Categories

Quick Performance Update

I have a few posts in mind that still need to be written, but I wanted to do a quick update since I have over two months of performance data now.

We got our second electric bill since installing the system, and it was even lower than the first.  I combined the readings from the bills with my production logs and figured out how much of our total usage is provided by the solar panels.  Here’s the results:

From July 14th through August 12th, 60% of the electricity we used was provided by solar power.  In the next billing cycle, August 13th through September 13th, that number rose to 71.2%.  The system wasn’t online for the first few days of the first bill, which explains why solar appears to be less of a source than in the second.  I expect these numbers to start dropping as the days get shorter, then pick back up again in the spring.

This Site Used in ToughTrac’s Literature and Webpage!

ToughTrac, the racking manufacturer I used, recently contacted me to ask for photos from the install that they might be able to use for their website.  I sent them a link to this site and said they were free to use what they wanted, so long as I was credited.

They ended up using a few photos for the page describing their Pitch Roof Set (the one I used), which is what I assumed would happen.

What I didn’t expect was for them to turn my post on the feet installation into a user’s guide!  One of my main purposes in making this site was to pass along what I’ve learned so that others will have an easier time installing their own systems.  Hopefully future ToughTrac customers will reach this site via the guide and find it helpful.

Production Data

Now that the system is installed and online it needs to be monitored.  Well it doesn’t need to be monitored, but I love data points and charts so I’m monitoring it.

The Real Time Production Data and Daily Logs page is where the current production is displayed, along with daily logs since the system went online.

The Xantrex GT 5.0 Inverter has a serial port on it that can be used to query to the inverter for various pieces of information.  They also have a piece of software written by Solar-Guppy called GT-View which talks to the inverter via this serial port and provides real time data as well as logging and daily totals.  Unfortunately, GT-View is horribly out of date and riddled with bugs.  The good news is that SG-View is pretty much the same thing, but with more features and less bugs (and not under contract by Xantrex).

SG-View isn’t perfect, but it does most of what I want it to do and is fairly easy to set up and use.  GT-View has the option to create a PNG file of the display every X seconds (minimum of 30) and then upload it to an ftp server.  This is what I’m doing (along with an auto-refresh script) on the Real Time Production Data and Daily Logs page.

Here are the current production values of the array:

The PNG generation feature literally takes a screen shot of the program.  This is great for an easy to implement web based display, but it means that the GT-View window must always be on top and that screen savers must be turned off.  Anything covering the window will show up in the screen shot.

GT-View can create daily logs with a data rate as high as 1 entry per second (I use 10 second intervals), which it can then upload to an ftp server at a specified time.  It also creates a Daily Totals log that gets updated at the end of each day, but it doesn’t seem to upload this to the ftp server.  I wrote a simple bat file to do this for me and then scheduled it to run each night.

Now that I have a log file that gets uploaded each night along with a Daily Totals log that also gets updated daily, I needed som way to make use of them.  Luckily Google provides an excellent chart API free for anyone to use.

Using this API and the log files, I wrote a small script that will generate an area chart when it is fed the location of a log file.  I had to limit things a bit because a full day of data logged every 10 seconds is too much for the API to handle.  I chose to pull a data point every 15 minutes, ignoring the points in between, and  using this to generate the chart.  It’s not perfect, but it gives good results.  Here’s an example using the log from my best day of production so far.

There’s also a section on the logs page that displays records and averages.  This is largely derived from the Daily Totals log.  The Daily Totals log contains information such as peak power, time online, total power produced, etc. for each day.  You may notice that the peak listed for some days doesn’t match the peak in the particular chart for that day.  There are a couple reasons for this.  First, as I mentioned earlier, the charts are created from a data point taken every 15 minutes, so the peak may have been skipped.  Second, the inverter keeps track of the daily peak and reports it when the Daily Totals log is updated.  if the peak happens in the 10 seconds between data points in the daily log, it will not show up in the log for that day, but will be reported in the Daily Totals log.

The peak values are pretty much self explanatory.  The inverter limits the DC power to 5300W and has a max efficiency somewhere around 96%, so even if I could be making more power, it will be limited for safety (this rarely happens, and usually only for short bursts).

Average production is just Lifetime Production/Days Online. It’ll jump around less the longer the array is online. Right now it is pretty volatile still.

Estimated Savings is my Lifetime Production multiplied by my current cost of 15.853 cents/kWh.

Estimated Pay Off is based on my after incentive costs of $3545.40/(Avg Production*kWh Cost).  This one is also pretty volatile right now and won’t be all that accurate until I have a full year of data.

Links to Resources Page Added

I added a Links to Resources page that contains a lot of good information regarding system planning, financial incentives, and the paperwork associated with NH’s rebate program.  This should be very helpful if you plan to do a project similar to ours.

Up and Running

The array is online and making power!

Well, it actually went online a little over two weeks ago, but I wanted to catch up on the installation related posts before going into anything else.

I took a short video of the array coming online and spinning our electric meter backwards as it pushed power out to the grid:

Sorry for the shaky framing, I filmed this right after finishing a lot of roof work in the hot sun and was a bit worn out.

Everything has been inspected and approved, and my final rebate paperwork is in the mail.  Hopefully I will receive a check from NH in the next 4-6 weeks.

Solar Panels

As I mentioned in my System Design post, I didn’t have many requirements for deciding which panels to use.  I pretty much just looked around for the lowest price per watt, then answered three questions about the panel:

1.  Is it UL1703 listed?
2.  Will I be able to make an array close to 5kW that works with my inverter? (the Xantrex sizing tool was very helpful in answering this)
3.  Will this array fit on the roof?

If the answer was yes to all three, I was happy with the panel.

I ended up using 26 Kyocera KD185GX-LPU panels arranged electrically as two strings of 13 panels and arranged physically as two rows of 10 with a row of 6 centered above them.

I would have loved to use an American made panel like those by Evergreen Solar, but the way their panels are designed wouldn’t allow me to get as close to 5kW as I could with the Kyocera panels.  At least Kyocera is building some panels in the US.

It may be enticing to buy non-UL (or comparable) approved panels due to the price drop, but it really isn’t a good idea, especially if you’re putting them on your roof.  It might be fine for an off grid ground mounted array, but most rebate programs require UL listed panels, and I highly recommend them.

I ended up buying my panels from Sun Electronics.  They had the best pricing I could find, but in my experience, customer service was so-so at best.  I couldn’t understand 90% of what the sales rep I got stuck with said.  It wasn’t a thick accent, he just didn’t enunciate anything and talked like he had a mouth full of cotton and molasses.  His reply emails were also rarely longer than 3 words, no matter how many questions I asked.  He initially agreed to toss in some free cables, but rescinded the offer after I refused to pay by wire transfer or personal check and insisted on using my credit card.  I then had to fax them a copy of both my credit card and my driver’s license along with a form to authorize payment on that card.  Hardly a secure transaction.  I’ve had better service buying things that cost 1/100 of what I spent on the solar panels.

I can’t personally give any reviews on altE, but I’ve heard good things about them from other people who have done solar installs.  They also have decent prices, but not quite as low as Sun Electronics did.  I hear they have excellent service though and will even check designs for you to make sure you are using the panels properly.

You may even be able to get further discounts if you order by the pallet (roughly 20 some panels). Googling “panels by the pallet” will turn up a few places.

Roughly a week after ordering them, this showed up in front of my house:

The panels weighed in at a little over 1100 lbs and had to be shipped freight.  Here’s how they were packaged:

Some of the more observant people reading this may have noticed that despite the blatantly obvious “DO NOT LOAD MORE THAN 10 CARTONS HIGH” stamped on each carton, there are in fact 13 cartons stacked on that pallet.  So in addition to putting mumblers on the phone, Sun Electronics also hires shippers without basic math skills like counting.  They may also be illiterate, I’m not certain.  Luckily the bottom panels appear to be undamaged.  This piece of cardboard was also in the pallet, further demonstrating the horrible math abilities of their employees:

I expressed concern about the shipping method in an email to Sun Electronics, but never received a reply.

Enough about them, back to the good stuff.

Each carton contains 2 panels:

Please note that the panels are sitting on the side of the driveway that we use for extra parking, and that the rest of our grass looks marginally better.  As I mentioned in the wiring post, each panel comes with a positive and negative lead terminated with an MC4 connector.  This allows each panel to simply snap into the next for an easy electrical connection.  This only works when wiring the panels in series.  Since our array is made up of two strings of 13 panels in series, this wasn’t an issue.

While having a helper will make things tremendously easier, it is possible to put these in by yourself. Each panel weighs 35lbs, and isn’t too bulky for one person to carry. Mariah was recovering from knee surgery at this time and couldn’t safely work on the roof, and it was hard to convince friends to either get up really early or work on the roof in July, so I ended up putting in a lot of these on my own. I did have help from a couple friends, which was greatly appreciated. We were lucky in that our porch roof has a more gentle slope and can easily be reached from the deck by a simple 7′ step ladder. Mariah and I worked out a system where I would carry a panel over to the edge of the porch and sit it on a chair, she would hold the panel while I climbed up onto the roof, then she would climb on the chair and help me lift it up to the roof where I set up a small staging area.

This way she could help me get 6 panels onto the roof, then go back to what she was doing instead of waiting for me to install each one.

I started out by test fitting the top row of 6 panels.  I’d end up removing them before I was done, but I didn’t know that at the time.

The panels simply sit on top of the rails, and are each held in place by 4 clamps.  There are  end clamps for the outside edges of the row, and mid clamps to go between panels.  Mid clamps are shared by adjacent panels, so for each row you need 4 end clamps and 2x(n-1) (where n is the number of panels in the row) mid clamps.  A square headed bolt fits shaft up into the end of the rail and can be slid down the length of it.  A clamp is then slid onto the bolt, followed by a washer and a lock nut.  Tightening the nut draws the clamp closer to the rail and secures the panel.  The recommended torque was about 15 ft-lbs.

I picked the top row as a starting point because it was the shortest row, and farthest away from the slightly scary edge.  I also had a good center reference on the top row.  From there I went down two rows:

Then over to the end of the bottom row.

In retrospect I should have just started on the bottom row, but doing things this way did allow me to bring the array online at half power a few days before I finished the entire thing.  Here’s half of the first row done:

Each half of the array made up a string – 5 panels on the bottom and middle rows, and 3 panels on the top row.  This is one complete string along with 3 panels of the second string.

Since the strings are combined in parallel, I could now hook the first one up to the inverter and collect power from it even though the second was not installed yet.  I was only running at half power though.

When I went back up to finish the other string and complete the array, I didn’t like how the rows were starting to drift apart from each other, so I undid most of the first string and did a better job squaring the panels. I then moved the other 3 top row panels to the lowest row and finished it off before moving on to the second row.  I probably should have started on the low row, put in all 10, then moved up to the next.  Doing it this way makes it much easier to fit and connect the panels.

Finally, all the panels are in!

You can see the bare ground wire laying loose on the roof.  To finish up the install, I grounded each rail and trimmed off the excess lengths of rail.

A few shots from other angles:

All Done!

Interior Wiring (Inverter to Grid)

The order of these posts isn’t necessarily how I did things. For instance, the inverter was worked on in parallel to the rest of the jobs. This was due in part to it being central to the PV system, but also because I could work on it during mid day while it was too hot to be on the roof.

This is the inverter, a Xantrex GT5.0:

The inverter takes the variable DC voltages and currents from the panels on the roof, and converts that energy to a constant 240V AC that can be used to power our house and the grid.

I knew the dimensions and weight of it beforehand, but was still surprised at how heavy (60lbs) and solid feeling it is.  Most of the front is taken up by a heat sink.

I picked an area in the basement near the electrical panel (behind that door) for the inverter.

The stones are the top of a brick cubby, not the floor. The inverter isn’t that huge. You can see where I outlined the inverter and meter locations, though I ended up moving the inverter farther left to center it on a stud. The inverter mount attaches to the wall, then the inverter hangs off of it and only needs to have a couple light screws to keep it in place.

You can see the type of screw I used to mount the hanger, as well as the inverter with the wiring box cover removed.

Here’s a better shot of the wiring box.  The large disconnect built into the box is the same type as the one I used on the outside of the house.

Around the time I installed the inverter, I also put in the production meter and related conduit.

The numbers on the meter will normally be level. It’s crooked because it is only loosely sitting the housing in that photo. It’s tough to pull back out once seated in the actual contacts, and I hadn’t wired it at this point.

To minimize losses, it’s best to keep the distance from the solar array to the electrical box as short as possible.  As shown in the previous post, after leaving the roof, the DC wires from the panels go through the disconnect switch and come into the basement through the side of the house.  The conduit containing these wires emerges by the top left corner of the door covering the electrical box, runs down the wall, and into the left side of the inverter.

The produced AC power travels from the inverter to the electrical box via a length of orange colored 10/3 wire.  The AC wire comes out the back of the inverter, through the wall, and into the box holding the electrical panel (behind that door).  In order to keep track of how much power the array has produced, the production meter was placed (in an electrical sense) between the inverter and the electrical panel.  After coming into the box, the AC wire goes through an elbow of 1″ conduit to the meter, then back from the meter to the electrical panel itself.

The production meter isn’t a requirement, but they cost less than $40 (including the housing and shipping) and I wanted a second way to keep track of how many kWh our array makes over its lifetime.

Once inside the electrical panel, the wire is connected to a 30A 240V circuit breaker.  Normally power flows from the utility, through a main 200A breaker, then through a smaller breaker to an appliance or receptacle.  Power produced by the solar array will flow from the inverter, backwards through the circuit breaker, then through the other breakers to whatever is switched on at the time.  Any excess power will flow out of the main breaker and feed the electrical grid that powers everyone.

Here’s the finished inverter and meter:

The inverter is designed to work indoors or outdoors.  It has a small back lit display, but instead of using buttons to navigate the menus, you knock on the front of the inverter.  One knock turns on the back light, the next cycles to a different set of values, and so on.  Since this system is tied to the electric grid, if the grid goes down (a power outage)  from something like a heavy snow storm, the inverter will disconnect and cease to produce AC power.  This is a requirement in order to connect to the grid, and is a safety measure to ensure that no one is energizing the grid while utility workers are trying to repair or maintain it.

Exterior Wiring and Grounding

I test fit a few panels before doing the wiring in order to be certain about where things would be placed and routed, but most of those were removed and replaced before the job was done.  Just pretend you don’t see any panels in these photos.

Wires

Section 690 of the 2008 National Electric Code (NEC) deals entirely with photovoltaic systems.  Read it before you do any planning, then read it a few more times to make sure you understand what is required.  Each of my panels came fitted with a 30″ positive lead, and a 72″ negative lead.  These leads are made from #12 USE-2 wire and terminated with MC4 connectors.

It is perfectly fine to leave these wires and connectors exposed on the roof, but once you leave the vicinity of the array, it’s a good idea to transition to THWN-2 in some type of conduit.  USE-2 is more expensive than THWN-2, so it may also be cheaper to do this.  Some people go into their roof at the transition point, but I wanted to keep the wires outside of the house until I could route them through a disconnect switch.  It is legal to route them directly into the house, but they must be in metallic conduit until they reach a disconnect switch.

Schematic

It might be helpful to look at the schematic I posted earlier while reading this post:

Panels

The array consists of 26 185W panels arranged as two parallel strings of panels.  Each string has 13 panels connected in series.  This means that the positive terminal of one panel is connected to the negative of the next, and so on for 13 panels.  This leaves an open negative terminal at one end of the string and an open positive at the other. As you can see in the photo at the top of the post, one wire has a socket, and the other a plug.  Connecting panels to each other is as easy as plugging one into the next.  The connectors are watertight and lock together once connected.  A special tool (or a flat head screwdriver and some wiggling) is required to disconnect them.

The inverter used in my design requires at least 235VDC to operate efficiently.  When panels are placed in series, their voltages add up.  The resulting voltage is dependent on many things like temperature and solar irradiation, but for the sake of simplicity assume that each of the panels produces 24V.  13 panels x 24V per panel = 312V for each string.  I could run the inverter off of a single string of panels, though it would only produce 2405W of power.  When two strings are combined in parallel, the voltage stays the same, but the current doubles, doubling the power.  Even though the strings are combined in parallel, the wires from each string are kept separate until they reach the inverter.

Grounding

The frames of the panels and the exposed metal racking all need to be tied to ground.  I used bare 7 strand #6 wire for the exposed portion on the roof and by the ground rod, and #6 THHN for the wire in the conduit.

Panels

The panels have a frame made out of black anodized aluminum.  Think of it as a coat of paint.  In order to make a good ground connection, you must get past this layer to the bare aluminum below.  To do this, I bolted a grounding lug to the frame of each panel with a star washer between the two.

Bolt, Lug, Washer and Grease

Here’s the whole assembly:

The bolt goes through the lug, then the star washer, then the panel, then the flat and lock washers, then finally the nut

The star washer cuts into the lug and panel frame when it is squished between them as the bolt is tightened, creating a solid ground path between the two.  The fastening hardware is all stainless steel, and dielectric grease is used around the star washer to help maintain a good connection.  Here’s one installed on a panel:

The bare ground wire passes through the “U” channel in the lug.  The set screw is used to securely bond the wire to the lug electrically.

This ground wire is bonded to the solar racking as well.

Racking

The racking is also made out of aluminum, which is a great electrical conductor until it oxidizes.  The oxide layer actually protects the underlying metal by preventing further oxidation, but it strongly resists the flow of electricity.  Because of this, each section of rail has to be electrically bonded within a row, and each row has to be bonded to ground.  To bond each rail segment, as well as add strength to the splice points, I used short jumpers made of #4 THHN with ring lugs crimped on each end.  These were attached to each side of the splice using a self tapping screw, with a stainless steel star washer placed between the ring lug and aluminum rail.

Each rail was grounded to the next using a grounding lug.  At places that 3 wires came together I had to solder a clamping lug to the top of the ground lug and bolt both to the rail together. A set screw is used to clamp the wire in place.

With the rail segments connected, and the end of each rail grounded, I checked the other end to see if the connections were good.

0.000 means no resistance, or good to go.

Earth

All of the grounded components are connected to a single #6 THHN ground wire that travels in the conduit with the other conductors down to the side of the house.

I’ll get to the disconnect in a bit, so ignore it for now.  Inside the disconnect box the ground wire from the roof gets connected to another section of bare 7 strand #6 wire.  The other end of this wire gets attached to an 8′ copper coated rod driven into the ground.

Lots of pounding later

I recommend putting the clamp on before you start driving in the ground rod. I had to cut the top off of mine to get the clamp on because it had flattened out so much.

Wiring

Conduit & Pulling Wire

Here’s the pile of wire and conduit:

120′ of bare 7 strand #6 copper, 90′ of #6 THHN,
65′ of 3/4″ liquidtight flexible metallic conduit (LFMC),
and the spool in the middle is 500′ of #10 THWN-2.

I was originally planning on using 3/4″ EMT conduit, which is about $3-4/10 ft. section.  The LFMC was $1.34/ft but has several advantages over EMT.

LFMC is a plastic coated, coiled self-interlocked ribbed strip of metal (usually aluminum or steel) that forms a tube.  EMT is sections of galvanized rigid steel pipe.  In order to route EMT, it must be bent, either by hand (hard to make it look good) or with a $40 tool.  Connectors (liquidtight for this application) can also be used to change the direction of the EMT, and must be used to join two sections of piping.  Each of these liquidtight connectors cost more than the EMT itself, and are still prone to letting water into the conduit.  EMT is galvanized to prevent rust, but it is doubtful that this coating would last the expected lifetime of the solar panels.  Pulling wires through EMT can also be challenging.

LFMC is UV resistant plastic coated, making it sun, rust and water resistant.  It can be bought in very long lengths, eliminating the need for joining connectors, and since it is flexible, it requires no special tools or connectors to bend.  Having one continuous length drastically cuts down on the number of places water can get in.  Since LFMC can be installed as a continuous run, it’s possible to pull wires through it before installing it, which is what we did.  You can also route it as you go, rather than having to measure, cut, and attach each piece of EMT.

My anaconda don’t want none unless you got liquidtight compression fittings hun.

As I mentioned, we pulled the wires through the conduit before installing it.  For my system the conduit contains 5 wires – 4 x #10 THWN-2 (2 for each string of panels) and a single #6 THHN ground wire.  I cut the wires to length leaving plenty of extra, then gathered up one end of each of them, and taped them all together.  I then tied a bit of plastic bag to a string, put it in one end of the conduit, and had Mariah turn on a shop vac attached to the other end to suck it through the whole length.  Once the string was through, I tied and taped it to the end of the wires.  We laid the conduit out as a long straight line, and ran the wire through it with one of us pulling on the free end of the string while the other fed the wire into the conduit and pushed it along.  After the wire was through, I coiled the conduit/wire back up and took it up onto the roof.

Down the Roof

The conduit needed to run from the termination point of the strings (the top middle of the array), across the roof, and down the side of the house.  It also needed to be elevated above the surface of the roof to keep it cool and prevent it from acting like a dam when rainwater is running down the roof.  Luckily, the solar panel racking worked perfectly as a place to mount the conduit:

Standard PVC clamps were used with stainless steel fasteners to attach the conduit to the rails as it ran down the roof.

Across the Roof

After running the conduit down to the lowest rail, I ran it across the roof toward the chimney side.

The conduit fit perfectly on the rail. It is just low enough to not interfere with the panels sitting flush to the rail, and the lip of the rail is wide enough to easily hold the conduit. All I had to do for this portion of the run was cable tie the conduit to the rail to keep it secure.

The rail did not go all the way to the edge of the roof, so I had a gap of 7 feet or so to cover before the conduit could run down the wall. The conduit still needed to be elevated above the roof, and after some head scratching I came up with this solution made largely out of extra racking parts:

This consists of a 3/4″ metal conduit clamp screwed into the underside of a panel mid clamp. I cut 2/3 of the sides off of a splice piece to make it flat, and screwed the mid clamp to the other 1/3. I then drilled a couple holes into the flat part. The flat part is designed to slip under a shingle and attach to the roof via screws through the drilled holes. This way the shingle protects the screw holes from water. The conduit then sits in the clamp on top. Here they are installed:

Disconnect Switch

After running the conduit to the edge of the roof, I lowered it over the side and down the outside wall near the service entrance. I trimmed the considerable extra conduit and wire, still leaving plenty to work with.

The conduit coming off the roof will eventually be connected to a disconnect switch mounted on the outside wall of the house. From here, another section of conduit will take the wires through the wall and into the corner of the basement to connect with the inverter. Starting in the basement, I drilled a hole through the wall to the outside, then went outside and widened it to fit the conduit. The spare chunk of conduit that I cut off earlier was pulled through the wall, with only a small portion left outside.

I attached a 90° liquidtight compression connector to the end of the conduit, then pulled the rest of the conduit inside so that the connector was flush with the wall. Before doing so, I filled the opening in the wall with caulk so that it sealed the connector-wall gap after the conduit was pulled tight. The other end of the 90° connector enters the bottom of the disconnect switch box via a knockout.  The box is then attached to the wall.

Here you can better see it coming into the box.  You can also see the bare ground wire leaving through a small hole in the bottom.

I trimmed more conduit off of the length hanging from the roof and fitted a straight through liquidtight connector to it, which was used to attach it to the disconnect switch box through a side knockout.

I ran the extra ends of the wire trimmed from the section coming off the roof  through the conduit section going into the basement.  I left them loose in the basement for now, then went back outside and wired up everything in the disconnect switch box.

The white wires from the roof (String 1-, and String 2-) are the grounded conductors in my system, and are simply joined to the white wires going into the basement with wire nuts.  Similarly the ground from the roof, ground to the outdoor rod, and ground to the basement are all joined together and to the metal box with grounding lugs.  The black wires are the ungrounded conductors .  They come down from the roof with each one going to a separate Line side pole of the disconnect switch.  The Load side pole is then connected to a black wire going into the house.  This allows me to easily disconnect the dangerous the array from the rest of the system in case I need to work on it or there is an emergency.  The inverter uses the exact same switch, so this was not a code requirement – I could have just run the wire directly into the house to the inverter.

Transition Box

Back up on the roof, the start of the conduit was still open with extra wire hanging out.  This is the point where the USE-2 wires with MC4 connectors from the two strings of panels transition to THWN-2 inside the conduit.  To accomplish this, I used a weatherproof enclosure and some special fittings to make a transition box.  I then cut a couple short MC4 extension cables in half, and fed the cut ends into the box.

This picture is taken from the top of the array, looking down toward the gutter.
The box sits just above the top row of panels, in the middle.

The fittings have a rubber sphincter in them that clamps down on the wire as the end of the fitting is tightened.  I ran the #6 ground wire through one, and the 4 USE-2 ends through the other.  The bare ends of the USE-2 was joined to the conduit’s #10 THHN in the transition box, as well as joining the bare and THHN grounds.  The conduit exits the gutter side of the box and continues on as you saw above.

The ends of the strings then simply plug into the transition box.

The wires coming into the basement are headed for the inverter, but that deserves a new post.

Solar Racking – Part 2: Mounting Rails

I fit, took apart, and refit things a few times as I figured out the best way to do this, but I’m only showing one run through here. I also had to replace the lag screws in a few of the feet with longer (and in one case wider) screws. The front row was particularly bad since I had drilled 5/16″ pilot holes. The other rows had 1/4″ holes, but If I had to do this again, I’d drill 3/16″ holes instead.

After peppering the roof with footings, it was time to install the rest of the racking system. Here’s a shot of the rails laying on the roof roughly where they will be installed:

There are a few pieces of hardware involved with mounting the rails to the feet.

First, the feet themselves:

On top of each foot goes a U bracket.  A lag screw and washer hold the bracket to the foot and the foot to the roof.  The slot the lag screw goes through allows you to adjust the bracket position over the foot to keep the rail straight even if the feet aren’t 100% aligned:

On top of the U bracket (inside it actually) sits a rail mount:

The rail is then attached to the rail mount by sliding it through the mount.  In practice, it’s a lot easier to put the rail mounts onto the rail sections first, and then laying the entire assembly across multiple feet.

The U bracket has slots cut into each side of it, and the rail mount has a hole drilled through it in line with these slots.  A bolt with a washer on it is slipped through one side of the U bracket, through the rail mount, and out the other side of the U bracket where it is secured with another washer and a nut.  The idea is that this can be used to adjust the height of the rail above the foot to correct uneven roof surfaces.  I found that I had one low section on the roof and had to raise 4 of these up an inch or so.  The other 37 were fixed at the lowest height.

The rails are free to slide in the mounts at this time.  Once everything had been triple checked for alignment, the rails were locked in place with a self tapping screw driven through the rail mount and into the rail.

Each rail was a bit more than 12′ long, so multiple rails were needed per row – 3 for the rows holding 10 panels, and 2 for the rows holding 6.  A splice piece was used where the rail ends butted up against each other.  The splice is an extruded U of aluminum with a hole drilled through it on each end that slips inside therail piece.

Out of the entire racking system I used (ToughTrac) this part was the most aggravating.  The hole locations in each splice were not consistent, so I had to test fit a few on each junction to make sure the holes in the splice lined up with the ones in the rail without leaving a gap between rail ends.  I was sent at least 4 extra splices, which I thought was an error, but may have been just for this reason.  The holes were also set too close to the base part of the rail, making it very difficult to install the bolts needed to secure everything.  One final complaint is that there was only one bolt on each side, which lead to the splice point being flexible.  A second bolt on each side would make the joint much more stable. ToughTrac actually gave me a follow up call to ask how the install went, and when I voiced these concerns they said that they were aware and were addressing all of these issues in the next revision of their product.  I doubt these problems will still exist for anyone planning on buying ToughTrac racking a few months from now.  Overall I was pleased with the product.

As you’ll see later, I added my own secondary fasteners while ground bonding the rails.  I also tried to line up at least one splice in each row over a foot so that it would be unable to sag.  Here’s a completed splice lined up on a foot

The rails were test fit into the brackets on the feet, then the brackets were adjusted as need to keep the rails straight and parallel to each other.  After the rails were in place, the height adjusters were used to level them in a few places.

Here’s my friend Les and I adjusting everything:

The end result was straight as a… well, rail:

Here’s the racking in its entirety.  I actually ended up removing the rails and re-setting them after this next photo was taken, because I had to make some adjustments to install the ground bonding.

I’ll talk about how the panels are actually mounted to the racking in a future post.

Solar Racking Part 1 - Feet

Something needs to hold the solar panels to the roof, and that something is racking.  I went with a pretty new racking manufacturer called ToughTrac.  Since they’re so new there aren’t really any reviews on them yet, and using them was a bit of gamble.  The big thing they have going for them is price.  They are new to the market and have very competitively priced racking products (at least for pitched roof mount).  They don’t have much documentation or sophisticated planning tools and installation guides like some of the leaders in this field, but the owner and chief engineer were happy to answer any questions I had, and it seems like a pretty solid product.  I’d recommend giving them a shot.  They also have a mounting foot design that is much less likely to allow water leaks than simple L feet.  I ended up getting 3 sets of their lite rail system.  Here’s what showed up (click for big):

Feet and Rails

Fasteners

Clamps and Rail Mounts

They do not include the hardware to mount the feet to your roof.  In my case I did the first row with 3/8×4″ hot dip galvanized lag screws, and then switched to the same thing in 4 1/2″  lengths for extra piece of mind.  Mounting the feet is by and large the most time consuming process of installing the racking.  First I had to find and mark all of the rafters.  All the stud finders I tried failed miserably at this task, even ones with a deep scan mode.  I finally figured out an awesome system involving two people – one on the roof and one in the attic, and was able to mark all the rafter locations on the roof.  Now that I knew where the rafters were, it was time to get started installing the 41 feet.

Here is the pile of tools I used for this job:

I measured and marked out the location of each rail (2 rails per row, 6 total), and snapped a chalk line across the roof.

Then I marked the center of the rafter on the chalk line with a wax pencil (giant crayon) The chalkline is also shown here.

Drill baby, drill!  The attachment on the drill turns it into a drill press of sorts, and makes it stupidly easy to drill at a 90° to the slope of the roof.

I used a putty knife to break the bond between shingles.  I sanded off many a knuckle by accidentally dragging them across the roof as I scraped the shingle.

The foot will have to slide up under the shingles so that it is properly flashed.  Some racking manufacturers just use L brackets for feet, but I was concerned that those would be prone to leaking.  These feet are self flashing and should keep water well away from the drill holes.  In order to get them under the shingles, a few roofing nails had to be removed.

It seems like roofing nails don’t really like being in the roof, and getting them out is pretty easy.  Just slip a pry bar under the shingle and tap it with a hammer to pop up the nail.

Then pry it out.

Now that the path for the foot is clear, I needed to notch the shingle to accommodate the raised portion of the foot.  Here it is marked.

Then cut (tin snips are the perfect tool for this).

Then the chunk is removed.

The foot slides into place.  It’s ok if it’s not 100% centered in the cut.  It’s no less water proof, and no one will ever see it once the panels are on.

I filled up the pilot hole with a generous helping of Black Jack Roof Cement.  This will add even more water protection as well as lubing the lag bolts slightly when they are tightened.

A U bracket goes on top of the foot, followed by a galvanized lag screw and washer.  The rectangular channel in the U bracket allows for some adjustment if the foot holes aren’t 100% in line with each other.

Crank that guy down.

I left them a bit loose because the final tightening happened after I had the whole row in and aligned.

I squirted Black Jack under the edges of all the shingles I had to lift to glue them back down.  This probably wasn’t necessary, but at $3 a tube, it’s worth it if it prevents even a single leak.

One down, 40 to go!

Up next are the rails.