Creating power for your home, off grid.
Emphasis on nuts-n-bolts, hands-on projects.
As the new inverter is basically physically exactly the same as the old one, it used all the same power couplings and so was a "drop-in" replacement. I'm in two minds as to whether to sell the old one or keep it "just in case"... You never know when a 12V 1kW pure sine inverter will come in handy.
This now means that I can use all of the 604Wp of solar panels installed and still have some headroom for more as the two controllers will handle up to 800Wp now.
I also put in an order for the optional battery temperature sensors for the charge controllers, as the built-in ones on the Morninstars were getting false readings from the heat sinks on the controller heating up the whole unit. I also ordered the meterbus adapter so I can plug a PC into the Morningstars to custom programme them and download the 30 days of data logging they keep in on-board memory.
As the days are getting longer now, I don't need to rely on the mains battery charger any more but I'll have to get or build a 24V charger before next winter as my 30A 12V power supply won't work.
I had to bring the Sharp panels indoors tonight as there was a big storm brewing this evening and they were still just perched on those garden chairs.
I've started making up the roof mount for these panels but the main problem was waiting for a sunny day to paint the wood with preservative that you can only do on a day warmer than 10C.
Annoyingly, some cowboy car park subcontractors at the hardware store (in a retail park) have sent me a parking ticket for 50 Pounds because I spent 3 hours in the shop trying to find suitable materials. They have a car licence plate recognition thing and if you are in the car park for more than 2 hours they assume that you were parking there to walk into town. I've complained to the hardware store head office to get them to cancel the ticket or refund me the charge as I was a customer of theirs the whole time and I did spend more than 115 Pounds in their f*&$ing shop
In other news... My long time CRT monitor finally blew up and I'm writing this on a used LCD screen I picked up for 30 Pounds at a car boot sale on Sunday... Oh, yes! The car boot sale season has started again . Now there's lots of cheap stuff being sold off from companies going bust - hence the van load of cheap computers and LCD screens at the weekend.
Even though the charge controller is designed to stop the battery back feeding the panels at night, during the day you can get significant back feeding between unshaded and shaded panels. This happens to me as some panels are part shaded by the garage roof or trees in the morning and others are in full sun.
It might even be why I've had two panels fail.
The instructions for the new Sharp panels were clear on the matter. It said that you could have a maximum of two panels in parallel without diodes and fuses but that more than that would require each series string to have a blocking diode and a fuse.
I had a look around for Schottky diodes but most have too low breakdown voltages of under 30V. Only some really heavy duty 16A ones had a breakdown voltage of 60V. The amorphous panels have a Voc of about 26V in strong light and so could put a 52V reverse voltage on a badly shaded pair. So I opted for regular silicon rectifiers that have a 1000V breakdown voltage and much lower reverse leakage. A 0.8V diode drop at 17.5V would be a problem but at 35V it's less of a big deal.
In other news... I got the parking ticket cancelled
I decided on a fixed tilt angle of about 45 degrees as the summer should provide more power than I can usefully use (yeah, right) and it's the winter that really needs the extra power. Mid-summer, the Sun climbs to 68 degrees above the horizon here while mid-winter it sinks to 15 degrees. Today it was really sunny and the battery bank got topped off by 11am and then I was rummaging around for other things to charge to make the most of the Sun. I charged another laptop, two phones, both batteries from my drill, my SLR camera, my Sony digicam snapper, and ran the works laptop the whole day with all the power saving turned off
To be honest, part of the decision to go with 45 degrees was the fact that the panels are about 99cm wide and it's easy to buy pre-cut aluminium extrusions in 1m lengths so I could easily make a stand with two 1m lengths, making a right-angle at the top.
The local hardware store was a bit short of stock so I had to improvise a bit.
For the rails on the panels I used two different profile L-shapes. One has a long side and the other has a short side. Bolted together at the ends by the M8 bolts that go into the panel frame, they made a unequal U-shape rail that allowed side bolts to go in for mounting the panel to a bracket at the bottom and the "hinge" at the top for the leg. I put a M5 bolt in the middle to keep them in line and stop them from warping. The U shape gives it strength against bowing and using two of them meant the overall U shape base was 3mm thick (each extrusion was 1.5mm). I let the extra 1cm of extrusions hang out over the base of the panel to give it slightly more clearance over the brackets.
I bought some 1m x 25mm x 1.5mm box section aluminium for the legs. This gives them strength in compression in all directions. At the top, the U shape rails long side gives a mounting surface. The slight overlap of the two L shapes meant that there is a lip that the box section ends can butt up against.
The base is some 70mm x 90mm building timber that I treated with garden furniture woodstain preservative. This is important as the wood will otherwise shrink / expand and rot.
The brackets were just heavy duty steel shelf right angles. They are painted but I'll probably give them a coat of Hammerite to stop them rusting. All the mounting bolts are nylon locknuts and zinc plated (I couldn't get stainless steel ones).
For the cross struts on the base I used some old picket fence boards that I picked up at a car boot sale last year and painted them. These stop the main base timbers from rolling over or pushing apart with the force of the panels on the brackets and legs.
I had intended to put the whole thing on the garage roof and use the roof as the cross strut by bolting the base timbers to the roof but it just made it easier to assemble on the ground to include cross struts... As it turns out, the whole thing is now far too heavy for me to lift on to the roof without help so it might permanently remain on the ground! So in lieu of bolting it to the roof, I piled some bricks on the base to keep it planted on windy days, although the last few days have been almost like summer than the end of winter. Today it got up to 17C outside.
The total cost was about 130 Pounds but with quite a lot of left-over nuts and bolts as they sell them in packs of 15 or 20. They're always handy. The main cost was the aluminium extrusions, which are pretty expensive these days.
Here's the other 60W amorphous array sitting at the edge of the garage, basking in the shadow-free Sun.
Meanwhile indoors, I rewired the two 12V LED wall lights in series so they could work on 24V and vandalised a variable power supply I bought at a car boot sale last year to convert it from a mains powered supply to a 24V DC to 2-15V supply. This replaces the old solar charger that I was using for powering 12V gadgets. Helpfully, it has a resettable 2A thermal breaker built in. With a cigar lighter socket wired on to the terminals I could plug in my NiMH cell charger and the mobile phone charger.
I need not have bothered though as at Sunday's car boot sale what should I find but an 8A truckers 24-12V switch-mode converter! It's truly amazing what obscure things you can find at car boot sales...
The Sharp array at 46.4V is working very well, charging better in low light than the amorphous array. In fact, on a couple of clear days recently, I've seen the 340W Sharp array put out 364W (just over 107% of nominal output). Now that it's up on the roof, I've seen more total power during the day as there is no shading from the garden fences. Today the battery bank made absorption phase by 11am and by mid-afternoon was sitting in float mode despite continuous draw from the laptop and the odd cup of tea made with a special low-power travel kettle I bought (it's only 850W, compared to 2-3kW for most modern kettles, so it can work on my 1kW inverter).
Having just reclaimed the lawn, I succumbed to another sale of the 60W amorphous kits! This time they were selling them in the store and not just from their web site so I could open up the box and check out all the panels in the sunny car park.
The diode losses and wire losses mean that the amorphous array is sometimes struggling to maintain output as it's maximum power point is now sometimes as low as just 2 Volts above the battery voltage when the pack is equalising at nearly 30V. By buying another set of 4 of the 15W panels, I can re-wire the array for 8 series triplets in parallel, running at 51.7V instead of 34.2V. This will push the array into the optimal conversion efficiency Voltage range for the Morningstar when charging the 24V battery bank.
The only problem with this plan is that the Voc of the triplets will be around 75V and right at the limit of the Morningstar controller. It will work up to 70V but above that it starts to shut down until at 75V it cuts out with a protection alarm. I'll get around this by connecting a string of 5W Zener diodes across the array output. By using 3x 15V and 2x 12V Zener diodes in series, I can clamp the array output at 69V. Near the Voc of the array, it has very little power capability (in theory, at Voc a solar panel can only put out zero Amps). As the Zener diodes are in series, they share the power dissipation (330mA into the 15V diode drop is only 5W and only 4W through the 12V diode). In all, the diode string should be able to dissipate up to 23W of power - more than enough to keep the array Voc below 70V. Of course, when the Morningstar is loading the array at its maximum power point of about 52V, the Zener diode string will effectively do nothing and draw no power at all.
I did consider making a "proper" shunt regulator with a Zener diode reference and a couple of power transistors but 5W Zener diodes are cheap and I won't even need a circuit board with this solution. It can only work with solar panels and only in this region of their operation though as they have such a curious IV curve that suddenly tails off past the Vmp point. A regular power supply would just blow up the diodes (as would choosing a Zener voltage lower down the curve where hundreds of Watts are available). That reminds me... I must put a fuse in the diode string so that if one of the diodes goes short circuit the others won't catch fire with the additional power they'd be subjected to.
The frames for the 15W panels are somewhat modular and I had a spare T bar (the broken one that I mended with super-glue) to make up a pair of two panel expansion units. All I had to do to the existing four panel unit was unbolt the end piece and replace it with the two panel expansion.
You can see the new six panel array here along with a new steel cored plastic rope as an added safety measure to stop the Sharp array taking off in any storms. It is secured to the roof support girder underneath.
I lengthened all the wires so they could comfortably reach the waterproof junction box and used two independent diodes for the two triplets of panels connected in parallel. This replaces the single diode at the main junction box at the end of the long feed wire. This should lower the diode forward voltage drop, as only half the current is now passing through each diode. It also means that part shading of this array will not cause local back-feeding through the other triplet.
Here's how it's wired up now:
You'll notice that some of the 12W panels have built-in blocking diodes. Despite this, they still put out 17.5V as their maximum power point as they have more cells than the 15W modules. I decided to still use external diodes as I don't know the breakdown voltage of the internal ones. You can bet they will be cheap diodes with less than 50V reverse breakdown rating (as they were only intended to work at 25Voc) so I decided to shore them up with the 1000V 1N5408 diodes I had already installed on the other pairs of panels.
The string of zener diodes is working well, capping the open circuit voltage at about 68V. The diodes have a voltage tolerance of 5%.
With the sunny weather lately, I'd been watching the battery get full at lunchtime and wondering what to do with all the power being wasted... Most of the kitchen is a non-starter as things in there consume way too much power. The fridge might be possible but there isn't enough solar power to run the thing 24x7 and if I forget to switch it back to grid power, the freezer would melt.
Then I realised that the ideal candidate would be the computer room upstairs. The comms kit and server are on 24x7 and chomp their way though about 3kWh per day and are fed by a UPS. That means I can swap the plug over from grid to solar and back again without interrupting the computer or network, as they're happy to run from the UPS for a few seconds.
So I ran a new extension of the solar power up to the computer room and put a trailing socket near the UPS under my desk. Now when it's sunny enough and there's spare power about, I can switch the computer room over to solar power. I'm toying with the idea of using a 240V coil change-over relay to do the switching automatically (or at least with a wall switch) so I don't have to crawl under the desk to swap the power lines over.
In the first two days, I increased my daily solar power utilisation from an average of 0.7kWh to 1.5kWh, using solar power for the computer room from about lunchtime (when the battery reached absorption mode) until about 9pm, when the computer room started to eat into the battery too much (what with all the lights, TV and kettle being used in the evening as well).
If I was really in the mood for a project, I could have the server monitor the solar power level and battery state by the serial port on the Morningstar controller. The server could then choose its own power source by activating the mains transfer relay automatically!
Since the start of December, I've been keeping detailed statistics on total energy used in my home (Gas, Electricity and Solar) and you can see the results below.
December and January are pretty much a dead loss for the solar system, with big negative spikes where I had to use grid power to charge the batteries to keep them from rotting. However, from mid-February, things pick up and I've not had to use any grid power to charge the batteries and have been able to extract progressively more power from the solar system.
But as the total energy use graph shows, solar power is only accounting for a tiny fraction of my total energy use (dominated by gas central heating). The peak daily offset for electricity has been that really sunny day a couple of days ago when I ran the computer room from the solar power (making just over 12% of my total electricity consumed that day). The best weekly total offset is only about 5.2% though.
The last graph shows the daily solar utilisation and a lifetime electricity offset trend line (total solar kWh as a proportion of total electricity consumed since December).
Of course, the figures are somewhat skewed as over the last couple of months I've installed a load more solar panels and changed their configuration and so on. But it's still an interesting record.
You can see some changes I made to energy use as a result of keeping records. Around mid-February, I decided to switch from gas water heating to electric water heating. The gas boiler circulates hot water to a tank upstairs and heats the water in the tank by a heat exchanger coil. Not very efficient. By switching to direct electric water heating in the tank, I increased my daily electricity consumption by about 5kWh but reduced the gas consumption by some 20kWh. Even though electricity is three times more expensive than gas, it saved me money and reduced my total energy consumption.
I would caution against using a double-throw relay or contactor to switch between mains and inverter power. I have done this, and you'll run into problems. If the two power sources aren't closely in phase when the relay drops out, there will be a ~very~ significant spike in your load systems. What happens is that any 50Hz transformers in appliances will have a hysterisis of magnetic flux built up from the one power source. Switching to the other source before this hysterisis has died (or crossed to zero) will result in a hammering EMF effect that can blow fuses or damage equipment.
I have a radio and TV translator site that has this kind of a setup to switch the equipment over to generator power, then back to mains when the power is restored. I regularly have to make a 4x4 trip to the top of the hill to replace blown fuses (sometimes by the dozen) in the equipment due to the kickback of the EMF.
In your case, things would be even worse. My equipment runs on 120 volts. If the mains power comes back on when the generator is 180Â° out of phase, a kickback of 240 volts is produced. In your equipment, the voltage would be 480 volts! This is not a static charge, but has the full reactive power of a big iron core transformer with a rapidly collapsing magnetic field behind it! Think spark ignition coil on a car, lot of potential discharge.
The only really safe way to switch between sources is to either make them 100% in phase, or give a moment's rest between disconnect/reconnect, which would require two relays and a time delay (a second would be more than enough time).
It almost certainly would have caused trouble as the UPS has a big transformer in it for powering its electronics and charging the 12V battery.
While I could design something with a bunch of op-amps and relays, there are more possibilities for things to go badly wrong. The idea was that the single change-over relay couldn't possibly connect both AC rails together simultaneously because the contacts all share the same mechanical armature. If it gets stuck because a contact has welded, then all that happens is that it gets stuck on one set of rails. As soon as you have two relays, there are all sorts of possible race or failure conditions that mean the two rails can get connected simultaneously by accident (a pretty expensive accident).
So the cheapest and safest thing for now is the good old two plugs manual change-over... At least until I can design and test something properly.
This energy efficiency stuff is even creeping into my day job [at this point there are bound to be raised eyebrows... "you mean this isn't his day job?"]. My current client is a sports car manufacturer that wants to save money (don't they all?) on their IT spend with us and try to grab some "green" cash if they can show the government that they are going... err.... "green". Not easy when their reason for existing is to make fast cars. However, I've put together some figures that show that if they let us install three big new servers, they can virtualise 38 of their 50 existing servers and save on our support costs but also save 45% of their electricity bill (more when they factor in the reduced need for air conditioning in the data centre).
We're also trying to get them to replace 700 old PCs with thin clients that look like TV set-top boxes and use a virtual PC in the data centre. Each old PC uses about 150W and a thin client uses about 70W (depending on what screen they use).
With the savings in consumption they make, they can afford to pay the 0.5p per unit premium to buy electricity at the "100% renewable sources" tariff to negate their IT CO2 altogether.
That's how I'm really trying to "Change the World".
When its time to change power sources, pull in the normally closed relay in series with the load while the double pole relay is doing the source swap. A little bit of delay in allowing the series relay to release and reconnect the load allows the magnetic flux to die and everything is fine. If you end up with welded contact in the supply relay, no big deal, you just get a short glitch in the equipment supply instead of a source change. If the series relay welds, well, you end up with what you were proposing in the first place, which is better than mixing the two supply voltages with each other.
No sooner had I mentioned that I have a project this summer to build a 24V battery charger than what should I pick up at the first of the mega car boot sales this year but a 3A fully automatic 24V traction battery charger for less than the price of a pint of milk!
The guy selling it had just binned an electric motorbike that one of his kids had broken and had no use for the battery charger any more. I didn't let on that it was worth about 50 Pounds.
It's a proper 3 stage charger with current monitoring of the absorption end phase so when it gets to about 200mA at 29V, it goes into 13.6V float and a LED comes on to say so. It's fully short circuit proof and I can even see a couple of presets inside that must allow you to fine tune the absorption and float set points or the transition current to float.
Obviously a 3A charger isn't going to top off a 220Ah bank in a hurry but it can be left on permanently and will keep the bank from rotting during the long winter weeks with next to no sun power.
Despite my cunning plan to limit the open circuit voltage to below 70V, the zener shunted array caused the charge controller to act a bit weird. It would go into absorption charge mode before the other charger (they used to be much more in sync) and when it got really sunny the MPPT scan would get confused by the voltage clamp. It got confused sometimes and would latch on to all sorts of weird "maximum power" voltages that gave sub-maximal power.
So although I was seeing lower losses on the line from the far end of the garage to the controller with the thing running at a higher voltage, the total power delivered to the battery was sometimes lower than expected. Re-wiring the whole thing for pairs again at 35V restored normal operation. The Sharp array still significantly outperforms the amorphous array, even though they are nominally similar powers (the Sharp pair being 340W and the amorphous array being 324W).
I also noticed that the maximum power point it mostly settled on was about 45V and not 51V, so adding the third panel in series probably caused the internal resistance of these thin film panels to start to degrade the square fill-factor of their combined I-V curve. There was also much more shading problems than before with partial shading of any of a triplet of panels degrading the total output much more than with only pairs of panels.
At least my eBay activities have netted me over 100 Pounds in selling off the various batteries, LED lamps and solar controllers that came with these amorphous kits so some of the panels have been effectively free.
It was an interesting experiment and goes to show that the proponents of integrated inverter designs might be on to something. They use local inverters on individual 200W panels in contrast to the traditional model of running 20 panels in series to give you 480VDC into a big central grid tied inverter. The problems of series resistance and shading must get magnified a lot.
Their idea is that low power inverters are cheap enough to install one on each panel at the roof and just run the mains down to the house. As each panel is working in isolation, you don't get the big series losses of a traditional system and because there are multiple inverters, you aren't at risk of losing all your power if one fails. They also use the mounting rails as the DC power rails so the panels don't even have frames and don't need junction boxes or wiring to the inverters. Makes the panels cheaper as the aluminium frames are a big part of their cost. It's safer too because the whole DC side is only operating at 24V.
The SunSine inverters can also all be connected via data cables to a PC that can monitor their production and set up alerts for non- or under-functioning panels. Repairing defective inverters and/or panels is a lot easier when the entire system doesn't need to be dismantled to trace defective panels or remove the master inverter.
That said, the small single-panel inverters are completely useless without a functioning utility grid to dump the power into. If you need backup power for utility failures, you need batteries, and they don't invert from battery power...
Curiously, those Enphase inverters will only work with the higher voltage panels. They have a quite high minimum voltage of 25-40Vmp. The new ranges of Sharp poly and mono panels are 23.2V but the ones being phased out are the 35V mono ones.
I've also just found this Mastervolt plug-in grid tie inverter for small systems up to 700Wp:
http://www.mastervolt.com/view_product. ... ro_id=5920
I've seen it going for under 400 Pounds on eBay and it's self-contained with the certifications required for grid connection in the UK but just plugs into a wall socket. Much cheaper than the SunnyBoy fixed installation inverters I've seen before. If I had another Sharp panel I could use it as it will work with strings as low as 45VDC but most of my power consumption is at night and without a meter upgrade and an agreed tariff for selling the power back to the utility (the government is dragging it's heels on that) there's little point in me doing it. In fact, it might cost me as the digital meter installed might be the type that runs forwards even if the power is flowing out of the house.
The two biggest loads I can't service with my current system are the cooker and the water heater as both require more than 1kW of power. The small grid-tie inverter would allow me to partially offset a 3kW load but both these loads get used primarily in the evening.
A grid tied inverter would however make a useful contribution to the fridge-freezer that runs 24/7. The fridge is a difficult load to drive from an isolated battery system like mine and doesn't like being interrupted for a change-over if the compressor is running at the time.
It's not even June yet and the array isn't angled optimally for the noon sun either, being set for about 45 degrees (about the right angle for noon in March).
It's been quite cool the last few days and partly cloudy. I see this surge in their output when it's cold and the air is clear so that when the clouds do part, the full force of the sun hits the cold panels and for a minute or so they put out current way above their rating. If the sun persists though the panels heat up and the output tails off a bit (but still above the nominal 340W rating at around 360W).
I wonder if glueing lots of heat sinks to the back would help them work better... I feel a mad invention coming on. PV panels encased in an evacuated glass fronted box with CPU heat sinks glued to the back and circulating water. Thereby combining a solar water heater with a PV system and the coolant makes the PV panels more efficient.
It's been tried. Doesn't work all that well. Water for domestic use is ~much~ too hot to act as a coolant for PV panels. Also, having PV and DHW systems integrated means that something valuable gets tossed out when to other half fails. In every case, it's more economical to have separate systems.AccordGuy wrote:PV panels encased in an evacuated glass fronted box with CPU heat sinks glued to the back and circulating water. Thereby combining a solar water heater with a PV system and the coolant makes the PV panels more efficient.
It often gets suggested that having a cooling mist of water flowing over the panels would increase output, but it gets pointed out just as often that the circulating pump for such a system would consume at least as much power as the additional cooling would produce. Just giving your panels free air circulation on the back side is about the most effective thing you can do for the money (panels not laying flat on a surface/roof with no air space).
The significant increase in output that you see is only partially caused by the panel being cool. Light is focused and intensified when the sun peeks from behind a cloud (going in or coming out). This is known as "edge of cloud effect". The fact that your panels are cool is contributing, but it's the concentration of sunlight that is doing most of the work.
Users browsing this forum: No registered users and 1 guest