Cobbled together PV in W.Sussex

[Outtasight]

Rob,

Thanks, I've seen them but two pallets of flooded cells weighing close to 1200kg isn't going to be easy to hide behind the comfy chair (where my 180Ah bank hides now)  .  I changed to gel packs to avoid being gassed.  Even the blue SLA 500Ah cells the same guy sells are too big physically to fit there. 

Eccentric,

Thanks for the diagram.  That's sort of what I was thinking of but thinking about it some more today, I'm coming to the view that the most important things to measure would be the available power of the array (quite easy to do) and the power demand on the array (already have this info to hand).  Then it's just simple...

if available power - (power demand + 650W) > 0 then turn the heater on.

This then makes battery charging and other loads a lumped variable that I don't need to care about the detail of.  Time to experiment with some bell wire and that old 6V amorphous panel hanging out the bathroom window...

[guydewdney]

Thats exactly what alan made for me....

http://www.navitron.org.uk/forum/index.php/topic,9072.15.html

[Outtasight]

I remember seeing that thread and the cool resulting gadget.  This incarnation is similar but different.

Similar in lots of ways as it's all based on a PIC that does the communicating (although the rules logic is in a PC attached) and in that the aim is to maximise use of solar power and minimise grid import.  The actual goals are slightly different though as I'm off grid.  A grid tied inverter always puts out whatever current it can as it's designed to export power so its available power is whatever it's putting out at the moment. Its primary load is the grid.  Your gadget looks to take all that available power and use it locally, rather than export it. 

I'm having the opposite problem in that unless a load is presented to the inverter, nothing is generated and the available power of the array is latent power that goes nowhere.  Unless I have some way to measure the actual sunlight, I've got no idea if there will be a torrent of power when I "open the tap", or just a dribble.  I also have to manage the battery so that if there's only a dribble of power available, it has to go into the battery first but after the battery gets full I need a way to use the latent power "overflow".

Even a dribble is fine though.  Keeping with the plumbing theme, I could make use of even a dribble of power as I can store it in the battery (like a toilet cistern) and when it gets full, I can "flush" it for a few seconds burst of high rate flow before having to wait a few minutes to recharge the "tank".

Up until now I've had to do it manually by looking out the window, watching the battery gauges and then running upstairs to "flush the loo".

You could say then that what I'm trying to invent (in true Wallace style) is a Flush-o-matic Dump Valve^TM



Today's tinkering brought some more logic to stop the heater cycling in low light when the battery is near float voltage (but not actually full yet).  It now reads the battery status from the TriStar charge mode and prioritises absorption while scavenging power if there is more than 900W available towards the end of the MPPT bulk charge or absorption phases.  This will limit the peaks (admittedly quite rare though) where the battery might be force fed 50A (somewhat hot for the size of battery).  Now, if the power looks low because the battery is nearly full and the heater turns on as we've guessed there's 650W going spare; two things will happen.  Either the guess was correct and the battery will continue to sit at the absorption set point with the excess  enough to drive the load or we guessed wrong and the volts will drop down from the absorption set point, the TriStar will drop back into MPPT mode and this will then turn the heater off and change the priority back to charging.  Some charge will still be lost from the battery but it should spend more time at or above float level and spend more time in the absorption mode on marginal light days.

The cunning part is using the charge status flags from the TriStar, as the float and absorption voltages aren't fixed but temperature compensated by a sensor on the battery bank.  So I don't need to measure the battery temperature at the load manager as that's already been factored into the status flags for me.  I still use a global low voltage cut-out of 27V because the Float flag remains set even if the battery starts to discharge.  The TriStar has a timer that counts the cumulative time below the float set point and when it's been discharging for a set time it drops back to MPPT charge mode again.  But I don't want to use more than a few seconds here and there of battery power for the heater so when the float flag is set I turn on the heater and watch the battery volts to make sure it stays above 27V so that I know I'm only using solar power for the heater.

As a quick measure of performance, I added a runtime counter that increments every time the heater has been on for one logger interval.  By multiplying the counter by whatever the interval period is (I usually set it to 15s but it could be anything) I can easily work out the total time the heater has been on. But even without working it out, I can see if today is a high-scoring day.

[guydewdney]

to simplify - you have a battery charger with three LEDs on it - bulk, absorb, and float.

You have detected that the float LED is on, and that turns on the heater? when the bulk light comes on, you turn off the heater?

[billi]

sounds complicated to me    has your new chargecontroller not an  in built secondary control relay  , that can handle these "dump load" ideas ?

Billi

[Mostie]

What about fixing a LUX sensor next to the panels and connecting it to your velleman kit?

[fred bloggs]

Outtasight

Just sent you a PM with contact details, I've just read you're further posts and it has given me a lot of food for thought as I want to run my system "off grid" although we are are grid connected. The panels will be on my garage roof (man kennel) and all the equipment will be inside, its a good excuse to be allowed to play

Many thanks

Fred

[rob26440]

I changed to gel packs to avoid being gassed.

Outtasight,

Some 320 Amp 2V YUASA gel cells for sale from the same place item: 220580804843
(15mins from where I live.  Wish I had the roof/garden space for PVs! )

[billi]

rob  

Wish I had the roof/garden space for PVs!

Nothing is impossible  


But these batteries ?  

Oh i have to modify  
i thought  it was for one battery at 2 volt , but its for 6 .... but still the supplier writes  like the yellow press

Billi

[rob26440]

Nice one billi!  I can't think of a suitable pun to associate hedgehogs with solar panels.  In fact, I haven't seen any hedgehogs in the garden yet this year - must be still hibernating.

[petertc]

Watch out for the solar spikes 

[Outtasight]

to simplify - you have a battery charger with three LEDs on it - bulk, absorb, and float.

You have detected that the float LED is on, and that turns on the heater? when the bulk light comes on, you turn off the heater?

Sort of... in an over simplified kinda way. 

There are two chargers.  Both have LEDs.  I don't read any of them because they don't show the charge mode in a simple way and besides it's easier to read the data from the MODBUS ports.  I then calculate their combined output power and make decisions based on that plus the charge mode, battery voltage and whether the load is currently on or off (having prior knowledge that turning the load on will present a 650W demand means I can try to anticipate what happens if I were to turn it on).

But don't worry, if that appears too simple then fear not, it's still under development...  This week has been busy with work but with the Easter Weekend comes some more software tinkering and a solar input sensor.

sounds complicated to me    has your new chargecontroller not an  in built secondary control relay  , that can handle these "dump load" ideas ?

Morningstar do sell a relay driver that works with the TriStar and a normal non-MPPT TriStar can drive a DC dump load directly.  But the relay driver is pretty expensive (£133) and you might need a hub as well.  My controller cost £30 plus some monkeying around with VB and left-overs from my parts bin.  The MPPT TriStar can't drive a dump load as it is a series switching battery regulator and besides my dump load is a 3kW 230V AC heater on the other side of the house.

In addition, as well as controlling an AC load remotely (as it is controlled over an IP network, the load could could be anywhere in the world with an internet connection), it also does system data logging.  Being fully programmable, I can fiddle about with just about anything.  Another project in the pipeline is to have the computers upstairs choose their own power supply.  There's enough relay outputs to make a source change-over switch for the computer room UPS that will implement the safety delay required to stop the EMF bounce-back that can happen if you just use a change-over relay on a transformer with unsynchronised AC phases.  One supply needs to turn off, wait for the magnetic field to collapse before re-energising the transformer with the alternate AC supply.

Some 320 Amp 2V YUASA gel cells for sale from the same place item: 220580804843
(15mins from where I live.  Wish I had the roof/garden space for PVs! )

Yeah, I've been watching that seller for a while and these 320Ah packs would be ok if they were NPC ones but they are EN ones that aren't rated for cyclic use.  They'll only last about 250 cycles from new.  Still they are cheap.  I found a UK distributor for the Deka ones I have (now branded as MK Battery).  They're expensive to buy new though (£183+VAT and delivery each).

But back to this week's developments...

Playing about with a lux sensor to improve the estimation of available power.  At first I thought of using a small solar panel to generate a voltage but I happened to have an old light dependent resistor so it was easy to just wire it into the A/D input that has a 5V reference voltage anyway.  I made the LDR just form the lower half of a voltage divider.

LDRs are commonly used in dusk to dawn sensors... with good reason, as I discovered.  They are very sensitive at low light levels - changing from a couple of megohms to about 50 ohms in full sun.  But they aren't particularly linear and so in all but dim light I just got a saturated reading so couldn't tell between 100W solar input and 1kW.  I had to attenuate the light reaching the sensor so that it would work in full sun but at its dusk range.

Cockle shells are a useful thing here.  Impervious to UV (unlike plastic) and whitish in colour with a high opacity (but not completely so), they make good light filters.  I glued the sensor to the inside of a small one and then silicone gooed a bigger shell on top to cut the light even more.  Not shown in the pictures is the final version with outdoor white gloss paint on the top to filter / reflect a bit more light to get the output "just so".  I also filled in the back with waterproof outdoor Polyfilla to seal it up and give it some weight so it wouldn't flap about out on the windowsill where it looks up at the sky.  The shells also serve as a diffuser (being curved) so that the sensor responds smoothly to light hitting it from a variety of angles.

 

Then I just had to modify the software to read the A/D port, scale the output to make 0 read dark and log the array output against the sensor data for a while to get the thresholds.  Actually, logging the data in Excel and then using the graph trend fit function, I could get a reasonable fit with a cubic term quadratic equation that I might put into the program so that it can estimate the solar input in Watts rather than just numbers but for now this works well enough.



The load now cycles much less, although there's quite a big margin of latitude between what the sensor detects and what the array actually produces due to shading of parts of it at early morning or late afternoon.

In other news... I finally got round to making an A frame stand for the old 48W array (rather than just propping it up on a water butt).  I also wheeled out some of the spare 15W panels and lashed them to some wood and threw them up on the roof, boosting installed capacity to almost 1.5kWp (in theory).

[Outtasight]

I've been away for a few weeks.

We were supposed to be going home to Japan for a month but the ash cloud over the UK put paid to that plan.  So it was a two week break in Germany and then two weeks at home "staycationing".  While out in Germany, I saw literally hundreds of large PV systems installed on houses and businesses and this sort of got me buying stuff again...

Most of the hotels we stayed in had free internet so I ended bidding on some more BP solar panels...

First up was a newish used BP 350J 50W panel for £100.  I thought it was a 12V panel so was bidding on two but only won one of them.  I'd resigned myself to having another "spare" panel looking for a partner but on returning home, I discovered that it's a reconfigurable panel.  That is, it's wired as two 25W 12V panels in the junction box so it could be converted easily to run as 24V.  So up it went on the wall next to the BP 3160S panel.  It's almost the same length so nearly looks like it's supposed to be there.



Then I found a seller that was doing brand new BP 380J 80W 12V panels for £165 in multiples of 10 but he was prepared to sell me five at the same rate so off we went to Yoorkshire to collect them!  I think they were so cheap because the 380J's aren't on the MCS list so they're no good for FITs installs.  Obviously, now I've got a spare 80W panel looking for a partner but to be honest, I'm running out of places to put these bigger panels...  Maybe I'll sell it to recoup the petrol money it cost to get to Yorkshire or maybe I'll keep it "just in case".



Having another week off work this week, I set to building the mounts for them to go up on the garage to replace the 6x 15W panels.  Half the frame for this job was already to hand in the form of the stainless steel bits from the used BP 380U panels I bought back in February and a couple of good outdoor treated bits of wood I picked up for like £2 at a car boot a couple of months ago (because a good bit of wood always comes in handy later ).  The wiring was also more of that cheap car boot cooker cable with the old cable added on in parallel to give at least 8.5mm² in both directions to the junction box at the other end of the shed roof...  Seemed to be doing the job as on a slightly hazy afternoon, the new group of four were pumping out at least 7.5A into the "network".

 

I designed this frame to have a lower pitch than the one for the Sharp panels behind it so that in the winter it won't cast a shadow on the Sharp pair.

That lot brings me up just short of 1.8kWp installed now and the 15W plastic panels will become car boot sale fodder...

All that remained was to tweak the load manager settings to take into account that extra available power and so the immersion heater ran for almost five and half hours today - time for a relaxing hot bath, me thinks.

[StBarnabas]

Outta
you put me to shame. Such enthusiasm! superb...

[Outtasight]

Ha hahahaha....  MMWwuuHAHAHAHAHA!!!!  Evil genius type noises from the trap door (remember that claymation series with Willie Rushton doing the voices? - I bought the DVD compilation the other day.)

The problem of having too much charge current reared it's head over the weekend.  The 180Ah bank isn't supposed to like more than 30A of charge.  So, obviously that meant I needed more batteries to soak up the circa 65A I've observed coming out of the solar chargers . The tired wet battery bank I tied in parallel doesn't accept charge very well and doesn't assist much unless the gel bank is depleted quite a bit (it has a lower resistance and higher hold-up voltage than the wet batteries).

I tried reducing the wiring resistance a bit by making up a 25mm² cable set for it but that didn't change the fact that the 110Ah wet batteries are now only about 35Ah (I measured it on a load test a few weeks back) and that means more than half the plates are inert material that just increases the cell resistance.

I was out in Wales yesterday for a meeting and this meant passing near by Wokingham, where the eBay dealer in used UPS batteries has his "lair".  At £50 each for two year old 105Ah AGM batteries with "one careful owner", it would have been rude not to.  He did also have some Yuasa 150Ah batteries but they were more expensive and too heavy for me to lift on my own.  The Marathon ones are about 35kg and I'm no Schwarzenegger.

Anyhooo...  Here they are



Just got them on a refresh charge now as there's no telling how long they've been sitting in that shed of his  .

I'll have to make up some different ends for the charge wires as the existing ones have those snap-on SAE post terminals and these new batts have 6mm bolts.  Oh, and the bolts he gave me were too long for the blind holes in the terminals .  So it's either off to the hardware store tomorrow or out with the Dremel tonight .

The other beauty with these batteries compared to the wet ones or even the big 6V gel cubes is that they can lay on their side and then they are flat enough to be secreted under the sofa.  See...?  There is a plan.