Request comments on simple solar charger

Hello Lewin,

Having trouble zooming png files such as this down so I won't comment on the overall schematic.

Just one word of caution: Many LDOs are like the princess on the pea. When the output capacitive load doesn't have just the right ESR range they can become unstable. The same can happen with some of them if the source impedance is too high. With solar panels being the source that can be a serious concern.

Personally I prefer switchers with solar. You could then even do MPPT but I guess that would be a bit overblown here.

Regards, Joerg

Hi Joerg,

Sorry. If you want to look at the EAGLE schematic, that is also up at

- but not that important I guess...

That's weird/bad news. I haven't had good results with switchers in battery applications as the internal resistance of the battery goes up. For example, a 1 ohm resistor in series with the battery has pushed the regulator into a Bad Place for me when pulling a lot of current.

Hi,

All the solar gadgets I've inspected were a resistor, a diode and a NiCd or NiMH battery pack. I guess these things are designed to cost $5, they work for a while, and when the batteries die you throw them away.

The 24V rail only supplies motors and solenoids, so I can tolerate variation and noise in it. There is a small MC34063-based switcher powering the smarts (they pull

Hi.

I also had a look for solar charging circuits, and the one I found on the web were very crude. They looked like they'd work but waste energy on diode drops and/or have little or no regulation.

I've bought solar powered gadgets to look inside them, and their circuits are equally cheap and nasty.

The A of E (see power supply section) points out that the solar panel I/V curve tends to flatten out at a maximum voltage and is thus a fair match for lead acid batteries. The latter seem the most robust battery in terms of low maintenance and management. The British Antarctic Survey use them rather than Ni-Cd etc.

I think you should be able to get away with fairly simple charging circuitry, and one diode drop in 24V is minimal. I suspect the battery to supply rail needs the regulation more. After all, the battery is smoothing the power input like the main electrolytic cap in a wall-wart (only much bigger!) before driving the regulator.

I expect you could replace the diode with a switched MOSFET to remove the diode drop.

Perhaps it could be gradually switched off as the solar voltage got too high, instead of having the crude shunt regulators.

Such simple circuitry should be okay in sunny climates, but in the UK I expect solar panels to spend much of their time way below their nominal voltage. They can still provide energy, but not at a high enough voltage. Therefore I would want to design a step-up SMPSU on the charging side. This would then make use of the low-grade sunlight we get here.

In your circuit I don't think you need LDO regs to drive the power rails. Lead acid batteries are not precisely 12V, in fact they are usually 13V8 when really fresh IIRC. If anything, you might want the LDO between the batteries and the power rail.

I've been thinking of using a solar panel to charge my bike light. Since this is only used at night when no charging can happen, I wondered about an SMPSU that could be switched from regulated charging to regulated discharging into the light bulb. Now there's an interesting challenge...

Hi,

No. A robot of sorts. If you think of an electromechanical scarecrow targeting nocturnal animals, that can drive around the perimeter of a field and photograph the beasts it's scaring off, you have an idea of the sorts of functionality in my device though that is not actually the application.

I've already speced in slack there. The device needs to run a ~1A motor load for three to four minutes every morning. It then sits around for the entire day soaking up rays and doing not much else. There is a scheduled comms session which could pull as much as 750mA for some time, but mostly it will be The problem is that solar cells seldom run at full voltage unless they are

I can't guarantee no clouds in my usage scenario :) But I can guarantee that the device is smart enough not to attempt anything strenuous while the batteries are depleted.

I won't quite say money is no object, but solar cells are not a major part of the project budget. I'm looking at products like the 04-1192 cell from this page:

(sorry there are no direct product links). It's a 6x6" cell, 0.54 Voc, 6A Isc, $12.80 apiece. An extra $50 or so for four more cells is not going to dampen my enthusiasm. On the other hand if the simple design with diode only and fewer cells will get me 365 days of operation, I won't waste money on greater complexity.

Reference to what, exactly? The comparator is powered off a 5V switcher. The various R dividers there are designed to trip the comparators when the difference between the two sides of one of the batteries (i.e. the nominal output voltage) falls below a predetermined lobat threshhold. Now I come to think about it, the low side comparators should have dividers at their -ve inputs also.

Well, do recall what I said in an earlier posting, that there are actually two sets of everything you see here, and the controller is capable of switching in a good bank to take over from a bad bank.

I'll look at this part, thanks.

Yep, the average consumer can't see how primitive it is, they just sunbathe it to perk up the cells.

Great, so no need for precision regulation.

Sounds like a sun-tracker?

Roughly speaking, they are big flat diodes. Illumination causes current to flow 'backwards', and the IV curve is just the normal diode curve shifted diagonally.

I think the top voltage is so that it can't force more charge in when the battery is 'full'.

The lower voltage just ensures there is enough voltage to force a full charge in.

I suspect that going slightly below that minimum voltage would only meant the battery never got to 100% charged, which isn't dangerous. It would mean you might have to change the spec to say it only ran for less than the maximum capacity of the battery.

Manufacturers write data sheets to show their products at their best. It may well be true that a panel reaches X volts and Y amps maximum, but only under Z illumination (sunny day, no cloud).

On sub-ideal lighting, it quickly drops down the IV curve.

A simple resistor load would draw power from the cell for all lighting conditions.

A diode/battery would not, of course.

I'd cautiously say yes, because solar cells tend to reach a maximum voltage (c. 0V45) then flatten out (equivalent to the 0V6 Vf of a diode). So they won't go 'over voltage' by much. Hence less need for 'dropper' regulators.

The problem is that solar cells seldom run at full voltage unless they are somewhere seldom cloudy. Such as on spacecraft, tropical beaches, in deserts, or on top of Mauna Kea (and other good optical telescope sites).

The crude solution is to have lots more cells so you have enough voltage to discard even when they are running at less than maximum voltage. In which case you will need dropper regulators. The snag there is that solar cells are not cheap.

The ideal solar cell load would follow the maximum power point on the power curves for all illuminations. If those points joined up to a straight line, a fixed resistor would do, but they don't, so it won't.

Essentially you need to design a DC to DC converter optimised for a wide input voltage range.

>

A switcher seems like the way to go. Since you are only using 300C a day, that is 10mA for 8 hours. Thus, you need about 24*10mA / E to run the thing. Assume E = 50%, then that is about 500mW. Assume 1/3 of the days have sun, and you need 1.5W (ie, one day will charge enough for 3 usages, on average.) If you use the cheapo 500mA panels you posted, that means you need 3V to supply this. Since a switcher isn't going to run very well at 3V, buy a few more (they are cheap) to get it up to 5.4V. Assume 10. Then, you have 5.4V at 500mA, or 2.5W, which is more than enough to keep your batteries charged, assuming a bit of self-discharge.

SLA batteries like CCCV charging, which simply means that you don't want to dump a bunch of current into them initially. I've heard 1/10 C as a reasonable figure, so assume 700mA is the max charge they'll take. However, if you charge them at 50mA, you can basically just leave them on forever. You may want to cut the charge when they get to 27.2V (which is the float value). Your call. I think that the cells will last longer if you do, but 50mA is a fairly tiny trickle, so it may not even make a difference.

So, use 10 panels at 0.54V, 500mA each, a switcher chip to get you 24V at 50mA from 5V, and stack the two 12V SLA batteries.

An LT1373 from linear could work nicely, or you could use a microcontroller to build you own with current limiting.

--
Regards,
  Bob Monsen

If a little knowledge is dangerous, where is the man who has
so much as to be out of danger?
                                  Thomas Henry Huxley, 1877

Hi,

? The cells I was looking at are the 6000mA ones further down the page.

These particular batteries are specified for 2100mA charge current. When I'm testing them on my bench, I generally set my PSU for 1A current-limit and set the open-circuit voltage to 13.7V, then let the battery work it all out. When fully charged, the battery continues to pull about 30-40mA depending on temperature.

Thanks for the advice. I'm going to do some experiments now and see how it works out. (Luckily I have a large box of sacrificial batteries that I picked up cheap!).

snip

Yep, it is a bit confusing. the best expalnation I came across was that it's a compromise between float and boost charging.

Normally a float system is 24/7, and maintaining a the correct cell voltage is a "good idea".

Since solar power is more like 8/7 to 13/7 it cannot be considered a true float, so the solar guys tend to overcharge to compensate. It does mean a reduced battery life, just one of the prices that you have to pay

martin

Hello Lewin,

Yes, that worked just fine. I don't see any reference in there. Hmmm... you need one.

If it was me I wouldn't split it up like that. What if Solar1 drifts into the shade while Solar2 doesn't? Or maybe got dirty, or a large bird decided to take a long nap on its top bracket? Or that large bird gets the feeling he shouldn't have eaten that last road kill and then, well, you know what I mean.

If you really don't want to use a switcher you could connect the panels and the batteries in series and get rid of almost half the circuitry.

A well designed switcher doesn't mind a high source resistance. You could, for example, use a switcher based on the LM3478. Then use a single solar panel (or two if you need the power) and convert up to the voltage you need. The nice thing is that it will charge even when the panel voltage drops below nominal.

To measure the status of the battery you need a reference. The LM3478 provide a nice bandgap but only when it's in regulation. Else, a TLV431 comes to mind.

BTW, if you absolutely have to use LDOs there are some that provide a signal when they go out of regulation. IOW when the input voltage drops so much that there isn't enough head room.

Regards, Joerg

Hello Lewin,

For comparing two sides, ok. But what for is the connection at + of BAT1?

Currently you seem to just charge away and there is no reference voltage against which a battery status of "full" would be measured. Regulator chips usually only come in discrete varieties such as 9V, 12V, 15V, neither of which is suitable for lead acid 13.8V limits. You'd need at least an adjustable regulator and then there has to be a resistor from ADJ to GND.

Amother thing you need is diodes to block discharge when there isn't enough light.

Then I would keep them completely autonomous but you have a connection in there. Also, allowing two panels to work on both batteries would provide more redundance than splitting it up completely.

Regards, Joerg

What I was trying to say is that if you buy those 6A 0.54V panels, you are wasting your money. The cheaper ones will work just as well, assuming you are right about your project using 1A only 5 minutes a day. You certainly don't need to buy 28V worth of them! That will cost you more than $600. Using a switcher, the entire power circuit will probably cost you less than $50 for the electronics, panels, and battery, and it'll be far smaller, only taking up 30 sq in. Using your 6A, 0.54V panels, you'll need 36 * 51 = 12.75 sq feet of panels! Also, you can't even charge your battery at more than 2.1A, so why do you need 6A?

A switcher like this is a relatively simple circuit. You can download a copy of LTSpice from

which has a built-in designer for switchers (using their parts, of course.) Or, you can just buy a DC-DC converter that takes 5V and outputs 28V, and use a SLA battery charging chip. Again, you don't need much current to keep your battery charged, assuming at least a bit of sunlight.

Anyway, good luck.

--
Regards,
  Bob Monsen

If a little knowledge is dangerous, where is the man who has
so much as to be out of danger?
                                  Thomas Henry Huxley, 1877

So the device does more than just run that 1A motor and do the comms session.

This is part of why I was asking the question in here - the cells are speced under ideal solar conditions, so I don't know how far to derate them for real life. Considering the cost of the device, adding another $300 to $600 in solar cells is money well spent if it significantly improves the chance that the device will last for its intended lifetime.

Thanks for your comments - I'm back to the drawing board.

that is the OPEN CIRCUIT volatge of the solar cells. Under load the volatge is less.

Think about connecting the solar cells directly (through a blocking diode) to the battery.

You need to take the load line of the cells and the load line of the batteries and see where they intersect to see how muvh current the cells will drive through the battery at what voltage.

Mark .

If you do go the DC-DC step-up and SLA reg, I highly recommend the UC3906 in the latter role. If you have trouble locating the Unitrode/TI data sheet just post back here - it is found under some arcane name.