High side MOSFET driver question

Discussion in 'The Projects Forum' started by russpatterson, Mar 26, 2011.

  1. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    I've been puzzling over this, off and on, for months. I want to switch the panel voltage (high side) on/off with an N-Channel MOSFET.

    I've been told that a high side driver is the way to go. I have a functioning circuit with a P-Channel MOSFET here:

    http://backyardsolar.blogspot.com/

    I have a LTC4440 (High side driver from Linear, datasheet attached) on my breadboard now and I'm trying to figure out how to make it work based on the datasheet. It says the the gate output (TG) swings between TS (source voltage, the 12V battery positive terminal in my case) and Boost (where I get the boost level voltage is what I'm struggling with).

    For some reason I thought high side drivers had internal charge pumps that magically created the voltage necessary to turn on the MOSFET themselves. I guess I need to create that voltage level myself? Is that what they mean by the term "bootstrapped supply"? How is this supply level usually produced?

    Thanks for the help in advace.
     
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    In your Eagle diagram, add a wire from node PANEL_VIN to the node connecting C1's positive side, D1's anode, and U$1 pin 3 (Vcc).

    One problem that you might encounter is if the panel has low or no output, the body diode in the N-ch MOSFET may conduct current from your battery bank through the solar panel. This would not be good.

    Another problem you may wind up with is the boost cap charging and discharging. This gate driver does not have an internal boost circuit; that's performed by your C4's low side connected to the MOSFET source terminal; C4 charges via D1 when the MOSFET is turned off. So, in order to charge the boost cap initially, the solar cell bank must output your battery bank's voltage + the Vf of D1 + the low threshold voltage of the driver IC (8v I think) in order to start switching.
    As far as keeping the cap charged; you'll have to turn the MOSFET off in order to charge the cap, and you'll have to do this fairly often - likely at minimum several times per second. I don't know what the leakage rate of your cap will be; that's another source of voltage loss.

    Might be better to just stick with your P-ch MOSFET. At least you can turn it on if your battery bank is > 10v.
     
    Last edited: Mar 26, 2011
  3. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    Great info. Thanks. I'll give it a try.

    I could add a diode between the panel and the Drain to avoid the battery draining into the panel at night.

    If I stick with the P channel design then I need to change the circuit or add a driver so that it switches cleanly. The 1K - 2.2K resistor makes the gate voltage rise way too slow.

    This N channel design here would turn on if the battery is anywhere between 10V - 14V as long as the panel voltage was high, which it would be with any light at all. Am I missing something?
     
  4. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    That'll be a permanent voltage drop, thus wasting power, depending on which diode you use.

    You need to re-post that circuit, or link to the thread where the circuit is already posted.

    What's "high"?

    Look at the charge path for C4 in your re-worked schematic.

    C4 needs to be charged up to higher than the low-value under-volt threshold of the driver IC.

    So, your solar cell voltage will have to be a minimum of:
    (Your battery bank voltage) + (the Vf of D1) + (the low-value under-volt threshold of the driver IC)
    in order to turn on the MOSFET.
    The Vf of D1 might be 300mV or so. The low-value under volt threshold is around 8v. Your battery bank might be around 12v. So, that's 20.3v needed in order to turn on the N-ch MOSFET.
     
    russpatterson likes this.
  5. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    The under voltage lockout threshold, UVLO. That's what I was missing. Thank you! I can get TG, the gate voltage to swing now. However it causes havoc on my bench supply. I can hear the current limiter relay switching tens of times per second. I'm using the bench supply for the panel voltage. Probably due to the cap discharging like you described.

    Open circuit voltage of the panels I'm using in about 22V or 30V, depending on the panel. The UVLO is max of 7 on the LTC4440. So worst case the battery is fully charged, 14V, plus the Vf of the diode 0.3, and UVLO 7 is 21.3. So you're right, it's pretty close to not being able to turn on. Plus the voltage will collapse immediately once the MOSFET turns on, then you've got to switch it off before the cap drains out. Sounds like you'd be wasting some power all the time keeping that cap charged.

    I attached the schematic that contains the P-Channel circuit. I thought about having two transistors, one to pull the gate low (like it is now) and another that would pull it high using the panel voltage. That way I could eliminate that big 2.2K 1/2 watt resistor. However there's the chance that it could short momentarily during switching. Also with the cost of the two transistors I could just get a driver IC.
     
  6. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Switching time shouldn't be much of an issue with your P-ch MOSFET, as it's very low frequency.
     
  7. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    I was hoping to take the charger to the next level and do a 3-stage charge and limit the current from the panel by switching it at relatively high frequency, hopefully 20Khz or thereabouts. Currently the MOSFET heats up pretty good at 7Khz switching with a 1.1K resistor in there, and only 1 amp max current. At 3 amps it would probably cook itself to death.

    Take a look at the attached scope picture. The rise time on the gate (scope channel 1) is pretty bad. Also I need to switch faster because the battery (channel 2) is swinging from about 11V to 13V on every cycle. I'd like to be able to hold it steady. Maybe I need to add an inductor/capacitor on the output to smooth things out? The scope's not showing the correct frequency for some reason in that pic.
     
  8. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    See the attached for a simple transistorized P-ch MOSFET driver. The Zener diode keeps Vgs from exceeding 12v. You could use a 10v of the same type. The Zener I chose needs 3mA current to develop the full voltage across it.

    The solar panel input supply has a high resistance; that's why you'll see an 18v swing on the gate voltage.
     
    russpatterson likes this.
  9. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    Thanks for that! Could you post the LT Spice .asc file for that too? What is the purpose of D2? To stop the leakage from battery to panel?
     
  10. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I'll attach it.

    If you don't have the MOSFET that I used, you could use another.

    Yes.
     
  11. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Rats - you won't have the Zener or the batt_SLA models either.
     
  12. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    The simulation ran fine. I think I got the battery model, and possibly the zener model from you previously. I ordered the parts for this design that I didn't already have. However when I looked up that Zener on digikey it said that part number was a 12V zener. So I swapped it out for a 10V one. I also sent out for a PCB of the circuit. I'm hoping my switching times will look great with this one.
     
  13. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Yes, it WAS a 12v Zener. The idea was to make certain that the MOSFET was fully turned ON. But you might do OK with the 10v Zener.
     
  14. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    Well it works! I used the 10V Zener. I will get some 12V Zeners and try them.

    It switches pretty clean. I'm happy with it. I get some ringing at the gate when it turns on. Considerably more when I run about 2 Amps through it. Check out the pictures. At 2 Amps the ringing looks it gets up to 37 volts at the peak of the ringing. Should I be worried about that?

    I'm using an IRF4905PBF, with Rds On of 28mOhm. At the 13V, 2.1 Amp run it gets pretty darn hot. I put a 1, 1/4" bolt through it and it got too hot to touch after a few minutes.

    Thanks for the circuit. This is the best high side switcher yet.
     
  15. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    This scope pic didn't make it on the last post for some reason.
     
  16. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    That MOSFET has a very large gate charge; 180nC.
     
  17. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    It seems like you either get a high gate charge or a high RdsOn with the P-Channel MOSFET's. Everything's worse with higher amperage rating. I tried a 20 Ohm series resistor on the gate and it didn't make too much difference with the ringing. I ordered some other P-Channels to try and compromise between RdsOn and gate charge.

    IRFU5305PBF-ND - 65MOHM, 63NC
    IXTP32P05T-ND - 39 MOHM, 46NC

    Is that kind of ringing normal? How do you go about fixing it?
     
  18. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    What about putting a 22V Zener on the gate. That might keep that first spike and the resulting ringing down.
     
  19. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    No, put a 2nd 12v Zener between the MOSFET gate and source terminals.
    The MOSFET's Vgs should never go higher than the source terminal plus the Vf of the Zener, and never go lower than the source terminal -12v or so.

    Instead of the 2N2222/2N2907 pair, you might go to transistors with a higher gain and current rating, like ZTX792A (pnp) and ZTX869 (npn) transistors from Zetex.

    Do you have caps to ground at the collector of Q1? That would help speed turn-on times, particularly on a breadboard.
     
    Last edited: Apr 1, 2011
  20. russpatterson

    Thread Starter Member

    Feb 1, 2010
    351
    16
    I put a 10uF cap from Q1's collector to ground and that quieted down the ringing quite a bit (see pics). It spiked a little more than that with a couple of amps running through it so I added a 100uF cap between the PMOS drain and ground. That helped quite a bit too.

    I think you are right on getting some beefier BJT's. I will order those next time.

    When I look at the data sheets for the P-Channel MOSFET's the Vgs is rated from -2V min to about -4, or -4.5 V max. So shouldn't I be holding the gate voltage to about 4 volts of the source?
     
Loading...