Sunday, February 24, 2013

Feathering your Nest

I have to admit that the Nest thermostat is one cool gadget. With almost everything else in our home in the North Carolina mountains on-line, the HVAC system was the one missing link to complete second home nirvana. When my wife gave me one of the new 2nd generation Nest thermostats for Christmas I knew I would have it installed and operating before the new year - I could not have been more wrong!

Pre-Christmas checks with the handy compatibility tool on their website were positive for our Waterfurnace geothermal system. I even emailed a photo of the connections to the existing TP32W03 thermostat provided by Waterfurnace to customer service just to make sure. A reply email and phone call assured me all was copacetic.



I knew something was seriously wrong as soon as I connected the base and I got a flashing eror indicator on the Waterfurnace Intellizone zone controller indicating a thermostat wiring error. After removing the zone controller cover I saw all the control lines from the Nest lit up meaning it was calling for heating and cooling at the same time and for all compressor stages. Fortunately the controller was smart enough not to relay these commands to our main geothermal unit and perhaps resulting in an expensive failure. Note that this was without the Nest main unit plugged into the base! My naive first evaluation of the Nest base was that it was just a nicely designed wiring interface between the handy push-in wire connectors and the mutli-pin connector that the Nest plugs into. To make a long story short, after repeated and lengthy calls to the Nest customer service line and conversations with their friendly if not too technically savvy staff, I finally gave up and put the old Waterfurnace thermostat back to get the house back up to a reasonable and wife-friendly temperature.

After the holidays and some serious googling on the state of the art in thermostats I had a clue as to the cause of the problem. A call to the Waterfurnace tech support line confirmed that the real source of the problem was the Nest's 'voltage stealing' technique that was putting enough current on the control lines that the zone controller interpreted as a 'on' condition on all lines. The customer service agent also said they had many calls from customers trying to install Nest thermostats and that some installers had experimented with adding resistors to the control leads without success.

Back in the day HVAC systems were all about relays and motors and thermostats were simple mercury switches that connected 24 VAC from a transformer to the right wire to signal for heating or cooling. Many older HVAC systems were wired with just one leg of the 24 VAC transformer and the signal wires to the thermostat locations. The 'common' (C) lead was omitted since the relays in the HVAC unit were already connected to the other side of the transformer. When electronic programmable thermostats came on the scene in the 70's there was no easy way for the 24 VAC transformer to power the thermostat logic. Batteries were only good enough to provide power through electrical outages. Some bright engineer at one of the leading HVAC suppliers came up with the idea of sinking just enough current through the still relay-controlled HVAC systems to power the digital logic in the thermostat. "Power stealing" became a common feature of digital thermostats as explained on ZenHVAC site. All was well and good until the HVAC systems themselves ditched relays and went to digital microprocessor controllers and the voltage stealing circuits had to evolve since there were no longer any relays to sink the stealing voltage. Getting back to the Nest design - it was now obvious that the base of the Nest had extensive logic inside (a fact I should have found from one of the teardown reports on line) and the voltage stealing logic was the culprit. A search for patents related to this madness came up with USPTO US20120325919A1 which is a patent application from Nest describing their voltage stealing technique.  It was time to head for the lab!

Nest on the bench
I dusted off an old 24 VAC transformer I found in the junk box and started hacking out what started as just as a HVAC 'simulator' so I could probe the Nest control lines but morphed into a relay interface that isolates the Nest from the zone controller. I made it totally from junk box parts I had lying around. There is undoubtably a simpler circuit that would accomplish the same task of letting the Nest harvest voltage from the control lines without putting any voltage on the HVAC system but my first approach using resistors, diodes and relays does the job.

To get an idea of how the harvesting works I probed the lines with a scope with everything powered up but with the temperature set so that NO signals should be coming from the control lines. Each control line was terminated with the circuit shown below.

Y1, Y2, W , G and O leads from Nest
Clearly the roughly 6 VAC negative AC waveform was enough for the Intellizone controller to think it was time to heat or cool and therein lies the problem. By replicating the circuit below 6 times (one each for the Y1, Y2, W, O and G lines) the problem was solved. You can use any low voltage DC relay (the ones I had on hand were 5 VDC relays -  Coto 9007-05-01). All parts can be found at RadioShack or DigiKey.



Once I installed the interface between the Nest and the Waterfurnace Intellizone controller all was good. Our home now has its HVAC system (at least the main zone) on line and we can turn on the heat before we head up to the NC mountains on cold weekends. I have no way of knowing how many other Nest customers have run into compatibility problems related to their voltage stealing technique. The simple solution would have been for the logic in the base to sense of a common wire was attached and then to disable all the voltage stealing logic if it was. Anyone at Nest looking into this for the next version?