an inverter repair 3000 watt

Discussion in 'General Electronics Chat' started by sdowney717, Mar 17, 2015.

  1. sdowney717

    Thread Starter Member

    Jul 18, 2012
    385
    13
    I blew this out a couple years ago when I inadvertently either shorted output to ground and interconnected inverter output to a running fueled engine generator.
    http://forum.allaboutcircuits.com/threads/rx3152-8-pin-dip-what-is-it.88105/

    I gave up on it then , but I have learned more since and might be able to repair it.
    I found out the output board has 8 Mosfets, IRFP260N, and all 4 on one side are shorted.
    I had before replaced the 16 input board Mosfets IRF3205 and some burnt 10 ohm gate resistors , but when it turned on several gate resistors burned and blew more mosfets and I gave up. But I think now they burnt again due to those shorted output mosfets.

    So if the output of a power mosfet is shorted, will the gate resistors burn?
    What caused these gate resistors to burn?
    RX3152 is a quadrature gate driver for the input mosfets and the inverter has 4 of these, an 8 pin DIP form factor.
    Would those RX3152's be damaged? They look ok, is there a way to ohm them out?

    Pics
    Daughter input board showing 2 black mosfet drivers RX3152
    [​IMG]
    primary input board showing those 2 drivers and other ics
    [​IMG]

    output board with 4 shorted mosfets lifted up
    [​IMG]

    burnt up gate resistors
    [​IMG]

    [​IMG]
     
  2. Dodgydave

    Distinguished Member

    Jun 22, 2012
    4,980
    744
    more than likely the resistors burned out because the mosfet drivers were blown also, i would have replaced all 8 fets and checked the low ohm resistors as well.
     
  3. MCU88

    Member

    Mar 12, 2015
    360
    35
    I feel your pain. Damn 3KW inverter. Now that is some serious power!
     
  4. sdowney717

    Thread Starter Member

    Jul 18, 2012
    385
    13
    How about before putting in new input board mosfets, remove the old mosfets, then power it with 12 vdc and measure the volts going out to the gates?
    IRF3205 specs a max 20vdc above which the mosfets will be destroyed.

    I can understand if the output board mosfets are shorted, then it will short the source-drain on the input board mosfets which will wreck them.
    Would that also cause the gate resistors to overheat?
    But trying to understand how gate resistors burn up. Is the current flow coming from the RX3152? Or would it be coming from the IRF3205 input board mosfets?
    What would they be driving the mosfets, a higher vdc than 12vdc that comes from the batteries?
    I did ohm all four of the RX3152, and they are not shorted.
    Those gate resistors, I think I used are pretty small.

    Do you know by looking if those are 1/8 or 1/4 ? 1/8 watt resistors can handle 125 ma?
    I still have a lot of the gate resistors I used, I also ordered 100 1/4 watt gate resistors. I really don't know what wattage to use for those and I don't remember what I used, looking at them what watt are they?

    I may have used too low wattage for the gates, so I will use the 1/4 watt that I am getting.

    Here it says this about burned gate resistors.
    http://www.powerguru.org/gate-resistors-–-principles-and-applications/

    "Overheated / burned gate resistor The power dissipation as well as peak power capability of the resistor is not sufficient. A non-surge-proof resistor is in use."

    Here says use 1/4 watt gate resistors.
    http://www.edaboard.com/thread297799.html
     
    Last edited: Mar 17, 2015
  5. Dodgydave

    Distinguished Member

    Jun 22, 2012
    4,980
    744
    unless you have the circuit diagram, you cant understand how its blown up.
     
  6. sdowney717

    Thread Starter Member

    Jul 18, 2012
    385
    13
    Here is the Data sheet IRF3205

    Can someone tell me about the gate current, is that in ma? (says nC)
    Trying to see how much current would flow by way of the gate resistor.
    (Total gate charge)?
    Also 2v min then 4v max for the gate?
    What about gate to source v +-20?

    [​IMG]

    First page
    [​IMG]
     
  7. Dodgydave

    Distinguished Member

    Jun 22, 2012
    4,980
    744
    gate charge is 146 nana coulombs @64Amp drain current, it wont give gate current just gate voltage as its such a high resistance.
     
  8. sdowney717

    Thread Starter Member

    Jul 18, 2012
    385
    13
  9. Roderick Young

    Member

    Feb 22, 2015
    408
    168
    The gate of a MOSFET is like a capacitor. Except for some tiny parasitic leakage, there is no current flowing to the gate while the transistor is on, nor while it is off. There is, however, the gate charge which must be deposited to turn the transistor on, and removed to turn it off. Since that happens once per switching cycle, the current is just 2fC, where f is the switching frequency, and C is the gate charge. That's not an actual net current, because it goes two ways, but you can use it to calculate power dissipation in the gate resistor. Apologies in advance to the seasoned designers for my hand-waving explanation.

    In my small experience with switching supplies, it seems that when MOSFETs burn, they become one big blob of resistance, and that includes shorting the source and drain to the gate. I have no way of knowing what happened in your case, but if there was a high voltage on the source of the transistor, it could have shorted to the gate, and burned out the gate resistors as a secondary effect. But why did the MOSFET burn in the first place? Again, I don't now, but if I was doing the debug, I'd look at the input waveforms, and see if everything looks right at the gates without the MOSFETs installed. I'd want to make sure that the drivers weren't stuck driving all gates high at the same time, or floating, which also lets the gates drift up to the threshold voltage and turn on the transistors slightly, making them burn up real fast.

    This writeup is not of an inverter repair, but may have some useful tips: http://pididu.com/wordpress/blog/repair-a-24-volt-electric-bike-charger/
     
    absf likes this.
  10. sdowney717

    Thread Starter Member

    Jul 18, 2012
    385
    13
    This inverter burnt up due to either the output grounded or interconnected to a running generator. That ruined the output board mosfets, they shorted one bank of four, then the input board mosfets were looking at direct shorts of their outputs which destroyed them. I was also thinking that as that link said, gate resistors of too small watts can burn. What made me not worry about the wattage the first failed repair was the smd gate resistors look really tiny.

    Most of the bad mosfets were simply shorted with low resistance, some mosfets were very visibly burnt.
     
  11. Roderick Young

    Member

    Feb 22, 2015
    408
    168
    Ah, very helpful! The failure was due to overload. That suggests to me that the output MOSFETs were the first thing to go, through no fault of the driving circuitry. If you still have the burned MOSFETs, take them out of the circuit, and test whether any of them have the gate shorted to drain, by any measurable resistance other than infinite. If so, then the high voltage from the output got back into the driver stage through the 10 ohm resistors, burning out the drivers most likely, before the resistors burned. It's imperative to check the waveform coming out of the drivers, or remove the parts and test them on a breadboard. You're only testing for whether they work at all. I'd place high probability on the driver being burnt out (it was in the case of the link I previously gave), and if the driver output is floating, then if you subsequently change the MOSFETs, the new ones will burn out, too, because their floating gates will inevitably drift up and turn on the transistors.
     
    sdowney717 likes this.
  12. sdowney717

    Thread Starter Member

    Jul 18, 2012
    385
    13
    That is interesting to think the mosfet drivers might be bad.
    Some kind of current backfeeding?

    Do these drivers put a positive voltage to the gates to turn on the N channel mosfets?
    Would that gate voltage be lower than the source-drain volts?
    So if the mosfets short, they backfeed power through the gate resistors into the drivers?
    I think they are that RX3152 8DIP.

    I dont have a scope, is there a way to check them otherwise? From your description sounds like they would be shorted low ohms internally on their connections to the gates somehow? I checked them with ohm meter and they have high ohms, nothing seems shorted?
    When I remove the mosfets, I will trace them some more and verify that.

    But if they are not shorted, then should be ok to try it you think?

    They are fairly cheap insurance to replace them.
    I appreciate your post! Thanks.
     
    Last edited: Mar 18, 2015
  13. Roderick Young

    Member

    Feb 22, 2015
    408
    168
    That's the idea. Generally the MOSFETs will have either source or drain connected to some high voltage. The driver chip is generally not made to withstand more than 20 volts on the output when off, and if it's on, it's not made to withstand continuous high current, only the brief surge to charge or discharge the gate.

    More likely, the output stage of the driver is blown OPEN, meaning that the gate of the MOSFET will be open, and you'll just be wasting any MOSFETs you replace. Make sure that the drivers are working, first. Those parts are simple - try looking up the MC33152 - is that your chip? You can get a datasheet from mouser.com or just about any vendor. If you can remove the chip from the circuit, you can supply a +5v or GND to the input of each section, and see if the output switches appropriately just using a DVM. You'll have to connect the power pins of the chip too, of course.
     
  14. sdowney717

    Thread Starter Member

    Jul 18, 2012
    385
    13
    I ordered some off Ebay, the chips say RX3152. Ebay has that exact part. Interesting that MC33152 looks to be same as RX3152. I think your right, good idea to replace them.
    Would they be the same functionality?

    Also looking at the pinout, where is it getting the power to send to the gates?
    vcc?
    Is it possible something beyond these 3152 drivers would also have burnt?
    Something that produces vcc?

    I assume logic inputs are coming from some kind of controller chip that signals the 3152 to go on and off?

    [​IMG]

    Can I use the same small 10 ohm gate resistors I show in the picture? I did order more which are 1/4 watt, but I dont know if the leads will fit the holes.
    I dont know if I used 1/8 watt or 1/4 watt gates in the pics with burnt gate resistors.

    Adding some more info.
    I separated all 3 boards, there are 2 input boards and 1 output board.
    I removed all the mosfets. only two output board mosfets were shorted and three input board mosfets were shorted gate to drain.
    The RX3152 drivers were difficult to remove.
    I finally decided to crush 2 of them and desolder legs one at a time. The other two I did remove intact.
    I then tested for a short between red and black input wires, and it was ok, all it did was charge up the capacitors.
    The input caps are 3300 uf 16v. There is an open position for more input caps , I could add 2 more total on the primary board. Do more caps help with surge?
     
    Last edited: Mar 20, 2015
  15. sdowney717

    Thread Starter Member

    Jul 18, 2012
    385
    13
    It is working!
    I got all the parts and I am running from a 3 amp hour 12vdc battery and a charger.
    The load display bar does not light up but that might be because it is putting out less than 100 watts.

    This sat for a while outside, I even think the output board was sitting in rain water for a few months in a tub on the deck. The tub cover had a crack. The green-red led pilot had rusted in it's plug. I used some rust buster oil on it and it is working good. Alarm on low volts is also working and shut down too. This inverter auto restarts if the 12 vdc power comes back up.

    Here it is running, I have it apart to see how it does. All the mosfets are warm. I assume if they are a little warm then they are working ok?
    [​IMG]

    I am still waiting on some heat sink grease. All the mosfets have those grey silicon pads, do they still need to be greased?
     
    Last edited: Apr 3, 2015
Loading...