60 cells (30V) PV modules for 24V off grid?

[lifer]

I've just bought 8 x 230W PV modules for off grid use. I was thinking of putting them in series (two chains of four modules) and using a MPPT controller to charge the batteries. I also have two pure sine 24V inverters (1500W and 700W).

I wonder if I could connect all those PV panels in parallel and feed the battery bank and the inverter directly.

The battery bank will be charged through some kind of PWM controller with a voltage limit cut off function. The battery output would be connected to the inverter through a diode (or a controlled switch - power MOSFET) so when the panels aren't producing enough power (panels output voltage is lower than the battery voltage) I could extract the power from the batteries.

Is this a common sense approach or am I missing something? I want to protect the batteries as much as I could (they're pretty older) and I haven't the MPPT controller yet (I'm going to build it along with a 3kW pure sine inverter so it's going to take a while but I need a temporary solution).

PS: I know that 60 cells (30V) PV panels are used mainly for grid-tied systems but it was a good trade so I had no choice.

[dnix71]

I doubled up cheap panels to get better voltage in low light and shade and it works for me (45v +/-)

60v is plenty to charge a 24v system, but you need to be careful about the charge controller you get. The cheap ones won't handle 60v. And strictly speaking 60v is considered high enough to require following the NEC.

30v would be fine for a 12v system, but 24v nominal is 28.8 fully charged. You won't properly charge a 24v system from a single 30v panel if you did that to avoid the 60v arrangement.

 

[Flux]

Your first proposal is the better one. The mppt controller will get most from your panels.

With plenty of solar coming in your batteries will only be working when there is no sun. Most of the time they will be more or less floating and cycling them within the surface charge will not shorten their life.

The second option has many problems, inverters are really designed to run from batteries at near constant voltage and with plenty of reserve current. I wouldn't chance the inverter with panels alone even with a controller to prevent over voltage. Even if it works you will be hammering the inverter input capacitors with the current limited source.

Without the mppt you will get much less production from the panels and as far as I can see you will loose out all round. I suspect the diode will not do a lot except conduct the peak inverter current from the batteries (safer than running th panels to inverter direct) but leaving out the diode and using the mppt seems infinitely better.

Your panels are absolutely fine for a battery system but will need mppt to get the most from them.

Flux

[lifer]

@dnx71: I'm going to build the MPPT controller by myself so it will have a higher input voltage limit (300V) as I might connect all the PV panels in series (240V).

@Flux: thank you very much for your support. Like I said, it supposed to be a temporary solution (maybe a week or two) till I get the MPPT done.

Speaking of inverters.. could you help me understand why can't (directly) connect an inverter to PV panels?

As I mentioned before, I'm going to design & build the inverter by myself. Inside the inverter, there's a step-up (boost) input stage which converts the slight variable PV output voltage to a higher and relative constant voltage (350V) using a feedback mechanism.

So, at this point, I assume the rest of the inverter will see a constant voltage source as input (for the DC/AC "H" bridge PWM driven stage). When the panels output power become insufficient, a controlled switch will dinamically connect the battery to inverter's input so those 350V still remains constant.

I have a small battery bank (150A/24V) because I don't need more overnight power. I also prefer to keep the batteries in full condition all day long if I have enough power coming from PV panels (almost 1.8kW). As long as the batteries are fully charged I want to "save" them for the night.

The MPPT charger output current could reach 75A for 24V battery voltage which are a killer for those small batteries so I can't use the whole panels power for battery charging at any time. That's why I thought I have to make the inverter to work with direct panels output.

(Please correct me if I'm wrong. Even I'm graduated of electronics engineering, I'm a software guy lately!)

[Rob Beckers]

Lifer, please be careful regarding the panel Voltage you're expecting from 8 of them (60-cell) in series: The open Voltage (Voc) depends quite a bit on temperature, and for example in my own neck of the woods (southern Canada) we see that go up to 44V on a cold day. That makes 352 Volt for 8 panels in series. Your controller better be able to handle that (adjusted for your lowest temperature). Keep in mind that running semiconductors at the edge of their rated Voltage makes for poor durability.

Regarding PWM controllers and 60-cell panels, just one thing to add: The Voltage that those panels will charge the batteries with is the battery Voltage itself, it pulls down the panel Voltage. That means you will see about 15V on a 12V battery system, and about 30V on a 24V system, just about half the regular MPPT Voltage that delivers maximum power from those panels. So, your 230W panels would be delivering roughly 115 Watt each when paired with a PWM controller (Power = Voltage x Current).

I know you want to build an MPPT controller, but I wanted to point out the mechanics of a PWM controller with 60-cell panels in case others are going to try that. As Dnx71 points out the Voltage is plenty high enough with two panels in series, for a 24V battery system (and it is of course), you just loose 50% of potential power output..

As to driving an inverter directly from PV modules: One very fundamental thing you need to keep in mind is that PV modules are a CURRENT source; you can change the Voltage from 0 through just about 30V and the output current from the panel will barely change (for a constant light input and temperature). Batteries on the other hand are a VOLTAGE source; you can change the current a great deal without doing much to the battery Voltage. Driving anything, including an inverter, should keep this in mind. I'm pretty sure it's possible to drive an inverter directly from PV modules, but they make a poor power source. Output constantly changes, depending on light and temperature. Even during the day there are huge swings as clouds drift by. If you don't have a need for constant output power from the inverter I don't see why you wouldn't be able to drive it directly. In effect you would have to build an MPPT charge controller for your rail Volage, that keeps the panels producing optimally, you'll need some serious buffering of the rail Voltage (lots of electrolytic capacitors), and that would drive the H-bridge.

Even so, my guesstimate would that you'd be loosing power frequently, even for a small load, as PV output varies constantly..

-RoB-

[lifer]

Wow, thats a great bunch of informations, Rob!

I don't know, but I always prefer higher voltage instead of higher currents. That leads to lower diode conducting and line transmission loses, thinner coils wire & cables and so on.

I will follow your advices, anyway. By "simple PWM charger" I meant a step-down (buck) converter, which stores the energy extracted from the panel in a coil then deliver it to the battery - so the panel won't be directly connected to battery at any time. But you're right, a classic PWM charger just connect/disconnect the panels from batteries on a cyclic basis.

After all, a MPPT charger is just a buck converter with some feedback, trying to extract the optimal power (voltage/current ratio) from the panels.

I have no experience (yet) with SMPS (just some PC power supplies debugging) but I've read a lot about this subject lately and I'm trying to take the right decision. Your advices were very helpful, thanks again!

Maybe the best solution is to have a larger battery pack (24V/500A or something) to really keep the inverter input almost constant but I'm hardly struggle to keep these "green energy" cheaper! That's one reason I've decided to make the MPPT charger and the pure sine inverter by myself. I'm also in the process of building a wind turbine and I can't go cheaper on that neither (it's not like I don't have enough money but I'm fighting for "diy" and "free energy" concepts!)

Sorry for being off-topic. Well, I'm going to build that MPPT charger in the next few days so I probably use those PV panels when I'll finish the charger. Thanks again for keeping me on the safe side!

[Flux]

" The MPPT charger output current could reach 75A for 24V battery voltage which are a killer for those small batteries so I can't use the whole panels power for battery charging at any time. That's why I thought I have to make the inverter to work with direct panels output."

I see where you are coming from. You can use your inverter without issue fed from the battery, I am still reluctant to suggest you try direct as Rob said the current source is not what conventional inverters are designed to work with. Your diode will work but I see little advantage in using it.

When in float charge the mppt controller will adjust itself to the inverter load and the battery current will be just that needed for float.

If your worry is excess charging current during the absorb phase then try to find a mppt controller with adjustable current limit or use a diversion controller at absorption volts. If you are not using full inverter capacity you could sense battery volts or current to divert to water heaters rather than a separate diversion at battery volts.

You seem to have some good ideas for the final solution with a pre-regulator for an inverter to widen the input voltage range but for now you need a short term fix.

Flux

[Flux]

I think we are on the same wavelength. I just underestimated you time scale in producing this device.

I wish you well you are young and well experienced. My own experience as an old codger trying to keep up with modern ideas was that there is a world of difference between a baby SMPS to run a computer compared to a large power converter for a few kW. The intricate layouts possible on a small pcb are not so easy to produce with larger power devices. Layout is everything above 10kHz, wires don't exist at 30k

Flux

[mab]

hi lifer,

I use 60cell/30v panels direct to battery (24v) and they work fine; mine have a Vmp of about 28v which is pretty optimal for my AGM battery. For my small system a MPPT wouldn't justify it's own cost - the money's better spend on more panels (and I don't the have the time/skill to make one myself).

If you have a 'wet' type battery and want to push the volts up to ~30, or your panels Vmp is much less than 28v then it will be less optimal.

if your Vmp is close to your desired charging voltage, your cable run is fairly short and you can cobble up an 'ideal' blocking diode (FET that turns off at night), then you too may find that the MPPT isn't worth the effort.

Even though they're low voltage compared to what is generally recommended for battery charging(72cells) I find they work very well even in fairly low light and outperform the 72cell panels (also direct connected). In very low light when the output of the panels is down to 10's of mA, the 72cell panels have the advantage; but in those conditions the amount of solar power generated is insignificant and I suspect a MPPT would probably use as much as it gained anyway).

[lifer]

@Flux:

Thank you very much for your kindly appreciations but I'm not quite young anymore (biologically speaking)! Regarding the MPPT charger, I have no commercially constrains/limitations (dimensions, cost control) so I'll go with oversized components (power mosfet, ferrite core) for a safer design.

@mab:

Thank you for your feedback! Actually, my panels have a Vmp of 29.6V (STC) but my main concerns are the higher current rate (60-70A) that my panels could generate when direct connected to the battery. By the way, I have 4 VRLA batteries (2 + 2 x 12V/75A). I know, they're quite a "babies" for off-grid but I'm going to take care of them thus waiting for the MPPT charger.

[Mary B]

Wires do work okay at higher frequencies but you need to take precautions in routing, filtering etc. Or you end up with something that will wipe out radio reception for blocks.

[lifer]

Thanks, I'll count on your advice!

Charging (and discharging) the power mosfet gate capacitance at higher frequencies it's my main concern (as I don't own an oscilloscope yet) but I'll go for a feasible design (using a very good mosfet driver) so I hope there won't be (m)any troubles.

[OperaHouse]

Some random thoughts.......

Get an Arduino...The UNO is like a temporary girlfriend that is still around twenty years later.  These are cheap, powerful and
easy to use.  I have one that controls the battery charging, operates the refrigerator, heats the hot water, operates the pump,
monitors system condition and records data.  It works just like I was there monitoring meters, flipping switches and turning knobs.
It is great for a temporary solution that somehow becomes final.  It will make a great charge controller but forget most UNO
projects you see on the internet.  Most of these are from first time designers who want to show everyone how smart they are. Maybe
it is me, but everytime I see a well laid out article a little voice in me says watch out.

ON/OFF...High speed switching isn't exactly a black art, but there are things you never would have thought of.  Every transition
is a loss.  150,000 of them a second amounts to more loss than if you have 500 of them a second. The circuitry is a lot simpler
and construction is not as critical. If you are not tied for space go with bigger transformers. I make my controls out of old UPS.
I have one simple FET driver that is only an opto isolator and a resistor. Who needs a high side driver. I power it with a special
purpose 14V medical battery snatched from the towns recycling center.  It has been powering the system for three years, I don't even
disconnect it at the end of season.  The water heater diversion control is also low speed PWM using just an opto isolator. This keeps
the panels at the power point and results in 40-80% more power than a simple diversion control.  The UNO also allows me to set limits
on the PWM duty cycle. With the heater control there is never a shot cycle PWM.  It stays between 7% and 93%.  When the control asks beyond that it turns fully on or off.

Scheduling....Bigger and bigger is always a solution.  Not a smart one though. I hate batteries and only have two car batteries in my
system. One is out of a vehicle I don't take with me, the other is for the boat when it isn't being used.  That gives me about 15AH
of capacity. The battery is there for short surges of current like thr fridge which draws 123A for a few seconds. There is an over
abundance of solar panels because the system has to run even on cloudy days.  Scheduling allows running everything, just not at the
same timr.  The wife is pressing me for a dishwasher.  She wasn't satisfied when I told her I married one.  It should be possible to
let it schedule a time of day and switch off all other power draws and do dishwashing.  It may have to switch off for a cloudy period
or when some other need takes priority, but it will be done at the end of the day.

Mathing it to death...There are those compulsive people who want to take a data point every nanosecond. I don't do power point tracking.  A panels power point voltage is pretty much fixed.  It only varies with temperature. Through empirical data I figure out the power point voltage and leave it at that.  Use the computational capacity to do something more useful. Getting system data every half second is more than enough

[lifer]

Well.. nice thoughts!

You're doing a great job with that little arduino. I have an old PC doing almost the same jobs as yours and a lot more (lights on/off scheduling, turtles feeding, various parallel/serial/usb devices and so on), the main difference being the internet connection and the web server, thus allowing me to control all those devices over the internet. I must admit the arduino it's a more compact and low powered device but I just needed a more powerful and versatile machine.

Speaking of dishwasher (the one you're not sleeping with!), I have even more power-hungry households appliances (A/C, microwave & electric ovens, dishwasher, refrigerator, water pump, few large TV sets and so on) so I really need more than one small battery and a PV panel to feed them. The main challenge of being off-grid it's to preserve your life style.

As for the MPPT charger, I've just find out a very simple and clever implementation of the power point tracking algorithm (more details here):



I think I'll give it a try as soon as I'll finish the power stage design.

[lifer]

Ok, guys.. I need special attention.

It just crossed my mind.. what if I don't use a charger at all? After all, every single PV panel works like a current source (8A in my particular situation). So I already have 8 "native" current sources to play with.

Thus what if I'm going to use some simple switches (ON/OFF power MOSFET) to dinamically connect every single panel to the battery bank depending on the battery charging state?

Moreover, I think about having two separate battery banks and alternate charge/extract power from each one.

So, while 4-6 panels feed a discharged battery bank, the other two panels are feeding the charged battery bank (connected to the inverter) so actually those panels are feeding the inverter (helping the charged batteries).

During the first bank charging (or depending on inverter load peaks) I can "move" back and forth some other panels between the two battery banks .

From the PV panels datasheet I-V diagrams (cells temperature and solar irradiance) seems like I'll stay around MPPT zone at any time (if the batteries are not deep discharged). The MPPT at STD is 29.6V (thus around the trickle charging point of lead acid batteries).

If my assumptions are correct, I only need some cheap power MOSFETs to connect each and every panel to a battery bank or another and some more powerful MOSFETs to move the inverter input from a (discharged) batery bank to another. And, of course, a cheap logic to manipulate all those inputs/outputs based on batteries charging status.

What do you think about it? By the way, the PV panels are closer to the battery banks (2-3 meters) so there won't be a problem to use 8 separate cables (one for each panel) or I could put the switching module (solid state relay) near the panels then only use two (thicker) wires - one for each battery bank.

[OperaHouse]

Thyat is OK to do, but I wouldn't tell anyone about it.  I had a system like that for a while.  Two 10A Morningstar controllers and the extra 20A of panels connected to a FET switch.  If the battery was below 14.4V it would turn on.   Your method is very inefficient at lower battery levels and lower light levels.  And that is when you need the panels most.    A PC is great but you have to feed it.  I have stacks of old UPS use the transformer and FET assembly.  I have a grinder with a thin cutoff wheel that makes nice assemblies of the circuit board. The UNO makes a nice controller that doesn't draw much.  Run four PWM controllers for panel groups instead of one big one.  Running a fixed power point is 95% as good as full tracking. 

[DamonHD]

You can take the basic UNO core and run it on microwatts.  For example the OpenTRV project that I have mentioned in the microcontroller section is based on the same 328 AVR core (and run-time system) as the UNO, wakes up every couple of seconds to check if the outside world is still there and sample some inputs, and draws only a few microamps at 2V--5V supply, ie can run off a couple of AA cells for a couple of years, nominally.  And all the tools to program it are free.  And it's easy to talk to over USB.

It can directly drive power MOSFETs as well as do PWM and digital and analogue inputs...

Don't knock some very smart very low power hardware control with a decent free widely-used programming environment built in!

Rgds

Damon

[OperaHouse]

It is a basic fact, you never have enough battery. The same is true for solar panels.  Yet you will end up wasting most of the power
you can generate.  Everyone here does that.  The future is learning how to schedule that power.  Switching battery banks is not a solution.  I've done it to protect good batteries from deep discharge or to get one battery back up to a full charge quickly.  These are all acts of desperation and a place you don't want to be.  I've thought or done all the things you talk about.  You've just got your panels and you want to try out all your ideas.  So go and have some fun.  I really suggest you get one of those little energy monitors like the Turnigy 130A.They can be had for under $20.  I have four of them.  They are an eye opener and will totally change your approach to off grid.  Numbers don't lie. Like Newton said, until you can express something in numbers, you know very little about it.

Once I get my taxes off the workbench I need to start working on my UNO based power point controller. Will be making it for 36V solar
bank since that will give me about a 50V power point for the water heater.

[lifer]

OperaHouse, you've just changed my life!

After reading your post (thanks @Damon for +1) I looked around for an arduino supplier and I was just fascinated! So many add-on (shield) cards, so many accessories.. I must admit I was in a market for something similar few years ago and I've read about arduino (or even raspberry) at the date  but I sticked with the plain microcontrollers.

Anyway, I've just made a (long) shopping list and I kindly ask you for an oppinion about it. So I decided to buy one Arduino Mega 2560 R3, one Serial Graphic LCD 128x64 and one Arduino Ethernet Shield. I want to use them for battery/panels monitoring, a MPPT charger and a pure sine inverter (that's why I choose Mega instead of UNO - much more (analog) inputs).

Do you think it's a good start up? Maybe we'd talk about it on a different area (microcontrollers). I really apreciate your help!

[DamonHD]

Note that the stock Arduino stuff is easy to program but not necessarily very power efficient out of the box eg by virtue of using linear regulators and not putting the CPU fully to sleep, etc (if you care about that sort of thing).

Though it is very busy, and difficult to follow at times, the Arduino forum is stuffed full of weird and wonderful stuff including most of what you might like to do.  As is AVRFreaks and a couple of other sites.

I'd go and point your favourite search engine at some of those and see what comes up with some of your key words for the projects that you have in mind.

One of the reasons that I'm sticking with AVR/Arduino for the moment is that it's pretty good and has a HUGE user base doing many many weird things.

So the stuff you've specified is probably as good as anywhere to start, but you may outgrow it quickly.

I can't tell you how many AVRs and custom PCBs I have in my house right now, but if it's less than 50 then I've failed!

Rgds

Damon

[OperaHouse]

I have some strong feeling on this.  I buy the 2011 UNO version because they range from $7 to $10 and uses all the same software.  At
that price you add a few components and have a powerful control.  Add a bunch of shields and it gets expensive and clunky.  Some make it hard to get at the pins you need and you also used up a lot of pins with these shields.  There is one I sorta liked, a L293 motor
driver board that was about $3 shipped.  It had 8 H bridges with wire connect terminals suitable for driving a FET or relay. It also
provided pin connections to the mother board.  The listing showed the chips in sockets but the chips came soldered.  I really needed
to remove the shift register to make the board do what I wanted. I picked up a lot of opto22 type driver boards for about $4 each. These are similar to the Majic boards that you write 8 bits to at 9600 baud, same as the UNO port.  The plug in modules are cheap at about 75 cents each. Makes a nice clean setup.  These Majic boards are quite a bit less expensive than the standard mother board, probably because people can't figure out how to use them.

But I'm off on a tangent.  If your hobby is computers then no expense is too much.  I don't see any need for all of that stuff.  The
UNO development system has a nice monitor screen to allow you to tune up the system.  After it is working it should just be a black box.A couple cautions.  Plugging in the USB now resets the UNO so any daily run data is lost.  Never got around to looking for a work around.I just fed important data to a spare PWM pin and read it on a DVM, quite linear.  If you are using A/D inputs, plugging in the USB will likely change your numbers because the supplied USB 5V is higher than on board regulator.  I modified a USB cable by cutting the 5V line.

The UNO is a "single chip solution".  You can program the chip in a UNO board and then just slap it in a simple board with not much more than a crystal of any frequency and have it perform some dedicated function.  If you are going to hang a bunch of I/O onto it, some other platform may be a better option.

[lifer]

I need that graphic LCD to locally display useful data (volts/amps, instantaneous power sinked/generated, some other electrical parameters and so on) and for local settings. I'll go with cheaper solution at any time but it has to be feasible for the project. Btw, what "other platform" could be a better option?

Damon, thanks for your confirmation. That's why I choose Mega instead of UNO - there's always needs for more inputs/outputs during project development.

PS: I already placed the order. I just can't wait to start the implementation!

[DamonHD]

FWIW I'm hoping to get my hands on an e-paper 'shield' for my UNO to play with.  Just chasing up my order today.

Rgds

Damon

[lifer]

Are you talking about something like this?



It's such a beautiful piece of art, indeed. The only downside - it's using a lot of ports (but then again, that's why I choose Mega).

[DamonHD]

Yes, though smaller.

And I intend to use it purely graphically, ie no or little text!

Rgds

Damon

[Mary B]

Damon have you played with the BeagleBoards at all? I was thinking of going that route for whole house DC control. Can add a standard video display for a monitor panel.

[OperaHouse]

I can see that you have gone over to the drk side.

I worked once for a custom board house that had very high insertion costs.  I would agonize for days on whether added
performance it was enough to add a 1/3 cent resistor. It was like Name That Tune, I can build that project with 13 components.
I loved that aspect of the job, being, a minimalist at heart.  My fridge program that handles about everything
uses the board mounted pin 13 LED of the UNO for the display.  The LED makes a short blinks every 8 seconds just to show it
is running.  Next it does a short blink for every degree above the set temperature, up to four. Then up to four long blinks
indicating the charge state of the battery, no blinks indicates near full charge. Normally all you see is just a single blink
indicating it is running.  From 20 feet away I can easily see the state of the system. I love data. The first day of it changes
the whole concept of the design.  After that it is just background noise.  Data is still always there if you need it.

I would think that a group that sees an old brake rotor as a wind generator would also see an old UPS as a dump or charge
controller with the addition of the UNO. One of my favorite chips is the three pin TL431.  Adding two resistors and a relay gets
you a simple voltage monitor. Perfect for all those who say, I just want it to do this.  Too bad problems are not that simple.
Add or remove a load the voltage changes, you need time delays for startup surges, timeouts before something is restarted and
a host of other math function.  Most of these programs can be written with only 15 lines of code.  Anyone can write 15 lines of
basic code.  Start adding screen drivers and it looks like so much spaghetti to everyone. So when you see my projects in the
future don't laugh. I'm just trying to instill an organic capability in others so they don't ask me to write code for them.
.

[DamonHD]

@MA: no; I've had my time cut out with SheevaPlug then PICAXE then AVR.  Friends have though.  I think I'll concentrate on AVR/Arduino for now as I have invested a fair amount in tooling and custom hardware, and the minimum current draw is difficult to beat.  But the Cortex M0+ stuff beckons, I think, by the end of the year, at least for R&D tinkering.

@lifer: it appears that my e-paper arrived yesterday but I was too knackered to notice.  Something to play with at the weekend!

@OH: we're going to have to get into that kind of production engineering very soon aiming to reduce factory gate cost from maybe £50 per unit to £2 per unit so that retail would be £10.  I'm pretty sure that we can get there, and quotes for full manufacture of the existing design in China have already come in at about £20 per unit without much effort.

Rgds

Damon

[lifer]

I'm just trying to instill an organic capability in others so they don't ask me to write code for them.

Hey, sorry if my cry for help sounds like a threat to you! (j/k)

Like I told you, I'm a software guy so as soon as I grab those arduino "toys" (and that will be today) I'll start writing codes right away!

Even though I admire your journey to simplicity, I still need that gLCD to simultaneously display all those status informations. Using one single LED it's like using morse code to chat on facebook these days. Anyway, for a particular case, one single LED might be enough (ie some non life-supporting projects like a fridge monitoring!).

Anyway, I've just ordered some 100/200A current shunts to start monitoring the battery and PV panles currents.

[jack11]

"I've just bought 8 x 230W PV modules for off grid use. I was thinking of putting them in series (two chains of four modules) and using a MPPT controller to charge the batteries. I also have two pure sine 24V inverters (1500W and 700W).
I wonder if I could connect all those PV panels in parallel and feed the battery bank and the inverter directly."


Going back to the original theme, I did some experiments lately that showed me I can't count on the operation of the direct-connect panels to be in line with conventional wisdom, probably because of the varying operating conditions, as follows:

1. Can't always assume that the panels' voltage will be pulled down to the battery's voltage, because it may be the other way around.
2. Can't always assume that the panels will operate in the current-source region (below Vmp), because they may operate in the voltage-source region (above Vmp).

For example, I've had my small portable system out on a couple of occasions, and saw it operate in two much different ways. The system is two 60-cell Sharp 230W panels (Vmp=30.0V and Voc=37.1V at STC) wired in parallel, connected directly to a 24V battery bank (two 12V lead-acid deep cycle 105Ah batteries wired in series).

On one occasion, I had it out in about 60-70 deg F weather, with medium sunlight, and fairly high humidity. The system voltage was about 28V at best, and the charging was pretty lame (looked like the batteries were always undercharged after each daily charging cycle). I believe the panels operated in the current-source region then.

On another occasion, I had the same exact system out in about 40 or so deg F weather, good sunlight, and pretty low humidity. The system voltage came close to 34V while charging, and the batteries started boiling. They continued to boil even when I took one panel out (the voltage stayed the same and the current was cut in half, as expected). I've lost quite a bit of water from the batteries.
I took some measurements then that showed me the system must have operated in the voltage-source region, even though it does not seem possible, but that's the only way I can explain what I measured.

I haven't yet tried operating this system with any sort of diversion or PWM charge controller to keep the battery voltage in check. This would have probably prevented the runaway voltage, but still the point is that unusual things can happen without any charge control, as the system makeup and the operating conditions change.

[Mary B]

As solar cell temperatures drop the output voltage rises. Example: It is 43 degrees today, I am seeing 87 volts off my array to the charge controller. At -20 it has topped out at 92 volts and at 90 degrees it drops to around 78.

[lifer]

Damon have you played with the BeagleBoards at all? I was thinking of going that route for whole house DC control. Can add a standard video display for a monitor panel.

I've been playing with Arduino (MEGA) for two days and I'm pretty disappointed about its capabilities. I just need something that could be remotely and dynamically programmed (not being completely erased each time I'd make a small modification) so I guess I'll go with BeagleBoards or OLinuXino (or maybe pcDuino? - there are many available options).

Regarding the PV panels, I've just mounted them on the roof and I have them directly connected (yes, I know.. ) to my two brand new 225A batteries (12V + 12V). I've just made a simple voltage monitor (3 stages: 11.5V > discharged, 13V > charged, 14.4V > trickle charging) that runs some 12V relays which connect/disconnect every single panel from the batteries.

Anyway, that's just a temporary solution till I've got my MPPT charger done.

[DamonHD]

There is a gulf between a microcontroller system such as Arduino and a small full general purpose computer such as an RPi; they are for different jobs.

I wouldn't want to monitor or control stuff on an RPi taking a few watts when my entire Internet-facing server only takes that in order to allow it to run off-grid year round.

However I do at this moment have several things providing simple monitoring and control services using milliwatts down to microwatts on microcontrollers, and with appropriate levels of autonomy, simplicity and reliability.

Since replacing the entire non-volatile firmware on an Arduino takes a matter of seconds isn't that *wonderful* rather than disappointing?

I think the fact that I can do what I did 20--30 years ago on a general-purpose CPU now on a microcontroller at a millionth of the power consumption, plus nice I/O and non-volatile store, or 1000 times faster with getting on for one million times the storage, in each case with broadly similar prices, is amazing.  I don't expect to achieve both in the same box.

Rgds

Damon

PS. I've had a preliminary play with the e-paper and it seems good, though it'll take some clever design to production engineer it for simplicity to work within the 328's memory constraints.  Today I have to do boring paperwork instead.