mosfet h-bridge ir2110 blowing issue

Discussion in 'The Projects Forum' started by vincent.aaron.vega, Mar 15, 2014.

  1. vincent.aaron.vega

    Thread Starter Member

    Mar 15, 2014
    34
    0
    Hi guys,

    i'm working on h-bridge with two ir2110 and ir3205 mosfet's.
    30Khz PWM comes from atmega168 (gradually rising form 0 to 200 and back to zero in 1.5 sec)
    H bridge runs a brushed DC 18v 1ohm motor form cordless drill (non loaded)

    I used this design to start:
    http://www.eeweb.com/project/circuit_projects/pwm-dc-motor-controller-using-mosfets-and-ir2110-h-bridge-driver
    and modified it when i found this: (figure 8)
    http://tahmidmc.blogspot.com/2013/01/using-high-low-side-driver-ir2110-with.html

    Designs are same except second has lower values of caps, superfast diodes and pull down resistors form gate to source, and bunch of decoupling caps.

    The reason i switched to second design is that previous missed the gate to source resistors. After few minutes mosfets opened themselves and short circuited the battery(2*12V 45Ah in series) making small fire. Board was scrapped but the mosfets were intact.
    Also values of the charge capacitors seemed too high(100uF) in the first design. (for 30khz PWM)

    I retained bypass schotky didoes on the second design.

    I tested the driver part of the scheme and it has valid outputs. Everything is connected correctly. I separated(different PCB) the controller form the drivers and the drivers from the H-bridge. It is pretty clean design. I have ceramic 100nf decoupling everywhere. Everything is grounded to a common household ground. I use anti static band all the time.

    I used 7812 for 12V charge pump, 7805 for 5V input

    When i switched to second design i got this problem.
    PROBLEM: when the controller applies the pwm and tries to run the motor, short circuit occurs trough one side of the H bridge and lower schottky diode burns and after that the motor starts.

    I will test in the morning again with the 100uF charge caps but..

    I'm asking you guys if someone can confirm this design as valid and maybe suggest why the problem occurs?
    http://4.bp.blogspot.com/-Lld4HStSKlw/UPvns0ckrZI/AAAAAAAAAaA/RFqfmsuv7bM/s1600/IR2110+-+7.png

    Any help will be appreciated.

    Regards
     
    Last edited by a moderator: Mar 16, 2014
  2. MaxHeadRoom

    Expert

    Jul 18, 2013
    10,564
    2,379
    Did you try removing a Mosfet pair and just run half bridge, either Q1 & Q4 or Q2 & Q3?
    Which are you retaining full on and which are you PWM'ing.
    OK I see you are PWM'ing both.
    Max.
     
  3. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    The schematic looks ok except for the bypass caps on the bus.

    You should put some electrolytic caps directly from drain of the upper MOSFETs to source of the lower MOSFETs. Something like 2000 to 6000uF

    The distance between 2110 and the MOSFETs should be as small as possible.

    Don't pass load current through gate return signal wires (that's the wire that goes form source of lower Mosfet back to the 2110.

    Use deadtime on the input PWM. I wouldn't rely solely on the diode/resistor combination in series with the gate.

    Have a look here: http://www.irf.com/technical-info/appnotes/an-978.pdf page 12

    Last but not least. Never work with batteries that can easily provide 100s of Amperes (when shorted) without a FUSE.
     
    vincent.aaron.vega likes this.
  4. vincent.aaron.vega

    Thread Starter Member

    Mar 15, 2014
    34
    0
    Hi,

    i will continue tonight, i will try half bridge and report back.

    About those bypass caps on the bus. Shouldn't they be obsolete when i'm using 10 times bigger battery then the motor needs?
    It is correct that when i was testing earlier, 15A PSU as charger was connected parallel to batteries. It probably has big caps on the bus inside.

    i will use shorter wires for driver to gate connection,

    "Don't pass load current through gate return signal wires (that's the wire that goes form source of lower Mosfet back to the 2110."

    How did i pass it? that would be from ground to pin 11, 13, 2?

    There is 4 seconds deadtime between switching the sides of h-bridge. It blows up on first rise.

    I will study app notes more if nothing helps form this list.

    And i will set the fuse when thing stop short circuiting. For now a hand on the bus switch will have to do. 60A Fuses are expensive.

    I noticed nobody commented the cap values. So boost cap(C1) can be 22uf or 100uf it's more or less the same?

    I will report back when i make the changes.

    Thanks
     
  5. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    1. The difference between a bypass cap and a battery is its internal resistance.
    Also, the caps are mounted NEAR the MOSFETS, they provide peak currents and they also do not allow the voltage on the bus to rise to abnormal levels.
    Do not forget that the wires from the bridge to the battery have an inductance. Believe me you need those caps on the bus, it's a necessity. Read the AN978.

    2. The bootstrap caps are being calculated. they depend on switching frequency, min/max duty cycle, MOSFET gate charge etc.. Should be ok with 22uf.

    3. What PWM uses 4s deadtime? Do you mean 4us?

    4. Do you have an oscilloscope?

    5. As for the shared load current: They wire that comes from the source of the lower MOSFETs back to the 2110 should not carry the load current. Maybe you should post a picture of your layout.
     
  6. vincent.aaron.vega

    Thread Starter Member

    Mar 15, 2014
    34
    0
    Ok i'll put 10,000 uF 100V cap. they are around 10$ here.
    I've read the an978(some parts), but the question about the battery bus cap is lack of basics on my side obviously. I'm not an electrical engineer.

    I'll put 22uF back.

    Is 100nF across the brushes ok. I don't need other cap here?

    I'm running simple testing program. One direction of the bridge 0 - 80% - 0 duty cycle, pause 4 seconds, other direction 0 - 80% - 0 duty cycle (in 1.5s rise and fall of duty cycle)

    I would love to post pictures but my phone has 5mp camera without autofocus and the pictures are blurry when shot at close range. Useless camera.

    about this wire that comes out of the lower mosfet's source(s). These are connected together and go back to battery.. i cant see a way i can ground trough ir2110? ir 2110 has it's own ground wire. i'm still confused about this.
    Can i add something to the circuit to prevent this?

    Thanks,

    Regards
     
  7. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Please read the AN again, it's all explained there. See attached picture.

    You can also attach a photograph of you layout.
     
  8. vincent.aaron.vega

    Thread Starter Member

    Mar 15, 2014
    34
    0
    Hi, thanks for the picture, i read the AN parts about return wire. i think in getting the picture.

    So to confirm, i need twisted pair consisting of low side gate wire and ground. Ground wire should be connected as close to the source as possible and routed(physically) away from bridge ground lead?

    Stores were closed on sunday so i got the 4700uF 63V cap today and i will continue testing tonight, and report back.

    I've got one more question, AN says that upper source to lower drain leads should be as short as possible. I placed mosfets in a row so one side has 6 cm lead and other has 2.5cm lead between upper and lower mosfet. Is this close enough? and it says that sharp bends should be avoided. I don't see a way to do this without some bends. Is 90° too sharp?

    I posted a schematic with explanation
     
  9. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    No, twisted pair is not necessary, only short they should be, gate and return.
    6cm is maybe not optimal but shouldn't be a problem.
    Don't worry about the bends.

    What you should know about the 2110, the driver high-side can latch up when VS goes below COM. Means it stays on even if you command it OFF. You see that's a bad thing. :D I managed to make it latch up easily.

    Therefore I recommend connecting a Schottky from:
    - COM to LO
    - VS to HO
    - COM to VS

    Any negative going transients will be attenuated by the schottkys
     
  10. vincent.aaron.vega

    Thread Starter Member

    Mar 15, 2014
    34
    0
    How did you trigger this state? meaning when does VS go below COM.
    Transient currents? shouldn't these gate to source resistors handle them?

    Can these Schottky have low A rating? i have some bat42 left from other project. Or should they be 1A like 1N5818 i'm using on H bridge.

    Oh and one more thing, do you consider adding a 33uF bipolar capacitor across the brushes, a smart thing to smooth out the shaft friction bumps. The motor will be running a CNC M16(yes metric thread, it's cheap amateur project) shaft directly.

    Oh and one more one more thing, something bothers me on this scheme(same picture as in previous post), wouldn't it be better if the R3 and R7 were pulled down with 1K resistor to the ground instead of source here. I mean, aren't they floating when the bridge is not working?
     
    Last edited: Mar 17, 2014
  11. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Small schottkys should do the job for now. The ones we use have usually 1 to 3A rating though.

    As for the negative going transients it's also explained in the app note. See every wire, every trace as an inductance and you will understand.

    One example: Output driver turns on, gate charges through the gate resistor and the wires and traces, i.e. through inductances. The driver turns off, the inductance (wires) can cause negative transients. It's on page 9.
     
  12. vincent.aaron.vega

    Thread Starter Member

    Mar 15, 2014
    34
    0
    Thanks i get it now. You've helped me a lot.

    I edited the previous post before i realized you answered already, so i'll repeat the question

    Do you consider adding a 33uF bipolar capacitor across the brushes, a smart thing to smooth out the shaft friction bumps. The motor will be running a CNC M16 shaft directly.(yes metric thread, it's cheap amateur project)

    Wouldn't it be better if the R3 and R7 were pulled down with 1K resistor to the ground instead of source here. I mean, aren't they floating when the bridge is not working?
     
  13. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Not sure about the brush caps, I think we use small ceramics inside the motor.

    NBo, R3 and R7 are correct. gate to source. Can be higher though. But 1k will also work.
     
  14. vincent.aaron.vega

    Thread Starter Member

    Mar 15, 2014
    34
    0
    ok, thanks,

    i'm putting it together as we speak, i will test this nigh and report back.

    Put some pictures if they turn out good.

    Regards
     
  15. vincent.aaron.vega

    Thread Starter Member

    Mar 15, 2014
    34
    0
    Update:

    I installed the 4700uF cap, dedicated negative return wires, shortened the leads. I've put 10uF bootstrap capacitors.

    Only thing i didn't do yet is schottky's on the driver how you suggested, i'll have to make a new driver board to accommodate for them. The space wasn't accounted for on this board.(i'll do it this night)

    However i had to test it. (i couldn't wait)
    Nothing burned this time. The motor ran one side for a second. Made a pause 4 seconds. However when it started turning the other side, it reseted the controller(atmega168) and stopped.

    Any ideas what kind of design flaws cause controller resets?

    I will prepare an updated schematic(complete) tonight with all the elements and correct values, here are some pictures to get the feeling what it looks like.

    From right to left:
    1. cnc controller board
    2. bridge driver board
    3. mosfet h-bridge board
     
  16. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    yes, a complete schematic showing also how controller and driver are connected will be helpful
     
  17. vincent.aaron.vega

    Thread Starter Member

    Mar 15, 2014
    34
    0
    Hi,

    I'm done, but i see where you are heading, i've got nothing between the Atmega and Ir2110's except wire.

    I've noticed an error while drawing the scheme. I've got 25V capacitor on 7805 input to gnd on controller board. And when i will use the batteries with PSU turned on, it will give 26.8V. I'll replace that tomorrow, and redo the driver board to accommodate for schottkey's on IR2110's.

    Here are all three boards, hope you find something.

    Regards,
     
  18. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Where do you create the deadtime? In the Atmega?
     
  19. vincent.aaron.vega

    Thread Starter Member

    Mar 15, 2014
    34
    0
    Yes. In atmega.
     
  20. vincent.aaron.vega

    Thread Starter Member

    Mar 15, 2014
    34
    0
    Judging from your question, i guess there is nothing wrong with schematic..

    I'm too tired to work on this tonight, however if anyone spots a design flaw, i would be more than welcome if they write it down.

    Anyone knowing the common causes of controller resetting when operating IR2110 and h bridge, or any switching circuit.

    Maybe to add some parts that are missing? Diodes on pwm signals, capacitors somewhere? higher value capacitor somewhere?

    I'm using 10cm cables from atmega PWM outputs to IR PWM inputs.. could this be an issue? I will have to use longer wires because i have to mount boards for two more cnc axis, to the left of the present ones (there are pictures above). Maybe optocouplers for pwm output?

    How does this reset even occur? voltage drop on the bus(i use 2 12V 45Ah batteries in series) or the circuit closes somewhere where is shouldn't?
     
Loading...