Solar Power

The design and construction of a small Grid-Tie Solar PV System

Solar Home Page

Introduction

I've tinkered with solar panels for some time on a small scale, and have been suprised at the amount of power that can be generated over a period of time, even in this temparate climate. On the whole, in an urban environment, I have found the net power produced by solar is greater than from my wind turbines although the peak output from wind can be much greater. My home built charge-controller has three inputs, one for solar, one for wind and a spare. Having the battery bank running at 12v is convenient for 12V appliances but in order to generate appreciable power through an inverter, the current drawn can be substantial. This means thick wires, heavy connections and bulky switches. One relatively easy solution is to push the generated power back onto the grid but grid-tie inverters often need about 40 volts DC to operate. So I find a 12 volt battery bank unsuitable. I could connect several batteries in series although increasing the voltage will mean my wind turbine will become useless as it is designed for 12 volt operation.

Well, I've played with wind turbines for a while now, so it was time to approach solar on a larger scale. With the new government feed-in-tariffs in the pipeline, the cost to install a solar system is coming down, in conjunction with the fall in price of solar panels and grid-tie equipment. So I set myself a small but achievable target:

Goal

Build a 600Wp Solar generator that is grid-tied.

Siting

The siting of the solar panels needed some consideration. Our house faces NE/SW but the SW side is shaded by two trees. The garage roof faces south although, being adjacent to the neighbour's house, means it will be shaded up until 10am. Similarly, it will be shaded by our house after 6pm. So, not a brilliant location but its the best I have. Being scared of heights also makes the garage a better option as it is only one story high!

Even though this is very modest grid-tie goal, it will still be expensive. Once you add up the costs of panels, grid tie equipment, and mounting mechanisms, the cost is looking at 4000-5000 pounds. Quite hard to justify that amount of money for the returns expected. However, I am sure there are savings to be made by substituting convenience for some extra effort. Fortunately, I managed to keep total costs down to 1800 pounds by using a non-standard mountoing system and by purchasing panels when they were on offer.

Panel Mounting

I pondered the problem of mounting the panels for some time. A common solution seems to be drilling through a few tiles and mounting brackets to which aluminium rails are bolted. The aluminium rails use special attachments that are used to clip the panels in place. I like the idea of the rails as these make mounting the panels easy although a solar mounting system seems grossly overpriced at over £1000. I do not like the idea of drilling through a roof slate as it compromises the integrity of the roof. Anything that involves sealant is just asking for trouble in my mind. A better solution is to come up with some form of bracketing arrangement that can be screwed to the top of the roof rafters and extend as a hook under the slates. As with any metals exposed to the elements, stainless steel or aluminium would be a good option here although galvanized steel is a cheaper alternative.

After taking off a couple of roof slates, I measured the slates, distance between rafters and cross-members, slate height and overlap and sat down to work out a solution. My father-in-faw, Frank, came up with an excellent idea which utilizes galvanized roof ties. One of these could be bent into such a shape that it can be screwed to a rafter and reach down under a tile and have a bracket sticking up which can be used to bolt a panel to it. The drawing below shows the basic idea.

There are 'standard' roof ties and the 'lite' version. The lite version can support a load of 8kg and I would use 4 brackets per 8kg panel. Some brackets are shared between panels but that still gives a minimum of 16kg load capacity which is more than enough for my needs. Being thinner, the lite ties will sit more neatly under the tiles than the thicker kind. The part of the tie that sticks up from the roof could be bent back through 90 degrees so that aluminium rails could be used. I then found I was back in silly-money-land as special solar brackets would be required. If, instead, they just stick up vertically, they can be bolted into the side of the panels using standard bolts, a much cheaper option. It means more work, more roofing, more brackets - but I find it hard to justify the cost of a 'standard' mouting solution.

There is enough room on the garage roof for 8 x 80W panels which would give 640W of peak power. In order to tie into the grid, a G83/1 compliant inverter is required. Having a poke around, the best value-for-money inverter I could find in this range was the Soladin 600 (see here). This will give a maximum of 525W AC power output so some solar power is lost during maximum sunlight but over-specing the panel side should give better production on cloudy days. Due to the UK climate, I think catering for bad weather is a sensible option!

I bought a couple of solar 80W solar panels and measured up to see how they would fit into the mass of roof measurements and whether the roof ties would be suitable. To mount 8 panels requires 20 brackets, 36 galvanized M8 bolts and lots of nuts. Total cost came in at £55, rather cheaper than the other options I investigated. Incidentally, I needed 60mm M8 bolts. I tried to buy these in stainless steel but had a terrible time trying to souce them. In the end, I had to settle for galvanized steel bolts due to availablity.

We made a prototype bracket and tried it on the roof to see how it would fit under the slates. It looked ok so we took 10 each and retired to our respective garages and hammered them into shape ready for the next weekend.

The brackets were drilled to accept 8mm bolts and offset spacers drilled for the the brackets that were shared between two panels. There was a space of about 70mm between panels which gave enough space to swing a spanner.

Regulations stipulate that solar panels must not be mounted more than 200mm proud of the roof. Since there was no bolting necessary underneath, I chose to keep them much lower to prevent the wind getting underneath them. The panels must also have an airgap to the roof to allow for convection cooling. I chose 3cm as that is just enough to get your hands under the edge to route wiring whilst still being very low on the roof. If you have big hands, 4cm might be a better option.

All Things Not Being Equal

As with most buildings, the roof was not perfect. The rafter spacing on the right hand side was different to that on the left hand side. There was also a small hump in the middle. These were not a problem in their own right, but had to be taken into account when fitting the panels. Those on the RHS would have different mounting centres compared to those on the LHS (to line up with the rafters). Here is the overall roof-plan:

Two separate lengths of wood were made up, with bolts pushed through at the mounting centres (one for LHS, one for RHS). These proved very valuble for making sure the panels were alighned correctly, and subsequently, as a good footing when climbing up the roof.

The next task was to lift the tiles and get the brackets up on the roof. I don't like heights so this was not a pleasant experience but after several hours shuffling slates around, the brackets had all been screwed into the appropriate rafters and the slates replaced. I did not want to go too high on the roof so as not to disturb the ridge tiles, but unfortunately, with limited roof space, I had to. This broke some of the cement on the ridge tiles so I spent all of the following day, sat on the rooftop, recementing ridge tiles! However, although a lot of work, I feel the bracketing solution is cost-effective and has not compromised the integrity of the roof. Here are all the brackets in place ready to accept the panels:

Installing the Panels

This turned out to be easier than anticipated. The panels had been pre-drilled and bolted ready for mating with the brackets. We started at the top and worked down the roof so as not to stand on anything glassy by accident! All the panels are connected in series using MC4 connectors and the excess wires and connectors are pushed into the channels that form part of the panels.

Trunking was predrilled and attached on the topmost and bottommost panels as to protect the cable between the two halves of the array. The ends of the cables were then fed through 4 inch wastepipe that was used as a kind of uber-trunking to get the cable from the roof and into the garage. 4 inch pipe was used as it allowed the cables (and the large MC4 connectors) to be fed around corners without too much difficulty.

Here are the panels in place on the brackets:

Internal Wiring

Once the cables were in the garage, they needed to wired through the inverter and up to the mains supply. This should all be done according to photovoltaic requirements (which I will not delve into here). The wiring back into the mains needs to be checked and approved by an electrician (part p).

All the electrical components were built onto a large wooden board which was mounted on the wall and then fed back into an MCB and back to the main consumer unit. I bought the various components from several sources (naviton, ebay, wickes, solar retailers etc). The black switch is the DC isolator switch. The red switch is the AC isolator switch. The Total Generation Meter is top left, and the soladin inverter is mounted in the centre.

And that was the bulk of the work. Now came the job of getting it working, measuring the output and making any improvements necessary

Teething Problems

The original design had 8 x80W panels in series. The Voc of these panels is 21v. So the Voc of the array is 8 x 21 = 168v (I measureed 164v). The maximum input voltage for the soladin is 155v. When I have used panels to charge batteries, the output voltage of the panels are pulled down to the operating voltage of the battery (eg 21v is pulled down to 12v under load). I expected the soladin inverter to do the same so had overspeced the voltage. This was a wrong assumption, I subsequently found that the soladin throttles current, not voltage. The result was that the solidin complained that it's input voltage was too high. To fix the problem, I took one of the panels out of the series string to bring the voltage down to 7 x 21 = 147v (measured 142v). This worked fine and I brought the extra leads from the unused panel back into the garage for use at a later date.

MPP Tracking

The soladin inverter has MPP tracking. MPP stands for Maximum Power Point. This occurs on solar panels where there is optimum power generated at a certain voltage, for example 17v is often quoted. I was expecting the inverter to load the panels to bring down the voltage to this sweet spot, but that does not appear to be the case. However, the soladin spec states MPP is active between 60v and 125v so the current architecture means that MPP tracking is not engaging. I believe the best soltion is to replace the extra panel and split the array into two parallel strings of 4 panels each. This will bring the Voc of the array down to 4 x 21 = 84 volts which lies in the MPP range. However, the maximum current generated by two parallel string would be up to 10A, which is greater than the maximum of 8A specified for the inverter. Now I know the device throttles current and not voltage, this should not be a problem. Having two strings should give me a these advantages:

• The inverter will use MPP tracking giving an output improvement
• The array should be less affected by shading giving an output improvement in the mornings.
• The extra panel can be incorperated giving output improvement in cloudy or shaded conditions.

Shading

The April weather has been excellent and I have been fortunate in having several cloudless days to monitor the power generated by the solar array. An instantaneous clip-on power meter has been used to monitor the output from the inverter. When all panels are exposed to the sun, approx 400Wp is generated. It should be able to go up to 525W. At this stage I do not know if it is weaker April sunshine or not being in the MPP range that is causing the descrepancy.

However, I have found partial shading has a much more significant effect than I would have expected. When it is daylight, but no sun is shining directly on the panels, they generate about 65Wp. If the lefthand panels are in full sun, the RH side in shade, 70W is generated. I would have expected this to be in the region of 150W. If the LH panels are in sunshine and the RH panels are 25% shaded (so most of the panels are in sun), I still only see about 70W. When the last part of shading disappears, then, power jumps up towards 400Wp quite quickly. At this time of year, the panels leave all shade at 12:00 whereas the LH side is fully illunated by 10:00 so there is a lot of potentail power being lost.

It appears that a shaded panel becomes a high-resistance load, thereby throttling back the current generated by the illuminated panels and by splitting the array into two parallel arrays of 4 may solve this problem. Before clambering on the roof, I will do some experiments on the ground to see if the situation can be improved or whether I just live with what I have.