Simple MPPT solar panel charge controllers

Discussion in 'OnBoard Electronics & Controls' started by BillyDoc, May 21, 2009.

  1. gonzo
    Joined: Aug 2002
    Posts: 15,601
    Likes: 1,096, Points: 123, Legacy Rep: 2031
    Location: Milwaukee, WI

    gonzo Senior Member

    There are some systems that produce a high frequency to "recondition" batteries. I don't know if they work
     
  2. BillyDoc
    Joined: May 2005
    Posts: 420
    Likes: 18, Points: 0, Legacy Rep: 266
    Location: Pensacola, Florida

    BillyDoc Senior Member

    Gonzo, I believe you are thinking of typical resistive regulators, also known as "linear" regulators. The regulator I posted the data sheet above for (LM5118) can take an input from 3 to 75 volts and produce an output with an even wider range. That is, it can produce voltage way above or below the input voltage on the output. In fact, in the example they give in the data sheet for a 12 volt output the input can be anything from 3 to 75 volts (it does need at least 5 volts to start up). Not only that, but it does this magic way more efficiently than a typical resistive regulator.

    It's an extremely complex subject, but if you want to learn more Google on "Switch mode regulator." These things are amazing!

    BillyDoc
     
  3. mydauphin
    Joined: Apr 2007
    Posts: 2,164
    Likes: 53, Points: 48, Legacy Rep: 575
    Location: Florida

    mydauphin Senior Member

    Anyone know of any cheaper Solar controllers that work on my BP panels, they produce over 24 volts dc and would like run serial so that is 50 volts. Outback controllers work great but are expensive.
     
  4. BillyDoc
    Joined: May 2005
    Posts: 420
    Likes: 18, Points: 0, Legacy Rep: 266
    Location: Pensacola, Florida

    BillyDoc Senior Member

    MPPT front end circuit

    Hi mydauphin,

    Actually, designing something exactly like what you need is what I'm trying to do here. So, keep an eye on progress and please pitch in if pushing electrons around is within your expertise!

    To those who love to inconvenience those poor dumb electrons:

    I think I have a front-end design worked out now, and have posted the circuit diagram below. The circuit will handle input voltages up to a maximum of 65 volts, and uses roughly 1.25 mA of current when the input voltage is below threshold plus another mA when above threshold to operate the enable output. So, in my case where I have a solar panel with an optimum power point at 14.036 volts, that means that the circuit "costs" a maximum of 0.03 watts to do the MPPT function.

    Here's how it works.

    The output from the solar panel passes through a "safety" diode to prevent the solar panel from shorting out the works when in the dark. This diode must be rated for the expected current (about two amps in my case) and voltage. I'll be using a 15TQ060 Schottky type myself, which is rated at 15 amps and 60 volts, because at 2 amps the forward voltage drop is only about 0.45 volts. The output from this diode feeds a capacitor, the value of which will be determined later. Lets call the voltage on this capacitor Vin.

    Vin goes first to an AD8214 comparator, which is a high voltage comparator rated at 65 volts. When the voltage at pin 2 of this comparator (the positive input) is higher than the voltage at pin 8 (the negative input) the current flowing out of pin 5 jumps from a low of less than 100 nano amps to a high of 1 milliamp. This current output is very handy, because we really won't know at what voltage this pin will rest under various conditions, and we needn't care either. If we find that we need a 5 volt logic level on the other end of the enable output, all we have to do is put a 5 Kohm resistor in series to ground . . . and we've got it. That five volts will be developed across that resistor regardless of the voltage at pin 5 of the AD8214.

    A couple of problems now arise: First, the inputs to the AD8214 will only tolerate a differential voltage less than half a volt, but if both inputs do stay within half a volt of each other they can be moved (together!) over the entire voltage span of the device, plus a little. That is, the "common mode" voltage on these inputs can range up and down the input voltage of the device, which can be from 5 to 65 volts. The second problem is that in order to make the MPPT actually track the maximum power point we have to be able to set a fixed voltage for the comparator to trigger on. This is best done with dedicated voltage references, but the widest range device I could find (the TL1431) only has a range of 36 volts. So, we need to make provision to "split the difference" and offset the triggering voltage to somewhere near the mid point of what we actually want. Doing this gives us a functional range of 72 volts for the voltage references.

    So, concentrating now on these voltage references, the two TL1431 devices, it should be appreciated first that the "mid-point" voltage needn't be all that precise, because the AD8214 has a nice wide common mode voltage swing. But the overall trip point voltage should be as accurate as possible! The TL1431 devices specified have an accuracy of 0.4%, but this is with precision resistors and calculated values. We are going to do better than that, because we will use sloppy old 5% resistors . . . and an adjustable potentiometer. Then the accuracy will be very strongly related to the multimeter we use to adjust the system with. Don't let this scare you, most decent multimeters are accurate enough.

    For now just take my word for the fact that the two resistors, marked R1 and R2 to the left of each TL1431 can be picked or adjusted such that the TL1431 acts just like a Zener diode, except a lot more accurately, at the voltage you want, and with far less temperature drift. So, ignoring the two R1s and R2s for now, Vin initially rises over the tops of a TL1431 connected to Current Regulator Diode - 1 (CRD-1) and over the top of CRD-2 which is, in turn, connected to a second TL1431. Both of the CRDs are designed to deliver 1 mA current (this is what a 1N5297 CRD is). BUT, the TL1431s both look like very high resistances until they reach their zener voltages, and these high resistances prevent much current at all from flowing to the CRDs. The CRDs in turn are trying to get some current going, so they look like low resistances.

    But wait! Initially Vin is too low to "turn on" the TL1431s but there's another possible current path. Assume that both TL1431's are simply not there because of their high resistances. Now, trace the current through CRD-2, then through the leftmost 11DQ06, then through CRD-1 to ground. That's a current path that works, with both CRDs in series with a Schottky diode. Even at low voltages this current path will draw the 1 mA the CRDs are designed for, with roughly a 0.35 volt drop over the Schottky diode. This voltage drop over the Schottky will be positive at pin 8 of the AD8214, and negative at pin 2 of the AD8214, thus leaving the "enable" output in a high-impedance state (not asserted).

    With the enable not asserted the voltage regulator yet to be designed will not be putting any load on Vin, so if the sun is shining at all Vin will continue to rise.

    My panel has 29 "stripes" of photovoltaic material, so I know from the Maxim app note (posted above) that the maximum power point for this solar panel is 29 times 0.484 volts, or 14.036 volts. I've got a diode in series with it that drops about 0.45 volts, so I subtract that and get 13.586 as the optimal voltage that I want Vin to be. I want to divide this 13.586 into two more-or-less equal voltage drops, so, for the lower TL1431 being fed by CRD-2 let's just see how close we can get using common standard value resistors.

    Half of 13.586 is 6.793, so the first thing I need to look at is how much resistance I can tolerate with R1 and R1 in series, because I don't want to draw any more than about 0.1 mA here. Using ohm's little law I divide 6.793 by 0.0001 and get: 67,930 ohms. I only need at most 3 microamps at the ref input on the TL1431 so this gives me lots of room to play.

    Next I go to the formula on the circuit diagram below, shift it around a little to solve for the R1 to R2 ratio to give me roughly 7 volts and see that it's 1.8 to 1. OK, so now I know that I need to get over 70 K total resistance, and R2 will be a little less than twice what R1 is. Now I go to my resistor trays and look for a resistor that is about 30 Kohms . . . and there it is! A nice 33 Kohm resistor. So far, so good.

    Now there is only one unknown in my formula (the one on the diagram below) and that's for R1. So, I plug in 6.793 for my Vzener, and 33,000 for R2 . . . and out pops 56,667.6. Fifty six Kohms is close enough, and it's a standard value as well.

    Back to that formula I see that an R1 of 56 Kohms and a R2 of 33 Kohms will give me a voltage drop of 6.742 volts. Close enough! Now I can go to the upper R2 and just guess that 33 Kohms will work nicely, with an adjustable R1 of 100 Kohms nominal. In the discussion that follows let's just call both of these Vzener voltages 7 volts.

    Where were we? Oh, yeah . . . we want the AD8214 to go on when Vin gets to about 14 volts.

    Remember that Vin is going up all the time we've been playing around with math, so let's just assume that things got a little out of hand and Vin is now up to 14.2 volts. Now the TL1431 associated with CRD-1 has enough voltage to drop 7, and so does the TL1431 associated with CRD-2. Vin will now pass through the top TL1431 with enough current to cause a voltage drop across CRD-1 of 7.2 volts, and at the same time CRD-2 will also develop the same voltage drop for the same reason. Now, however, because CRD-1 is tied to ground the voltage at the junction of CRD-1 and it's associated TL1431 is 7.2 volts, relative to ground, and CRD-2s voltage at the same relative point is 6.8 volts. This will cause a current flow through the rightmost 11DQ06, and also make pin 2 of the AD8214 positive relative to pin 8, thus causing the AD8214 to assert the enable output.

    This is what we want. As Vin rises to a pre-determined level the enable is asserted and subsequent machinery will suck current from Vin to charge a battery at a rate high enough to cause Vin to drop below the threshold of the AD8214. If the sun is bright, Vin will soon be back up and the process repeated. If it is dim, it will take longer, but in both cases power will be extracted via Vin at very near the maximum power point for the solar cell.

    I'll get to the battery charger next, but meanwhile please look over the above and find those inevitable screw-ups for me!

    BillyDoc
     

    Attached Files:

  5. mydauphin
    Joined: Apr 2007
    Posts: 2,164
    Likes: 53, Points: 48, Legacy Rep: 575
    Location: Florida

    mydauphin Senior Member

  6. BillyDoc
    Joined: May 2005
    Posts: 420
    Likes: 18, Points: 0, Legacy Rep: 266
    Location: Pensacola, Florida

    BillyDoc Senior Member

    Hi mydauphin,

    The short answer is "no, not at all!." Both of your links are to DC to DC converters, but that is only part of the problem. Solar cells are subject to highly variable inputs (full sun, clouds, rain, shade, etc.) and at every "situation" that the solar panel finds itself in there is an optimum current draw that maximizes the usable power (volts times amps). An MPPT charger implements some means of Maximum Power Point Tracking to adjust the circuitry to find and use this maximum power point dynamically. It's a very good thing to do, because doing so can add roughly 30% more useable power to your batteries (other conditions being the same). Google on MPPT or Maximum Power Point Tracking for more information on this.

    BillyDoc
     
  7. BillyDoc
    Joined: May 2005
    Posts: 420
    Likes: 18, Points: 0, Legacy Rep: 266
    Location: Pensacola, Florida

    BillyDoc Senior Member

    Opps! I forgot something . . .

    I forgot to make a provision for the case where the battery is fully charged and you want to just let the voltage from the solar panel rise to it's maximum with no current being drawn. In this case Vin would cause an excessive current through the two TL1431s and one of the 11DQ06s. The fix is easy, just add two resistors as shown below, as R3s.

    To calculate the value of R3, subtract the maximum power point voltage from the solar panel (in my case 13.586 volts) from the solar panel maximum no-load voltage (29 times 0.55 volts, minus the Schottky drop of 0.45 volts in my case, that is, 15.5 volts) and divide this value by 0.002. Then pick the next higher standard value, which in my case is 1K. So, again just in my case, both R3s will be 1K, which will limit the maximum current draw through the 11DQ06 diode to 1 mA when the system is in "idle," but will otherwise have no effect.

    BillyDoc
     

    Attached Files:

  8. pistnbroke
    Joined: Jan 2009
    Posts: 1,405
    Likes: 34, Points: 48, Legacy Rep: 404
    Location: Noosa.Australia where god kissed the earth.

    pistnbroke I try

    I glanced over all this wonderfull techno stuff but nowhere do I see any mention of the fact that to charge the battery to 80% you need to get it up to 14.4v ....at this voltage it will not gas. 100% but gassing 15.3v .... personally dont see the problem in fitting a shunt regulator to bypass panel output when bat volts reaches 14.4v god its only a 20w panel ..I am with gonzo fit a diode and let it get on with it .....
     
  9. BillyDoc
    Joined: May 2005
    Posts: 420
    Likes: 18, Points: 0, Legacy Rep: 266
    Location: Pensacola, Florida

    BillyDoc Senior Member

    Yes, pistnbroke, you are quite right . . . and if this was the only charger I need I would do exactly what you say, just hook up the panel and forget about it.

    But this isn't the only charger I need. And it serves nicely to learn a bit about MPPT chargers. And grabbing an extra 30% more power isn't such a bad thing either when you are only starting with 20 watts under optimal conditions . . . provided the cost is minimal, and it is.

    As for the voltage levels you mention, that part of the circuit isn't implemented yet, so you have to wait a few more days for that. I have a couple of other things I have to do first, but will be back at it by the middle of the week at the latest. As a preview, I will be implementing the LM5118 buck/boost controller with a mechanism to step through the various charging stages you mention as appropriate.

    I am also trying to keep everything 60-volt capable so the charger can be adapted to work with 12 volt, 24 volt, 36 and 48 (nominal) volt systems. A "universal" design, if you will, and this does add some complications. So, please be patient a little longer!

    BillyDoc
     
  10. gonzo
    Joined: Aug 2002
    Posts: 15,601
    Likes: 1,096, Points: 123, Legacy Rep: 2031
    Location: Milwaukee, WI

    gonzo Senior Member

    I think you fail to grasp the basic concept of any regulator: You will lose power. Even if the circuit increases voltage, it will decrease amperage in direct proportion minus a loss. With the panel you have it will still be barely a trickle charge. Also, internal resistance of the circuit will sharply limit what you can do to increase voltage.
     
  11. BillyDoc
    Joined: May 2005
    Posts: 420
    Likes: 18, Points: 0, Legacy Rep: 266
    Location: Pensacola, Florida

    BillyDoc Senior Member

    Gonzo, Here's a quote from Outback Power:

    http://www.outbackpower.com/products/charge_controllers/ (emphasis added).

    I'm trying to do the same thing here, but with a DIY device for those with electronics experience to save a few bucks (like me!). The Outback Power MX 60 controller (which is extremely nice, by the way) sells for about $500, and is worth every penny; but only if you need that much controller and can't DIY, or just don't want to bother.

    BillyDoc
     
  12. gonzo
    Joined: Aug 2002
    Posts: 15,601
    Likes: 1,096, Points: 123, Legacy Rep: 2031
    Location: Milwaukee, WI

    gonzo Senior Member

    Controllers are like a faucet. If the water pressure is not enough, not amount of opening it will give you more water.
     
  13. BillyDoc
    Joined: May 2005
    Posts: 420
    Likes: 18, Points: 0, Legacy Rep: 266
    Location: Pensacola, Florida

    BillyDoc Senior Member

    Well, I guess those guys at Outback are just lying then.

    BillyDoc
     
  14. gonzo
    Joined: Aug 2002
    Posts: 15,601
    Likes: 1,096, Points: 123, Legacy Rep: 2031
    Location: Milwaukee, WI

    gonzo Senior Member

    Read their specs. The claim is that their controllers are 30% more efficient than other controllers. It is very vague because there is nothing to really compare them to. You still need sufficient power to control. A weewee panel does not produce it.
     

  15. BillyDoc
    Joined: May 2005
    Posts: 420
    Likes: 18, Points: 0, Legacy Rep: 266
    Location: Pensacola, Florida

    BillyDoc Senior Member

    Actually, Gonzo, I think if you look a little closer you will find that they are claiming to: " increase your renewable energy yield by up to 30%." They do this by tracking the point of maximum power output (watts), as it is produced by the solar panel under varying conditions AND by matching the voltage requirements of the batteries being charged to that available power. If solar panels were linear devices this would not be an issue, but solar panels are definitely not linear. What Outback and others are doing is well beyond mere "efficiency." If you were to use a 100% efficient controller that did not do the tracking part, you would not even get close to what Outback does.

    BillyDoc
     
Loading...
Forum posts represent the experience, opinion, and view of individual users. Boat Design Net does not necessarily endorse nor share the view of each individual post.
When making potentially dangerous or financial decisions, always employ and consult appropriate professionals. Your circumstances or experience may be different.