monitoring (think TED) - Many new, remotely electronic items are coming with some form of networking built in.
Integration	Modest to high	Low - each system stands on its own
Ease of use	Modest	High for the limited single application uses

When I first started looking at this, my initial impression was that $50 was way too much for an individual light switch (the Belkin WeMo has an MSRP of $49).  Then I searched for the z-wave switches that I've used in my house (GE-branded Jasco switches) and saw that they are $42.50 each on Amazon.  Not terribly different.  I remember having bought most of mine on a good sale at SmartHome, but they were in the mid-$30s.  Of the two switches, my guess is that the wifi one has the greatest potential to get much cheaper than the Z-wave switch due to economies of scale.

Other Technologies

Zigbee is out there, but is stupidly non-standard (i.e., high integration costs).  This will be in single niche products or will serve as just the last 15 feet of control (think Philips Hue) and wifi will be useful for controlling before there.

General Considerations for Adoption of New Technology

• Cost: Cheaper is good.  The potential to become crazy cheaper is even better.  Should be considered on a total system basis.
• Current Embedded Base of Related Products: Tends to reinforce the cost point.  But also should incorporate what people know about the technology and how to implement it (both the user and the designer).
• Complexity: Highly complex (to the user) solutions generally fail.
• Reliability: In the real (not virtual) world, users do not accept high rates of failure.  (I'm not sure how high, high is in this case.  Is a failure rate of 1% of commands to devices too high?)
• Product Features: Ensure that the product solves the real problem that the user has, not the problem that the technology is capable of solving.

What is Currently Missing In Home Automation?

[DRAFT section]
Home Entertainment
HVAC/Hot Water
Lighting/Shades
Physical Access Control
Spa/Pool Equipment Control
- Chemicals not there
Energy Usage Monitoring

Conclusions / Forecasts

I guess its a bit of a cop out to suggest that its still anybody's game out there right now.  Belkin is a big name and has a fair amount of brand power behind it, they could (potentially alone) shift the playing field.  Overall, though I don't think Belkin will be a dominant player here.  My guess is the following:

• Wifi wins:  Wifi will eventually come to dominate the controlled/connected home; wifi networking is ubiquitous and the hardware has gotten very inexpensive.  That said, I think that some support for z-wave will exist for a very long time.
• One at a time: Most homeowners will take a very piecemeal approach to home automation.  A few light switches here or there.  Simple schedules will dominate.  The next five years will be mostly DIY, making even three-way switch replacement a bit tough.
• Physical/Virtual Interface Design:  Apps will be useful for setup, but people like to be able to turn on and off a light without having to pull out their iPhone.  Expect to see relatively small number of "central load control" systems.  We will still have light switches in our rooms.
• Integration will matter: That said, how people interact with the disparate systems will become a bigger and bigger issue as they add more smart items to that home.
• There will be the parallel growth of a home controller that can bridge wifi, z-wave, x10 and all other previous technologies.  Think something like the MiCasa Verde.  
• For these boxes to work, they will have to plug into IFTTT, which I think has the potential for changing how users interact with their digital and real worlds.
• High End will Still Exist: Proprietary systems will continue to reign on the high-end (Control4, Creston, AMX) where systems implementation are a high portion of the system cost.
• These should die out or just serve a smaller and smaller niche as improvements elsewhere in the ecosystem are made.
• Security Matters, Eventually: Maybe after some noteworthy hacking event, security will become a big deal for home automation.
• Where the Control Sits:  There will be some diversity in where the control exists.  Should it exist in the switch or in the device (i.e., lightbulb)?  Obviously as it gets cheaper, the end device gets more and more achievable and in the end, mostly more flexible.  Mostly, because you will note that no matter what control signals are sent to my Hue lightbulbs, they won't turn on if the switch is in the off mode.

End Note: Anybody have a really good WYSIWYG HTML editor for the Mac that handles tables as well as Word?  Blogger has nothing.  Tried SeaMonkey and the results are above - which I'm not terribly happy with.  I've been using Confluence at work a lot and find it to be very fit for its purpose - especially for tables.

Robots are taking over?

[Under Construction - but I've run out of steam - will get back to it when I get more - steam]

Recently, Wired published a story about robots taking over the workforce. I really don't see that happening quickly, though I would like to capture my thoughts about things that I would like to automate for me.

One thought at the start is that I think these tasks that I'm seeking to be done are all part of normal life.  Everybody (more or less) has to tackle them.  I also consider all of them, "work".  As in, they are things that we have to do, but don't seem particularly enjoyable.

The tasks are also repetitive and thus warrant some amount of specialization and I talk about each of them in that context.  If these specialized 'bots will exist in the next 5 or 10 years, I don't see the generalized 'bots coming for another decade.  But I'm probably optimistic on both fronts by about a decade.  This goes back to the adage about forecasting: we tend to over-forecast short-term change and under-forecast long term change.

Here are my separate categories.  I'm definitely open to additions from anybody out there that stumbles on this post.

1. Driverless Car: Yes, I would like to sit in the backseat.  And have it automatically valet my vehicle (that includes paying the garage fees, if any).  I don't know what I'm going to do that is productive when I'm commuting in the back seat, but it would still be awesome.  Long trips would be even better.  This sure wasn't me when I was younger and saw driving as a path to freedom.  Now, it's just a chore.
2. Home Organization/Cleaning: a broad category, yes.  Here are the subtopics:
1. Pantry: Combination of inventory management and organization.  Ideal situation would be that it would stock, reorganize and order more whenever a stock-out occurs.  For this to happen, the device would have to be able to continuously scan present items (I wouldn't restrict access to the pantry to the robot) and know how to order more.  This wouldn't be literally limited to a pantry, but over-counter cabinets, as well.  Lots of flexibility necessary for this one.  Would need the flexibility to recognize items in the pantry that it didn't put in there (even asking the owner what an item is if it doesn't recognize it).  Should also be able to respond to fetch commands.
2. Refrigerator:  Same as pantry, except with refrigerated items.  Should also clean/disinfect regularly.
3. Closet/Laundry: Collect, load, clean, unload, dry, collect, fold, stack and arrange.  Super complex and varied set of skills.
4. Kitchen: Clean counters, floors, tables.  Load dishwasher.  Dispose of food.
5. Bathroom: Mostly looking for cleaning skills here. Who wants to have to clean their own toilet?  Floor, shower and sink cleaning would be nice additions.
6. Trash/Recycle-ables: Remove trash from all trash receptacles around the house (but primarily the kitchen), put into a larger trash bin outside and replace the bag.  For bonus points, it should take the garbage cans to the street at the right time(s) of the week.  I think that the receptacles would have to be redesigned to make this an easy activity for a machine.
3. Cooking: Not in a huge hurry for this one.  I don't mind cooking (what I do of it, mostly grilling) and my wife likes to cook.  Its the cleaning up, mentioned above, that we really don't get any satisfaction out of.
4. Yard Maintenance:
1. Lawn: Yep - something like this.  There was an article recently (in the Financial Times, I think, but I couldn't find it when I looked) about lawn robots that suggested a couple of interesting things: (a) Europe is in the lead because lawn services are more expensive due to higher labor costs and (b) when people have reliable systems they are often used to cut the lawn daily instead of weekly.  They machines are battery based, and will cut within a defined geography without any human intervention.
2. Pruning/Bush Management: Complex.  Trim my bushes.  But it would be awesome.  Tree pruning is a reasonably infrequent and probably very difficult due to the heights.  
3. Fertilization/Watering: This could be a simple means of depositing some fertilizer on the tail end of the lawn robot's activities.  Simple task.  More interesting would be for the system to know how much water and fertilizer to deposit based on an array of sensors around the yard.  
4. Weed Control: How cool would it be if you were able to plant what you wanted and then not have to worry about weeding through the entire growing season?  If we can mow lawn, presumably we could get a smaller device that would work around delicate flowers or herbs or vegetable garden.  If it could also patrol the garden for deer and rabbits, that would be pretty kick ass.
5. Debris Management:  We get a lot of sticks and branches that fall from mature trees onto our yard.  I'd love our lawn to be thoroughly scoured of this after every big storm or at least every couple of weeks.  Some of this happens now just from a large lawn mower blade chopping up small branches.  Pretty sure that the lawn 'bots aren't  to be quite this big (at present).  When they are, this is probably just an infrequent activity that I still have to do in person.
5. Pool Maintenance: So, we already have some form of "pool robot" that we use to clean the bottom of the pool.  Our Polaris has no intelligence but is still useful.  I'd like something that knows when its done and turns itself off (potentially climbs out of the pool and then turns off the auxiliary pump).  Some better designed pool equipment would be important, too.  Here are a couple of thoughts: (a) Pool filters should automatically back flush when they need to be (I can't believe that this isn't the case already) and (b) skimmer baskets should be designed to automatically empty.  I don't see an easy way for this to work in their current configuration, but that doesn't mean that it shouldn't be changed.

Automation of a Garage Door

I was inspired by this video that I saw on YouTube a while back to both automate my garage door opener and to also install a position sensor. I finally got off my butt today to get the project done. All of the purchases had been previously made, and I was pretty sure that I had the appropriate tools.

Equipment List:

• Already installed garage door opener
• Radio Shack Relay (125VAC/10A DPDT Plug-In Relay)
• Radio Shack Quick Crimp-On Disconnects (found here)
• GE 45604 Z-Wave Technology Outdoor Module for Lighting Control (at Amazon)
• Two prong extension chord (to be sacrificed).

Process:

1. Wire up the relay to the extension cord providing power to the relay.  Specifically, cutting off the extension cord and then using the end with the prongs.  I've never wired up a relay - so this was a first to me.  I used simple wire strippers and cut the wires such that they would fit into the quick disconnects.  I crimped them using my wire strippers.  
2. Wire up the relay to a blank section of the extension cord (just using some of the wire).  Added crimps on the end to be connected to the relay and left the end to be wired to the door opener stripped but bare wire.
3. Here's what it should look like when all done:
   
4. Add the Z-wave Outdoor Module to my MiCasa Verde.  I did this by plugging it in near the Vera, turning it on and having it add it.  I referred to it as Garage Door Opener.
5. Take everything to the garage and plug it all in.  This involves wiring the stripped end of the extension cord to the same wiring terminal as the existing switch and remote control. This is what it looked like for me when it was all done. Note that the Z-wave switch and the relay were just sitting on top of the garage door opener (hopefully things don't move around enough to warrant zip ties).
   
6. Add a scene to MiCasa Verde called "Garage Door" to turn on the switch and then turn it off in 5 seconds (the shortest amount of time possible).

Note: The wiring of the relay should be pretty straightforward.  The one I had, there was a simple wiring diagram on the back of it showing the state of the relay when it was open (off).

[Update 2013-12-16]
To add a little bit of clarity to exactly what I am doing with the relay here are some additional pictures and some explanation:

My relay was set up like this; the diagram depicts the configuration of the pins when the power is off.  When power is switched on, the link between 2 and 6 or 1 and 5 flips to 4 and 6 or 3 and 5.  This is essential to trigger the garage door, which needs a circuit to close for a brief moment (that is what your normal garage door switch is doing).

A little bit hard to tell here, but the power outlet from the Z-wave controller (the two white wires with the blue warning label around it) are plugged into the 7 and 8 poles.  The garage door opener is connected to poles 4 & 6 (normally open to one another, but when the relay is active, closed).

Another view from the business side of the relay; here you can see the numbers clearly on the left side of the relay (not used).

[End Update]

After it is installed, you may want to ensure that your previous means of raising and lowering the garage door still work (remotes, hard-wired All set.  Assuming the relay is wired correctly, when the scene is run you are good to go.

Bonus Round

I also and almost primarily wanted to know whether the garage door was up or down.  To do that, I needed an window or door sensor.  I chose an EverSpring Z-Wave Door / Window Sensor.  Prior to going to the garage, I added the sensor to the MiCasa Verde.

It turned out to be pretty easy to physically install, the key was finding a spot where I could keep the gap between the sensor to a minimum (less than 40mm) that was added to the door and the portion screwed in to the garage door frame. For me, this required adding a door shim behind the sensor to make sure it was close enough to the magnet in the traveling sensor. After final installation, it looks like the following:

After a few test runs making sure that there was no interference and that the MiCasa Verde was reading the position properly, I declared victory. Victory may be premature - my guess is that this setup may have issues in the cold - there is an air gap at the top of the garage door that might hamper the batteries in the sensor. We'll have to wait and see if that is the case, though. A 4.5V DC power supply could be brought to bear if need be.

Finally, I see the following in the room "Garage Door":

Nesting

I decided to take the plunge and purchase some Nest thermostats when I read some of the early press about them. You couldn't actually buy one at the time, all you could do was put your name on a list to get them first when they became available. I did so and then waited.

My motivation was pretty simple: I wanted something that was easy to program and internet accessible. In fact, the joy of a "self learning" system wasn't one of my primary motivating factors. The interface for the Nest looked good and my understanding was that it should fulfill my other wants - so I went ahead despite the reasonably high $250/unit starter cost. If I really could schedule easily, I could easily save the purchase price in a couple of years. Despite the fact that we had two programmable units in the house, I found their interfaces to pretty much be impenetrable and, for the most part left them unprogrammed.

On February 14, I got a message saying that I could order my Nest. But, "Click below to order, but please note that you can only order once using this personal reservation page." So I decided to take the plunge and ordered two - one for the upstairs systems and one for the downstairs. They shipped within two days and were here quickly - it was just a matter of me finding a few hours to install them and configure.

The installation was not difficult and the materials provided were well thought out. They even include a screw driver with interchangeable heads to allow you to easily unscrew pretty much anything that you'd encounter. There were also nice little (blue) stickers that you could use to label each of the wires positions before undoing anything; this should ensure that you don't mess anything up.

Despite all of the good documentation, there was a cryptic note about not having to jumper the Rc and the Rh (both red) lines together. It wasn't clear if I needed to attach one of them or both, if they were separate wires. I eventually concluded that there is only one red control line coming from the furnace and just connected it to the easiest place on the nest panel.

One of the more painful parts of installing the Nest is finding a part of our wall that wasn't chewed up with past thermostat installations. It would appear that they have been replaced multiple times. Strategically choosing a good position took the most amount of time in both cases (and even worse when I had to pull off the Nest upstairs after the first time due to one of the wires not being connected well).

Here are all of the install pictures:


So downstairs, I was starting with a pretty old Honeywell control unit - this was the starting point for the downstairs thermostat - not sure of the vintage


The old on the floor - with the new


Wires all labeled


Connected to the Nest mounting panel - conveniently with a built-in level


Mounted!


Mounted - upstairs!

After all of the wires were connected, you simply pop the eyeball (in Christa's words) on. It will want to connect to your wireless network, so the first step is to put in your password. After that, it updates to the most recent software level (the one downstairs did this several times before it was done - the one upstairs, just once). After that you can begin the setup; the first step is to show you which wires it thinks are connected. It does this in an incredible simple visual way, making sure that there is no question about whether its done right.


The rest of the setup to do on the machine is a breeze. The fun part comes when you sign up for a nest.com account. When you do, it recognizes the thermostat on your local network and you are good to go. From there, I took two different approaches. For the upstairs thermostat, I decided to set a schedule and for the downstairs I let the system learn our behavior to set the temperature.

The web interface for scheduling is pretty good, though not great. It does include copy and paste functionality. I don't know if its just an early quirk, but I lost the schedule the first time I put it together. The upstairs Nest had lost its connection to our wi-fi network, so that could have been a contributing factor. The primary iPhone application doesn't do much more than allow you to adjust the temperature, but that is exactly why I bought it.


What Its Missing:

• My chief complaint so far is that Nest doesn't share any data with me. I should be able to access a log of all of the events of the thermostat (turn on, turn off, all of the relevant parameters at that point in time, such as why it was turning off or on).
• The units are designed to learn about your behavior to help you save energy. But when that happens, it doesn't tell you anything about what its learned or what you have saved.
• My wife doesn't like the aesthetic. It looks like an ugly eyeball in her opinion.