Cutout vs Short type solar panel regulation circuit

Discussion in 'The Projects Forum' started by russpatterson, Jul 30, 2011.

  1. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    This moved over from this thread. http://forum.allaboutcircuits.com/showthread.php?t=57345

    If you're builing a solar controller save yourself some time and take a look at this.

    I've been using a cutout type solar panel regulation circuit (to control the amount of current that flows from panel to battery). There are several troubles with the current design as it requires a P-Channel MOSFET and some extra parts and doesn't work that great.

    The suggestion is to switch to a short type panel controller and use an N-Channel MOSFET, which are cheaper, run cooler, more efficient etc.

    Here's the original panel regulation circuit

    [​IMG]



    Here's the schematic for the short type method using an N-FET.

    [​IMG]

    It seems as though it will work, however I can't get the simulator to play along, which is probably an issue for another forum.

    Any input on the proposed circuit?
     
    Last edited: Jul 30, 2011
  2. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    I have my own question: I have been using my cheap bench supply, which can regulate current, as a substitute for the panel when I test designs. If I set the bench supply to a max of 500mA is it going to be OK with being shorted like that? I know the panels don't mind but I'm not so sure about a switching supply like that.
     
  3. Bernard

    AAC Fanatic!

    Aug 7, 2008
    4,170
    395
    Add some resistance between power supply & FET to represent panel internal resistance.
     
  4. nickelflipper

    Active Member

    Jun 2, 2010
    280
    35
    How many watt PV panel are you using? Open circuit voltage? shorted panel amps?

    To be fair, a high side PFET switch is perfectly OK. It looks like in the original circuit you tried to use a NFET for a high side switch, which was discussed in the previous thread. The PFET switch uses less parts in a high side configuration as an NFET, or the same as the shorting circuit NFET.

    It would be pretty easy to find a PFET with an Rdson of less than 0.1Ω. The power loss would be P = (I)2 x R. So using your 500ma, that would only be 0.025W, or less than 1% for say a 5W panel.
     
  5. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    Thanks, I'll try the series power resistor. I've got 3, 100 Ohm, 10W sand resistors, giving 33 Ohms. That should get me in the 500mA range at 15V.

    I tried the circuit just now on some perf board with a 12 watt panel and it works. I can turn on or off the panel with a logic level 0 or 5V on the N-FET gate. Current flows and the battery charges when the gate is 0V and doesn't flow when the gate is 5V, and the panel's shorted to ground. Pretty cool. Thanks for that one THE_RB.
     
  6. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    I'm using anywhere from a 10Watt to 210Watt panel, depending on the application. So from 500 mA to 10A. My plan is to populate the board with different parts depending on the amperage.

    I guess it's hard to see, but the original circuit is using a P-FET. I've found that P-FET's have a much higher RDSon, say 39mOhm vs. 8mOhm for an N-FET. Also the lower the RDSon in a P-FET the higher the gate charge and the slower they are to switch (due to the whole RC time thing driving the gate).

    I'd be interested to see the high side P-FET driving circuit that uses the same number of parts as the N-FET short style one above. To be fair there should be a 150Ohm series resistor on the N-FET gate since I'll be driving it with the PIC.
     
  7. nickelflipper

    Active Member

    Jun 2, 2010
    280
    35
    This is what I get for entering a topic late , my apologies, the 500ma reference in post #2 threw me. So while a 500ma load can easily be switched by a logic level pfet (no driver required), a multi amp load can be more of a challenge as you say. The topic seems vaguely familiar.
     
  8. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    Yes, I've been working on this particular section for quite a while. So it takes more amperage on the gate to switch more load amperage? I guess that makes sense, and I've observed some of that behavior but didn't know that to be the case. Do you know what the symptom is for switching a large >3A load with low < 20mA gate current? Would it not switch as quickly and spend more time in the linear region?
     
  9. nickelflipper

    Active Member

    Jun 2, 2010
    280
    35
    Slow switching times mean more heat in fast switching circuits.

    Finally found the appnote I was looking for. The Microchip AN874 talks about component selection for a Buck Led Driver. Based on a micro pin drive current of 20ma, 5% of time spent switching, and a switching frequency of 100kHz it came up with a total gate charge Qg of 10nC.

    Plugging in info into the Digikey search is kinda a hunt and peck situation. There are very few logic level pfets that will drive a load over 3A with low Q at 4.5V. A possible candidate for a solar smps powered by a micro would be a DMP3056LSS
    It has a Vds = 30V, 6A Id, and a low Qg of 6.8nC.
     
    russpatterson likes this.
  10. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    How fast do you want to switch it?? Commercial solar regulators usually are limited to switch about once per second, so anytime battery volts >X the panel is shorted for 1 second (no charging to the battery for 1 second) then the battery voltage is then tested again.

    They usually have a LED indicator to show regulation in action, which is user friendly as the frequency of the 1 second regulator LED pulses gives the user an idea of how charged the battery is.

    You don't need hardly any gate current at all, as long as you have enough gate voltage (Vgs) to cause the FET to turn on well.

    If you are driving the FET gate directly with the PIC you should use what is normally called a "logic level" FET which will turn on well with just 5v Vgs. You can drive it through any resistor really for this application, I would use 10k as it will save your PIC in case of a FET failure short circuit where solar 22v may appear on the FET gate.
     
  11. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    I have the "switch very infrequently" design working now. However this doesn't really charge a SLA battery to its potential very well. So I want to switch it at 2Khz - 5Khz so I can have a three stage, voltage controlled, charger. It holds the voltage to a saturation level ~14.4V for a few minutes (each battery has all these specs in the data sheet) then down to a float charge level of like 13.8 until it's ready to run. So it effectively limits the current to the battery to hold the voltage at the pre-defined level. You can also compensate the voltage levels based on battery temperature, which is on the list.

    I built a 3 stage charger using a buck converter that works pretty well but I'd like to have my standard board do the 3-Stage, voltage controlled, charging too, (w/out the whole DC/DC buck circuit). If you want to run lights at night then getting the battery fully charged is key.
     
  12. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    [​IMG]

    So after some testing I've noticed a problem with this design. When the power is off for the controller, the MOSFET still conducts power to the battery and can/will overcharge it if left alone. Does anyone have any clever fixes for this? I need a way to ensure that current won't flow from panel to battery when the circuit is not powered.

    One thought I had was to just make the on/off switch a "request" to turn the power off and just put the uC in a sleep state with a 5 minute cycle to check and make sure the pin driving the MOSFET gate stays at 5V. However this is not optimal.

    Any input is appreciated.
     
  13. wayneh

    Expert

    Sep 9, 2010
    12,087
    3,027
    I believe you can turn the FET around to avoid current flowing through the body diode. That's the way they get oriented when used in "ideal diode" configurations.
     
  14. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    Do you mean connecting the panel to the source and the drain to ground? I'm not a MOSFET expert but I didn't think it would work that way. Current flowing in the reverse direction. I would need to drive the gate voltage 5V higher than the source voltage, ( the source voltage would be the panel/battery voltage, 11-14V). At that point I might as well do the whole high-side driver thing.
     
  15. wayneh

    Expert

    Sep 9, 2010
    12,087
    3,027
    I spoke without thinking too much. But yes, the active rectifier approach is to look for a small voltage across the body diode, when Vs exceeds Vd. You then turn on the FET (Vg >> Vs, as you noted) to eliminate the diode voltage drop, and current flows in the same direction it would normally through the body diode. That's in the reverse direction we normally use for N-channel MOSFETs.
     
  16. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    [​IMG]

    @Wayne, this is what you intended? I can see that without any power to the gate; the current from the panel would be shorted to ground, which is great. However when I put 5V on the gate, the MOSFET will not allow current to flow through the body diode and all the panel current would flow through D3 to the battery? Vgs at that point would be somewhere in the area of (5V gate, and 14V battery) of -9V.

    I Googled active rectifier and did not find any useful information. Can you point me at any that explains that theory?

    Thanks!
     
  17. wayneh

    Expert

    Sep 9, 2010
    12,087
    3,027
    I'll try to dig up some references. The only one I have handy is here.

    I'm pretty sure that a MOSFET conducts in either direction whenever Vg > Vs. So again, when off, only the body diode is carrying current and there is a voltage drop across it. That voltage triggers application of voltage to the gate, and then current flows with little voltage drop.
     
  18. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    My gate voltage will be significantly lower than the source voltage (Vg = 5V, Vs = 12V-14V (when panel not shorted and connected to the battery). So will a Vg of 5V stop current from flowing through the MOSFET when the Vs is ~14V?

    I will give it a try on the bread board and see what happens.
     
  19. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    I finally got around to setting up the experiment with swapping the ground from source to drain and it does not conduct current from panel to battery. However I think the addition of this 15V zener will add some safety to the circuit.

    Even if the uC is non-functional, if the battery gets above 15V the MOSFET will conduct and short the panel. There will be some heat due to the linear nature of the control but worst case the MOSFET burns up and the battery is safe from an uncontrolled overcharge situation.

    [​IMG]
     
Loading...