Peter has been asking us about doing the final installation of the monitoring system for our house, so that he can get some data out of it (finally!). In order to make that happen, we needed to finish the electrical work, and that's been a bit of an obstacle because of various lighting needs and also because our electrical subcontractor, Andy Harkness, has been in high demand recently. But we managed to lure him and one of his electricians, Karl, up to the house over the past two days to finish up the electrical work.
Imagine my excitement when Andy called me down to say "Ted, can we combine any of these circuits? The panel's full!" It turns out that we are able to keep all the circuits separate, at the cost of having _zero_ free space in the panel. If we need to add even one more circuit, we'll have to add a secondary panel.
Why so many circuits? Peter wants to be able to measure the power consumption patterns of individual devices and rooms. Along with the per-room temperature sensing and an outside weather station, this will give him (and us!) a really good idea of how the house is actually performing.
The snarl of wires coming out of the wall here are the wires to individual temperature sensors in each room, all of which terminate in the utility closet. There's also network wiring, of course—that's the other, smaller snarl of wires.
This is the control/status monitoring panel for the solar hot water. We may collect some data off of this as well.
This is a flow meter that will detect the flow rate of the hot water. We also want a flow meter for the glycol, but had some trouble sourcing one that would be reliable—Peter originally sent us a flow meter that's mostly made of plastic, and Gary, our solar subcontractor, took one look at it and refused to install it. When I called Peter about it, he told me that there had actually been some problems with the device in the field (nothing serious—one leaked a tiny bit, and another failed after a year). So we aren't installing this particular device—Peter's researching other options.
Oh, the reason for the weird bend in the pipe that comes into and out of the hot water flow meter is that we need a certain length of pipe after a bend or a valve before the flow meter, or the turbulence caused by the water flowing around the bend or through the valve will affect the measurements.
This is the flow meter for the well. We're not actually measuring this—it's just used by the sewer department for billing. But it's a meter, so I included it... :)
Ted and I were all set to install LED strip lights, as described in my recent post about choosing LED lights. But then we spoke with my father's friend John—a major techie at a major technical manufacturing company—and he suggested we wait a bit longer.
He said that for the next two years or so, the best LED products will be Edison-style replacement bulbs that use remote-phosphor technology. LEDs do not produce a wide spectrum of light on their own, but when LED light strikes a phosphor, the phosphor emits a wider range of colors. You can see this in the Philips LED replacement lamps: the unlit bulbs look yellow, but the light that comes off them is a nice warm white.
Those kind of replacement lamps are the best short-term approach, but the longer-term approach will be multi-string ("but not RGB"). He said, "They will be phosphor-shifted blue LEDs picking up green-yellow (called BSY), with some combination of red/orange/amber LEDs at the longer wavelengths."
He added, "CRI is only a start at analyzing the problem. It's very outmoded, made a lot more sense in 1950 than today. Doesn't measure reds well (which are very important to human perception), and the spectral absorption are too broad-band." This confirmed our experience of CRI — the lights we were going to buy had a good CRI (85) but was noticeably weak in the red part of the spectrum. Ted looked OK under our test lights, since he has fairly rosy cheeks to begin with, but they made me look a bit more wan than usual.
Our informant likes two lighting models right now:
CREE LR6 (BSY/orange two string) — "This dims well, but the color changes a lot"
One thing he likes about these models is that neither has 120Hz ripple, admitting that not everyone is sensitive to this, but that it drives him nuts. He also notes that efficacy is approaching 100 lumens per watt, "which is a good benchmark for a warm white bulb."
He suggested we wait at least a year before installing LED strip lights for the following reasons (direct quotes):
The models that you're looking at don't actually use DC, but rectified high frequency AC that tracks the input line (120Hz modulated 25kHz). This means a lot of flicker.
They are also are "local phosphor single string", which means bad color.
The efficacy with transformer is probably about 50 lumens/W, not awful, but not good either.
Our new plan is therefore to postpone installing the strip lighting, but to leave all the rough wiring in place. We'll make do with floor and table lamps for a year or so and then install strip lights once they've improved the color rendering and efficacy.
I told him that most of the light fixtures we're buying take regular A19 lamps (Edison standard), but a few will take B10 lamps (Edison candelabra bulbs). He warned that it's harder to make good replacement lamps for smaller bulbs because there's not enough mechanical volume to make a good LED ballast. I asked whether CFLs at that size are any good, and he said, "Most of the fluorescent at that size are CCFLs, which have good life, but won't dim well, and have lower efficacy than larger CFLs. There's a effect called cathode drop which fundamentally decreases the efficacy of these small lamps."
He concluded, "Maybe this gives you something to think about. A lot is going to change in the next few years."
I could not have designed and built this house without our good friend Google. I created all of the 3-D drawings in the free version of SketchUp, and I researched literally every component of this house using Google's indispensable search engine.
Unfortunately Google is nearly useless for researching items that are aggressively marketed online, particularly LED lights. The problem is that discount LED vendors use every trick in the book to rank among the top Google search results, so it's nearly impossible to find helpful online advice about how to buy LED strip lights.
Until recently I had only a dim (ha!) idea of what components we'd need for LED strip lighting. But now that we've figured it all out and completed our lighting schedule, I'd like to share what I've learned about illuminating a room primarily with LEDs.
Not all LED lights are created equal
I do not usually use all-caps (the Internet equivalent of screaming), but this is hugely important. It is tempting to order LEDs online from discount vendors, and you might know people who like the lights they purchased that way. Ted and I have some friends in Austin who purchased a lot of accent lighting from Eco Light LED, and the lights look really cool. Installed inside bookcases and behind valances, they have RGB controllers and can be adjusted to display a huge range of fun colors. Ted and I assumed we'd light our house with something similar, only on a larger scale.
We subsequently learned, however, that you can only get away with cheap LEDs if you aren't using them as a primary lighting source. When you look at a person, you're seeing the light that's bouncing off of them; the surface of their clothing, skin, etc., absorbs certain parts of the spectrum and reflects the rest back out. So if your light source is missing crucial colors, that person will look downright creepy.
Any decent LED manufacturer will publish the product's Color Rendering Index (CRI). This is an adequate (though incomplete) measure of the light's color fidelity. CRI is measured on a scale from 1-100, with ordinary incandescent lights at 100 and everything else somewhere below that. For your primary indoor lighting source, you shouldn't go below a CRI of 80.
Color temperature is also important, and it is easily misunderstood. "Warm" light actually has a lower temperature — incandescent bulbs are 2,700K and those blue-white LED xmas lights are around 6,000K (compact fluorescent bulbs usually range from 2,700 to 3,500K). All of our LED lights will have a color temperature around 3,000K and a CRI in the mid-80s.
Once you know a little about CRI and color temperature, the discount LED vendors no longer look so good. Eco Light sells Warm White LED Strip Lights for about one-quarter the price of the strip lights we're buying, but a closer look at their downloadable spec sheet reveals that the color temp is a not-so-warm 3,500K, and they don't mention CRI at all. I found CRI info on a few other discount sites, but the numbers were unacceptably low (70-75).
Most of our LED strips will be installed behind a valance, with the light shining upwards across the ceiling (the sole exception is the under-cabinet lights in the kitchen). The LEDs we selected have a good overall CRI (85) but are a little weak in the red part of the spectrum, which means we need to choose ceiling paint with a hint of red pigment so it won't absorb all the red coming from the LEDs. Our lighting consultant told us a cautionary tale about how indirect non-incandescent light bouncing off cream-colored walls can turn everything yellow, and we don't want that to happen to us. In fact, we had planned on using pigmented plaster for our walls, rather than the usual paint over drywall, but the lack of fine-grained color control pushed us back to the standard approach.
LED strip lighting basics
The first thing to know about LED strip lights is that they run at a much lower voltage than most other electric devices in your house. In North America, we use 120 volts AC (alternating current), and most LED strip lights run at either 12 or 24 volts DC (direct current). You will therefore need a transformer to convert from line voltage (120V) to the voltage of your LED system.
In case you're rusty on how electricity works, I should point out that the voltage has nothing to do with how much energy the lights draw. Voltage is analogous to water pressure (not the total amount of energy used), so the important number is how many watts a fixture requires.
Each circuit of strip lighting will require its own transformer (by "circuit" I mean a strip that's controlled by its own switch), and the total wattage of the strip lights cannot exceed the maximum output of the transformer. Our strip lighting draws 3W per foot and our longest stretch on a single circuit is 14'-4" (the upstairs hallway), which means our heaviest circuit will only draw 43W. The smaller WAC Lighting transformer is rated up to 60W, so we'll be well within the limit.
In case you're curious, here's what you'd need to run 14'-4" of strip lighting:
Two 2-inch LED strip lights
Four 1-foot LED strip lights
Two 5-foot LED strip lights
One 12-foot lead wire (connects the light strips to the remote transformer)
One end cap
One remote transformer (driver). Converts from 120V (line voltage) to the 24V required by the lights.
Each segment of strip lights connects to the next, and the end cap terminates the string.
Incidentally, the reason we're fiddling with the 2-inch segments instead of just trimming a 1-foot segment (the lights are trimmable every three inches) is that it's marginally less expensive. This will give us the exact length we need.
Why bother with LEDs?
I confess that writing this post has made me wonder why we're installing LED strips when CFLs use roughly the same amount of power. People criticize fluorescent bulbs (compact and otherwise) for containing mercury, but it's just a tiny amount and can be recovered if the bulb is properly recycled. LEDs don't contain mercury, but various toxic chemicals are used during production, and their heat sinks are made from valuable metals like aluminum.
So why are we using LEDs? I suppose it's partly the eco-bling factor. But we're installing plenty of conventional fixtures too, and in the short run we'll probably fit most of them with CFLs. LED replacement bulbs tend to shine in a single direction; this makes them a good choice for recessed downlights, but not so good for wall sconces and multidirectional fixtures.
We considered using a series of T5 fluorescent tubes for all the indirect lighting, but then we'd run into "socket gap" — dark spots where two bulbs meet. We would also have to build a larger valance to hide the bulbs, and there are a few places where we need to keep the valance quite small.
We therefore succumbed to the blingy lure of LEDs, and I sincerely hope they'll earn their keep during their projected 50,000-hour lifespan. That's 17 years, assuming we run them eight hours a day, which I doubt we'll do. Again, this is why it's important to choose the manufacturer carefully, since we're likely to be stuck with these things for a long long time.
I keep thinking I can cover lighting in a single post, but each time I try I wind up having too much to say. In a future post I'll share diagrams and photos of our indirect lighting setups. I'd also like to talk a little about switches (whee!), and now that my eyes have been opened to the subtle art of lighting design you can expect a rant or two about bad lighting.
Andrea promised a while back that I'd explain how the foundation works, but I haven't done that yet. We went through a fairly long and painful process to get to the foundation we have now. I think it was worth it—the current foundation looks very cool, and I think it will function well. So let's take a tour.
What you see to the right is the pier foundation as viewed from the west side, looking east down toward the road. You can just see the roof of the car behind the foundation of the garage. That's the sort of U-shaped bit of concrete to the far right, which I will not be talking about, since it's pretty much just a normal foundation.
The horizontal bar of concrete that's closest to the camera is a grade beam. It's mostly resting on footings that have been pinned to ledge, although there's a point in the center where the ledge comes so high that there's just grade beam on ledge, with no special footing underneath. What you can't see from this angle is that the ledge drops off pretty steeply on the other side of the grade beam.
In the center, you can see a wall heading east from the grade beam. This is a shear wall—it's there to prevent racking in the east-west direction. It's solidly on ledge all the way down. The big dark grey wall that's further down the hill is another shear wall that's to prevent racking in the north-south direction.
Racking is a situation where there is a differential force on a wall that tends to try to bring it out of square. The force is called a shear force. Shear walls are shaped so as to resist the shear force. Shear force in earthquake country comes from the ground moving, while the inertia of the house tends to want it to stay in the same place. So the foundation accelerates in the direction of the shear force, and the top of the wall has to move to catch up to it. If you don't have a good shear wall, the wall can fail.
A house that's on piers has limited shear strength, because the piers are just big sticks poking up out of the ground. The footing might look like it will have some shear strength, and indeed the rebar that connects the footing to the pier, combined with the rebar in the pier, does give it some shear strength. But many homeowners in the San Francisco Marina learned the hard way that a house standing on piers with no shear wall is likely to collapse.
But we don't get earthquakes in Vermont, right? Well, not as often as California, but it's best to be prepared. What we do get, though, is high winds. With winds, the ground wants to stay in place, and house wants to move, which will tend to make the piers fall over. We have a pretty tall wall to the south. So we need enough shear strength to accommodate that wind force without failing. So our structural engineer, Ben, specified these shear walls.
The other advantage of the shear walls is, I hope, entirely psychological: there are two piers that are not directly pinned to ledge. I worry a little bit that these piers might somehow fail—they might start to wander, or we might have gotten the drainage and cover wrong, and they might move due to ice expansion. I don't think this is a real risk—we examined this issue carefully and concluded that the footings were secure. But if for some reason we blew it, the shear walls give us a huge additional safety factor that will allow us to correct any problem that should arise. Even though this is extremely unlikely, the fact that they are there gives me a lot of peace of mind.
Referring back to the picture at the top again, you can see a slot in the north end of the north-south shear wall, and a notch in the south end. These are to accommodate the laminated veneer lumber (LVL) beams that will be running east-west along the tops of the piers. So let's talk about wood for a minute.
Anywhere where there is horizontal contact between the foundation and the concrete, we will have pressure-treated lumber. So the grade beam at the top will have a 2x (two-by, not two layers) of PT on top of it. The shear walls will have 2x or 4x, depending. There are notches on the east side of the grade beam on the north and south ends to accommodate the LVL beams; each of these notches will have first a layer of PT, and then the LVL beam will rest on top of that. The LVL beams will be attached to the piers with metal brackets.
Ultimately, when all the foundation wood is in, we will effectively have a sill plate to rest the floor box on. In places, it will look just like a traditional sill plate. In places, the top of the plate will be an LVL beam.
What ties this foundation together is the floor box. The bottom of the floor box is a structural membrane made of zip sheathing nailed to the bottom of the floor joists according to a nailing schedule provided by Ben. This will create a box that is very resistant to deforming as a result of differential forces. This is what allows the shear walls, which are not otherwise connected to the piers, to keep the piers stable.
By the way, I should point out that while it might sound like I know what I'm talking about here, I'm relaying to you a layperson's understanding of how this all works, not a structural engineer's. Hopefully what I'm saying here will be of some benefit in terms of telling an interested reader what sorts of things to look out for, but it's no substitute for actually hiring a structural engineer to analyze your specific project. If you're flying your house the way we are, you definitely want expert help.
This week Eli and his team got a start on cutting wood for the house. I want to talk a bit about Eli's process, because it's very different than what I normally see done by builders. Rather than trucking raw materials to the site and cutting them to order on site, according to a blueprint provided by the architect, Eli (who is actually an architect among his other talents) has a really high-end CAD package he uses to prepare detailed drawings and cut lists.
This gives him a degree of precision in his plans that allows him to hand his cut lists off to his team, who pre-cut all the wood in the house and pre-assemble what they can in Eli's shop. This means that they can use tools that are set up on a nice flat concrete floor, and store the wood under the roof, and not haul stuff to the site that's going to have to be hauled to the dump afterwards.
I personally find this process both frightening and inspiring. Frightening because I would not have the confidence to go from a CAD drawing to cutting in a shop and shipping the results to the site. Inspiring, because this is exactly the right way to do a custom house: you design the whole thing in a piece of software that is intended to produce accurate cut lists. You have a crack team that can follow the cut lists and reliably produce a stack of pieces that really do fit together. You do as much work as you can in the shop. Then you load it all on a truck and bring it to the site.
This is where we are in the process now. Eli calls it "making sawdust." Eli's crew is going to be cutting and assembling all next week. Nothing's leaving the shop. I think they may still be doing a bit of cutting the following week as well.
Another key aspect of this process is that as the pieces are cut and assembled, Eli's crew marks them, so that when the time comes to put them together on site, all the measuring is already done. So at that point all the team has to do is put the pieces together according to the markings.
Why do I think this is so cool? Because it's efficient. A well-built house, which I think everybody should get when they buy a new house, requires a lot of labor. But a lot of that labor is dead time where you're hunting for things, or switching tools, or whatever. Builders put a lot of thought into saving time on the site, but by doing things in the shop, the time savings are substantially more.
Also, because Eli knows precisely how many pieces are going to go into the building, we started out with a clean bill of materials that we don't expect to have any substantial surprises on it, and Eli is also able to predict with some accuracy, based on his past experience with his team, just how much labor is going to be involved in assembling this.
Many builders brag about building to code minimum, as if that were an achievement, but really building codes are intended to enforce the absolute minimum level of quality, so that if your house was built to code, you at least don't have to worry that it's going to catch fire for no good reason or fall down in a minor windstorm. Many houses built to code minimum have flimsy walls, poor air quality, poor noise isolation, noisy floors, and crooked, flimsy fixtures. This saves money, and allows the builder to sell the house for a lower price, or take a higher profit. There's nothing wrong with either of these things, but if it were possible to make something better, wouldn't that be nice?
Andrea mentioned money in her previous post, and money is a real worry in a well-built house. The problem is that a home buyer has no real idea what actually went into the house that they are buying. They can't easily tell that they are getting a well-built house or one built to code minimum. Some things are obvious, but some things aren't until you've moved in. By using his CAD/CAM system, Eli is building our house for a very competitive per-square-foot price.
The result is that when this is done, we hope to have a well-built custom house without it being so expensive that if we should need to sell it, it would be impossible to recoup our costs. Anyway, that's the plan.
Our building site was relatively quiet last week. Concrete is curing, and our electrician set up the main panel and meter in anticipation of CVPS turning on the electricity this week. Ted and I also spoke with several solar installers to see about getting some PV panels at the roof ridge and also a solar hot water system. More on that as it unfolds.
The biggest news is that we recently partnered with Efficiency Vermont to pursue Passivhaus certification [follow the link to read their "About Us" page]. The cool part is that our house will be part of a research project to evaluate the suitability of Passivhaus construction for Vermont. They'll install monitoring equipment in our house and closely study its performance.
Peter Schneider, Efficiency Vermont's Passivhaus consultant, was particularly interested in studying our house because it has several unusual features: a pier foundation and partial shading. Vermont's abundance of sloping, ledgy lots makes pier foundation a tempting solution, and of course trees are rampant hereabouts. So hopefully we'll provide useful data for would-be Passivhausers in North America.
Peter was on vacation last week, so he hasn't gotten farther than the first few rounds of PHPP tweaking, but Marc helped pick up the slack. This will all probably change this week, and I'm probably jinxing things just by typing this, but so far it looks like we can pull off Passivhaus performance with the following general specs:
11-7/8″ I-joist floor deck (16 oc), stuffed with dense-pack cellulose and with 4″ of polyiso underneath.
9.5″ I-joist wall framing (24 oc) filled with dense-pack cellulose and with 4″ of exterior polyiso.
Schuco SI-82+ windows, which we ordered this week from European Architectural Supply in Lincoln, MA. The windows are PH-certified and made from uPVC. Yes yes, PVC is evil, but this is unplasticized PVC which is apparently a bit less evil. It's made without phthalates and can be recycled, at least in Europe. But hopefully the windows won't need recycling for a long long time.
Climatop Max and Climatop Ultra-N glass. The glass offered by Schuco is pretty darn impressive. For the south windows we upgraded to Climatop Max, which has a SHGC of 0.6, but for the rest of the house we went with the Climatop Ultra-N, which has an SHGC of 0.5. All the glass has a Ug of 0.105 (which PHPP callously rounds up to 0.11).
We haven't decided for sure on the HRV yet, but we'll probably either do the Zehnder ComfoAir 350 or the Paul by Zehnder Novus 300. The latter adds about $1,400 to the already formidable cost, but the efficiency is 93% as opposed to the ComfoAir's 84%, which would win us quite a bit within PHPP. Another knob to turn would be to add more polyiso under the floor or use larger I-joists — we'll hopefully do the cost-benefit analysis this week and reach a verdict.
It seems like the biggest advantage in our design is the ludicrously simple house shape. We're basically building a shoebox with a shed roof, which means there aren't many corners or thermal bridges undermining our envelope. Marc, Ben, and Eli already minimized thermal bridging before we decided to go for Passivhaus certification, so we're picking up a lot of PHPP points without having to change our plans.
We're waiting on a few more details, though, including some THERM data Peter is confirming with PHIUS. Hopefully that won't kick us back out of the ballpark, but as I said we still have some knobs left to turn.