Battery charger design for solar panels

Discussion in 'General Electronics Chat' started by TexasTony, Sep 29, 2012.

  1. TexasTony

    Thread Starter Active Member

    Jul 15, 2010
    38
    0
    Hi folks,

    OK, got a new project, looking for some guidance. I'm looking at getting a handful of solar panels (90-130 watt, 18 volt). In looking into battery chargers, they're nearly as expensive as the panels! I've read about how the MPPT are more efficient than PWM chargers, but I don't think they're worth the price.

    In fact I'm planning to just design some switching regulator boards, and that's the project. I'm thinking 1 to 2 solar panels per set of batteries (12v), and run some inverters I have already off each bank. So that's the background.

    I've worked with guys who do switching regulator designs before. I understand the basics (controller, FETs, inductor(s), capacitors, routing requirements, rising/falling edge critical-to-power-loss details...).

    But I've not designed one, and I've been looking but not found one that meets my needs. Here's the needs list:
    * 16-22v input
    * 12-14v output
    * 5-20 amps input and 8-30 amps output (solar panels will be 90-130 watt 18v nominal, and I'd like to do 2-3 panels per regulator if possible)
    * I may add some feedback resistors to a micro-controller to tweak the output voltage (uC will read temperature, battery voltage and adjust the output voltage accordingly. If I can get fancy, I'll do some of the MPPT stuff and maybe load-match the output current/voltage to maximize the panel output power. This will mean using the regulator in a current limiting mode. And I'll want a regulator that can be disabled).
    * Maximize efficiency, I'd like to see 92-95%.

    The regulators I used to work with were for computer motherboard CPUs, so they could drive 100+ amps, but only at about 1 volt. I've seen boards actually catch of fire due to bad layout (that was a fun one!).

    So my weak point is finding a regulator that meets these needs. Any suggestions?

    Thanks
    Tony
     
  2. Kermit2

    AAC Fanatic!

    Feb 5, 2010
    3,795
    951
    How does the 16-22 volt input work if you have 12 volt batteries?

    The inverter will expect to see 11-16 volts not 16-22.

    The solar panels will work perfectly to charge 12 volt batteries. When charging a battery they will output a little less current and have a voltage slightly above the batteries they are charging.

    A reverse current diode should be used so the batteries cannot discharge back through the solar panel.

    an overcharge cutout would be a nice feature to have. If you could build a circuit that would switch the batteries out of the solar panel output path when fully charged, then your installation would be hands free and you could leave it un-monitored for long periods.

    Again, the solar panel output will be lower than the 18 volts stated when useable amounts of current are being consumed. No special charger circuit is needed.
     
  3. Kermit2

    AAC Fanatic!

    Feb 5, 2010
    3,795
    951
    I apologize for my grumpier alter ego above. He's just getting old and cranky. :)

    If your real goal is to design the converter circuit you describe and not simply to charge the batteries with solar panels as my previous incarnation believed, could you forgive me and describe in further details any aspect of this project you need help with specifically?
     
  4. Dyslexicbloke

    Active Member

    Sep 4, 2010
    420
    19
    I am in the process of doing that, building a charger I mean, right now and I cant give you a definitive answer RE a design but I can tell you what you should be considering.
    I have AGM cells and a mixture of different panels.

    1.
    The voltages you are talking about are probably either the 'pmax' of the panels. That is the voltage at which the panel will produce the most power given an appropriate load.
    Mppt tracks this pmax point, which will change with illumination, and delivers the maximum charge current to the batteries at the current terminal voltage.

    Since the panels will function just fine delivering current to your batteries directly, IE the voltage will fall as the current rises, you need to decide if MPPT is worth while.
    Given that the best MPPT design, basically just a buck converter with external voltage set-point, will only be circa 95% efficient you need to be gaining 5% just to break even.

    I plan to try it just for the sake of it but I must admit I am not expecting to gain much, if anything.

    2.
    Your batteries specific charge requirements. Most will benefit from, if not require, temperature compensated charging and bulk charge current limiting.

    3. Terminal ripple voltage, which most batteries do not like at all.

    My basic plan ids to use PWM to control the charging current.
    I will have several switches, FET's, on different panel circuits which essentially puts multiple switches in parallel reducing the on resistance and therefore loses in the form of dissipated heat.
    Each 'channel' will be out of phase with its neighbour, thus reducing ripple and therefore size of the required chokes and caps.

    A small uP will orchestrate it all generating multiple PWM signals whilst monitoring voltage and currents.

    I would be happy to collaborate if you are interested, I'm no expert but have been researching for a while now.
     
  5. TexasTony

    Thread Starter Active Member

    Jul 15, 2010
    38
    0
    Kermit first...

    If you're old... what am I? I recently got (an underserved as yet though!) an AARP card...

    But to get specific to your questions. You don't want to give a battery too high a voltage, it'll shorten the battery's life, and that varies with temperature and other factors. Back-draining is always an issue, so something like a diode is necessary (but not a diode, it'll be too much power drop, use a FET which is part of the switching regulator design).

    But honestly, if you're asking these questions then I think I'm already ahead of you.


    Dyslexicbloke,
    I'd be happy to work with you. I like you're numbering system, I'll build on that. Let's have this discussion here so that others can benefit from our discussion. If we need to have private conversation, we can always go to private messages later.

    1. Agreed. MPPT is a feature and can be added/deleted as desired. But it will need some 'hooks' to enable it (ie ability to measure power). Let's not focus on this for now.

    2. Temperature compensation I think is a necessity. But this will be easy to implement. Here's my strategy: I'll have a uC with a couple of external temperature sensors. One close to the FETs (over temp check). One close to the batteries (maybe multiple), to see what temp they're at and if they start overheating. So the idea is measure the battery temp, and use some outputs with resistors to the feedback voltage. So if I want 12.7, 12.9, 13.2, ... I can use a combination of these outputs to adjust the regulator voltage.

    2b. I split out the bulk current limiting as that is a completely separate issue. Some regulators have this feature, some don't... I'm hoping that controlling the voltage will help limit current, but I know it will be likely more than that.

    3. Let me first ask you to clarify what you mean by ripple voltage. Obviously switching regulators have ripple. That's what the capacitance and inductor are designed for. But the reality is that the batteries will look like a massive capacitor. And caps are relatively cheap. So I do not see any issue with ripple noise.

    Let me add to this ripple question. The power drop in a switching regulator is mostly in the rise/fall edges going into the FET. That's why the switching frequency is a key factor: faster means less ripple but also substantially more inefficiency. I plan for the lowest freq and use caps/batteries to compensate for ripple. (for any perfectionists, yes the inductor also drops a meaningful amount of power, but there isn't much I can do about that). I'm saying all this to say that a real switching regulator controller is needed to drive these FET controls, you can't just use outputs from a uC to drive the FETs (directly or indirectly). In reality you can, but I don't think it will be a good idea.

    (let me start adding bullets)
    4. You said multiple FETs to reduce losses. I disagree. You're only looking for trouble if you try this. First, I'm not sure if switching regulators can run in parrallel. They could fight each other. It's better to get a switching voltage regulator that drives multiple sets of FETs if you need higher current. This is part of why I want to use multiple banks of batteries (this is multiple purpose, if one battery shorts/fails I don't want it to drain the entire bank, also I have a few 2kW inverters, and they cannot work in parallel).

    5. You had said AGM. Love the batteries, hate the price. 8 years back I bought them (used but 90% good from data centers) for $25/battery (80 amp-hour, so they were full sized). I saw new optima AGM batteries the other day for $220 each. OUCH. I not going to get into that right now, but when I'm ready to start that project I'll start another thread to talk about a much cheaper alternative.

    6. Switching regulator choices. OK, here is an couple of options:
    http://www.irf.com/product-info/datasheets/data/ir3859m.pdf

    It's only 9 amp output. The problem is that most regulators can't handle the higher voltage, this one goes to 21v max. But it's nice in that it has the FETs integrated, so really I only add a few R's, the big inductor and capacitance. But it's a fairly straight-forward design. I'm open to comments on this selection, and any other options anyone else knows about. The biggest shortfall is the 9 amp limit, it'll be only 1 panel per regulator. The best feature is intergrated FETs, modest price ($2.8 / 25 on Mouser). I can tweak it for voltage adjustment, disable...

    http://datasheets.maximintegrated.com/en/ds/MAX8664.pdf

    Here is a switching regulator with external FETs. $4.5 / 25, so cheap enough. This one is nice in that it can control two charging circuits, each up to 25 amps! That is two solar panels on each side, so 4 panels of charging. But circuit design is more complicated.

    Comments?
    Thanks
    Tony
     
  6. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,394
    1,606
    Well... an MPPT converter is slightly more then just a buck converter, though it uses one. An MPPT converter tracks the maximum power point of the cells and runs at that point, using the highest voltage and current (thus power) it can simultaneously draw. A PWM converter just switches the battery in and out to set a current point.

    If we accept the 95% of the MPPT then it will be running at the 18V for .95 8 180W = 171 effective charging power.

    A PWM will drop the voltage down to around 14V while charging so the power input is 14/18*180 = 140 watts, and then you need to give it's efficiency in powering the load.

    So MPPT here would gain some 22% over PWM, and probably more.
     
  7. TexasTony

    Thread Starter Active Member

    Jul 15, 2010
    38
    0
    Ernie,

    I think you're confusing linear regulators and switching regulators, which is not the MPPT vs PWM difference. I'm going to reply just to clarify for all the other readers.

    A linear regulator will take in the 18 volts (let's say 5 amps just for example). It drops the 18v down to 14 volts, keeping the same 5 amps. So it takes in 18x5 (90 watts) and charges with 14x5 (70 watts). So it is only 78% efficient in this example. The larger the voltage difference, the less the efficiency.

    A switching regulator works very differently. I'm not going to redefine how it works here (google "switching voltage regulator" for education). But basically it uses an inductor, lots of capacitors, high-power FETs and a controller. Typically you can get 90+% efficiency, regardless of the input/output voltages. But the closer the input and output voltages, the more efficient it can be.

    OK, so most PWM chargers and MPPT use switching regulators. PWM is "pulse width modulated", which is another term for switching regulators.

    Here's the advantage of MPPT. Ignore how you convert with linear vs switching. Depending on the temperature, light levels, etc, a solar panel will have a curve with current vs voltage. Let me give you a few sample points.

    voltage current effective wattage
    17.5 4.75 83.1
    18.0 4.7 84.6
    18.5 4.3 79.5

    So if you torque down with your regulator and drive down the voltage to 17.5, you only get 83w. While at 18.0v it's almost 85w. The feature with MPPT is that it searches up and down the load line to optimize the solar panel's output power. And allegedly it can be a 15% difference.

    This is the difference with MPPT. I hope that helps... I thought it was just linear vs switching the first time I looked into it as well but later realized this detail.

    Tony
     
  8. Kermit2

    AAC Fanatic!

    Feb 5, 2010
    3,795
    951
    Don't look at me...

    I've apparently been doing everything wrong all these years and have only now become aware of it.


    (I never trusted just the word of my wife alone) ;)
     
  9. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,394
    1,606
    No I am not.
     
  10. TexasTony

    Thread Starter Active Member

    Jul 15, 2010
    38
    0
    Ernie,

    Can you give me some evidence to support your claim? I have no problem if you're right, but please give some support.

    PWM = Pulse Width Modulation, and that's another term for Switching Regulator. See:
    http://www.tmworld.com/design/characterization/4388418/Inductance-and-capacitance

    There are LOTS of references tying PWM to switching regulators, just google them.

    Believe me, I don't have a horse in this race. My design will be a switching regulator. If anything, I'd be VERY disappointed that designers used linear regulators in PWM chargers. But that's their issue not mine. My goodness, switching regulators have been used on nearly every board design for at least a decade in PC's? But it would be good to verify this.

    Do you have any guidance on a switching regulator design to use here?

    Thanks
    Tony
     
Loading...