Simple MPPT solar panel charge controllers

[BillyDoc]

I spent some time today trying to find circuits for implementing simple Maximum Power Point Tracking charge controllers. The commercial ones I have seen are all overpriced, in my opinion, or way too capable for my needs. So, I thought I would try to design one. At this time I only want to keep a single marine battery charged with what I think is about a 20 watt solar panel. It's for a cabin I have deep in the woods.

So far I have found two interesting circuits from the same man, W. Stephen Woodward (pdfs attached). The first (MPPT battery charger) describes pretty much what I am looking for, but uses a buck-only Switch Mode Power Controller (SMPC), thus limiting charging to those conditions where the voltage output from the solar panel exceeds the battery voltage by some amount. It also has a feature for dropping the charge current back to "trickle" when the battery voltage reaches fully charged, which I think is a necessity.

The second circuit (Solar array controller) is more sophisticated and uses a boost-buck SMPC, however the one chosen is a surface-mount device with 141 tiny pins!!! It's also damn expensive ($45). I really, really don't want to solder that thing up. The circuit as a whole also seems to lack a "fully charged" mode.

With a lot of study I think I can extract the best features of the two designs, but as I've mentioned before on this forum I'm lazy on principle and would love to get some help here. The MPPT approach seems to me to be the only way to go for this problem. So, has anyone else found good circuit ideas that implement MPPT for small systems? Or, have any good ideas of your own?

BillyDoc

 

[marshmat]

Hey BillyDoc,

I see what you mean about the regulator chosen for the second circuit... who the hell uses LGA141 for a power device, anyway? (OK, Linear Technology does, but still....) Do note, though, that only 14 of those pins have to be precisely soldered. The rest are soldered up in six banks, where you're allowed to be sloppy. Still, it means buying a bunch of $45 chips and LGA sockets to match. Annoying.

The MPPTs I've used (for solar cars) have all been very sophisticated units, made by AERL in Australia and with prices around $1000, give or take a few hundred. A typical car array would be a kilowatt or two and would have at least two, usually six or eight separate MPPTs. These things worked anywhere from twenty to a couple hundred volts, great units, very compact. I don't have any experience with cheaper and/or simpler ones.

The easy way to go would be some variation of the first circuit you posted, and make sure your solar panel has a higher operating voltage than your battery. Full buck-boost capability is complex, thus expensive.

I'll give it some more thought, though.... there has got to be a cheaper way....

 

[BillyDoc]

Hi Matt,

From what I've read this morning AERL originated the concept several years ago. They and Outback seem to be the leaders in the field still.

Most of the "big guys" seem to be using microcontrollers in their design, which certainly makes sense if you are going to sell a bunch of them and charge for "features." But I think that Woodward's analog approach is just as good if you can do without digital displays and the like, and I can.

Probably the sensible way to approach this is to look at buck-boost controllers first and work from there. I think I'll give Maxim a visit.

BillyDoc

 

[Guest625101138]

Have you worked out what benefit you will get from peak power tracking on your system. If the panel is a good match to the system voltage does it make much difference. You will need a regulator to protect the battery but is anything more worth the effort and expense.

I can certainly understand the benefit of peak power tracking with wind turbines. Also if the solar panel voltage does not match the battery system then you need some form of voltage switching.

Rick W

 

[BillyDoc]

Hi Rick,

The solar panel I have is 12" x 36" and seems to put out roughly 16 volts in full sun with no load. Beyond that I don't have any data. It didn't even have any leads (it was a gift), I just attached some with conductive epoxy this morning. I think that I could put a series diode on it and attach it directly to my battery and not go too far wrong . . . but I hate the thought of boiling off the battery juice, and if it isn't too expensive or troublesome . . . why not go all out? So, I've been exploring the MPPT concept today.

And I'm beginning to think this might not be so difficult after all. I found an application note on the Maxim site (App. note 484) that had some very interesting info in it. The thing that really grabbed my attention was their contention that since in almost all cases the maximum power point was at a voltage of 0.484 volts per cell you could connect the solar panel to a small capacitor, monitor the voltage on the capacitor with a comparator until it exceeded 0.484 volts times the number of cells, then switch on a plain-vanilla switching controller to the battery until such time as that voltage dropped below the threshold. Then wait until it built back above threshold, and so on. The rate at which the voltage builds up will be entirely dependent on insolation, so just this procedure effectively tracks the MPP while providing enough stored current to supply a switcher further down the line for a few cycles.

Many switching controllers have "enable" inputs, so this could be a means to a very simple charge controller. Obviously, I need to study it more though. The Maxim app note is attached below, if you're interested. I also found a switcher (boost/buck) that might work and is cheap ($4.18), National's LM5118, data sheet also attached.

So at this point I'm thinking, solar panel connected to capacitor with comparator looking at the voltage and enabling the LM5118, the output voltage of which is set by some flip-flops to go through two or three stages to trickle.

Or something like that.

BillyDoc

 

[Guest625101138]

I am interested in what you are doing because I am looking at panels outputing 60V (peak power at 50V in full light) feeding a 44V system. I also do not have much idea how the voltage-current changes with light level.

In your case I just wonder if you will get much from it by the time you allow for the losses in the controller.

The attached curve is typical of what I have seen for cells. You can see that shifting the operating point around the optimum does not change the power output a great deal.

Rick W

 

[BillyDoc]

You can see that shifting the operating point around the optimum does not change the power output a great deal.

Rick W

I think that you are absolutely right on this point, but only if the shift magnitude is pretty minimal. To put some numbers to the graph you provide, the peak voltage on the left with no load will be about 0.55 volts (I'm talking about a single cell) and according to the Maxim app note I cited above the peak power point will be at 0.484 volts. My panel appears to have 29 cell "stripes," so the peak voltage is 15.95 volts and the power optimum is 14.036 volts. Let's assume that this panel will put out a current of 2 amps under optimal conditions.

Now, if you look at the graph you see that increasing the load so that more current is drawn quickly moves the operating point of the cells onto a steep downward slope and thus reduces the power available. The question is, by how much.

In my case, my battery will sometimes get discharged to 10.5 volts, so if I attach the solar panel directly to it at this time the battery load will shift the power point to the right until the voltage at that point is 1/29th of 10.5, or 0.362 volts per cell. This will clearly have a bad effect on the power output because of the steep slope of the curve at that voltage.

We don't have numbers on this graph, but from what I've read so far the claim is that the power will be reduced roughly 30% by this effect (neglecting controller losses). So, if my solar panel will put out 2 amps at 14.036 volts optimally, that works out to 28.072 watts. If I reduce that by 30%, then I'm left with only 19.650 watts with which to start recharging my battery. I think its worth it to try and avoid this loss, at least as much as I can.

Figure one in the Maxim app note is the same graph you provided, but with additional light levels. They show the slope of the drop off with increased current to be almost vertical! But more importantly, they show that the mid point of the "knee" on each data line is more or less at 0.484 volts from 0.1 sun to full sun. So, if we have a capacitor attached to a solar cell that is already charged up to say, 0.48 volts, then whatever the light conditions above that solar cell it will be trying to increase the charge on that capacitor with maximum power. If there is only 0.1 sun it is going to take longer than if there is a full sun available, but in each case the maximum power available will be utilized. Capacitors, moreover, give back almost all of the energy stored within, they are close enough to 100% efficient not to worry about the difference.

If we now use a comparator to detect when the voltage on the capacitor gets to, say, 0.488 volts and turn "on," and also build in a little hysteresis so it turns "off" at, say, 0.480 volts, then we can use the output from the comparator to switch on a switch-mode regulator connected to the battery we want to charge.

As an aside, I have a large ancient, battery operated, fork-lift truck. The battery on this thing is huge, and when I got it you could see yellow crust all over the plates. I had read that hitting such a battery with high current pulses with enough time between pulses for the released gases to be re-absorbed could "de-sulfate" the plates and put the battery back in business. So, I rigged a device to do this. I used 70 amp pulses of 1 millisecond duration every half second ... for a couple of weeks. It worked beautifully! The yellow crust went away, and strangely enough the volume of the liquid in the cells increased considerably. I had to suck out the excess several times, and I still haven't gotten around to adjusting the acid density. But the battery now works very well, and it sure didn't before this treatment. The reason I mention this is that I think that bursts of current are a very good way to charge a battery!

So, when a cloud passes overhead and the battery is only getting a shot of current every so often, it is just cleaning up the mess in there, nothing to worry about at all!

Rick, the LM5118 is good to 75 volts input, so I will try to make the entire device capable of this so you can use it as well. You don't happen to know of a comparator that can handle this voltage do you?

BillyDoc

 

[gonzo]

I don't think a 20W panel needs any regulation. In ideal conditions it will put out 1.7 A. Unless you are charging a cell phone that is barely a trickle charge. I use panels without regulation and never had problems

 

[BillyDoc]

Hi Gonzo,

I'm sure you are right! But I also think I can get a significant improvement. I know it's probably silly for this application, but I hope to learn enough by working the problem at this level to be able to do the same thing when faced with a much larger problem, like in a boat. I think of it as "education."

It's good to know that I could just directly wire my panel up to the battery though. Thanks!

BillyDoc

 

[Guest625101138]

I have had a look at the actual curve for one of the panels I am considering. The open circuit voltage is less than I thought. I will not charge my batteries directly from them because the voltage is too low.

However if you take the curve and assume it is working from 32V to 42V, similar to your 10.5 to 14V then you get an idea of what is lost by not optimising continuously.

At 32volts they are 5.8A so 186W.

At 42V they put out 5.1A so 214W.

The optimum is 225W. So the losses are not that great considering these values are the range limits.

The worst case will be in the high temperature situation when the batteries are almost charged. Under this condition current is down to 3A. But then it would be going into trickle anyhow.


I do not know what sort of comparator circuit you mean!

Exercising batteries over their range usually restores some capacity but I have never seen heavily sulphated cells.

Rick W

 

[Guest625101138]

Hi Gonzo,

I'm sure you are right! But I also think I can get a significant improvement. I know it's probably silly for this application, but I hope to learn enough by working the problem at this level to be able to do the same thing when faced with a much larger problem, like in a boat. I think of it as "education."

It's good to know that I could just directly wire my panel up to the battery though. Thanks!

BillyDoc

I am certainly interested from the learning aspect as well. I want to buy a big panel to play with. Things like the temperature rise are important factors.

I think the peak power tracking on a turbine will be even more interesting.

Rick

 

[BillyDoc]

Hi Rick,

Don't give up on those panels just yet, the LM5118 controller will charge your batteries with an input from 3 to 75 volts. And funny you should mention turbines, it's basically the same problem as with solar cells, if you try to suck off too much power you just bring them to a halt from back emf. I did a design for an alternator conversion a couple of years ago that this same circuit could be used for. (http://PoiesisResearch.com/Power.php)

A comparator is an electronic circuit (almost always in an integrated circuit chip) that compares the magnitude of two voltages, and produces a digital output in response. So if one voltage is higher than the other the output is "true" if it is lower the output is "false." You can also design in a "window" between the "true" and "false" responses, called the "hysteresis." Hysteresis is like on a thermostat that turns the furnace on when the temperature drops to 70, but only turns the furnace off when the temperature rises to 73. That thermostat would have a hysteresis of 3 degrees. In this case the comparator would be comparing the voltage on a capacitor being charged up by the solar panel with a reference voltage which would be fixed at 0.484 times the number of cells in series. When the comparator went "true" the controller would turn on and suck the capacitor down as it charged the battery, the comparator would then go "false" and the controller would stop. I'm looking at a comparator from Analog Devices that works from 5 volts to 65 volts that I think will work nicely here (AD8214) but I've yet to read the spec sheet.

I was more than a little shocked when I looked at the plates on my big battery. Which is why I started looking for a way to fix the problem. Those big babies are expensive! It works great now though, and the fix cost me nothing but some time.

BillyDoc

 

[gonzo]

The improvement given by charge regulators can only be achieved with a power source that surpases the charge in current and voltage that the battery can achieve. Unless you install a huge array of panels, all you are accomplishing is to drop the voltage without any advantage. To learn about regulators of any kind, you are better off using a simple battery charger as a power source.

 

[BillyDoc]

The improvement given by charge regulators can only be achieved with a power source that surpases the charge in current and voltage that the battery can achieve. Unless you install a huge array of panels, all you are accomplishing is to drop the voltage without any advantage. To learn about regulators of any kind, you are better off using a simple battery charger as a power source.

Hi Gonzo,

Can you elaborate this? I have no idea what you mean! Are you saying that an MPPT charger has to be larger than the battery capability to achieve any benefit? Why?

BillyDoc

 

[gonzo]

Regulators reduce voltage to control the rate of charge (amperage) . They can adjust to type of battery, state of charge, temperature, etc. To achieve that they start with a higher voltage than necessary and drop it.