H-Bridge/Supply Motor Drive Ringing

Discussion in 'The Projects Forum' started by Deadstar1312, Apr 24, 2012.

  1. Deadstar1312

    Thread Starter New Member

    Apr 2, 2012
    16
    0
    Hey,

    I am having some problem with some bad rining on my power supply which is causing all kind of problems. I would like to understand why this occurs and figure out how to stop it. Any help resolving this would be appreciated!

    I am trying to drive a linear actuator but am finding some strange ringing when I initially retract the actuator which dissapears quickly. The problem is this ringing appears all over my system! When using an alternative rotational motor I do not see any ringing, the problem is I am stuck with this linear actuator!

    I have made an H-Bridge using the HIP4081. I am using the demo board provided by Intersil described in the data sheet with IRF520 FETs and have put two 680uF caps across the supply as suggested. I am driving the HIP4081 with an 8 bit AVR, feeding two 5V phase correct PWM signals into ALI and BLI and pulling BHI and AHI high.

    I can see the rininging on the battery, 4 gates, and two PWM drives. The ringing seems to be full scale and have a frequency of 10 or 20 MHz.

    I've uploaded two pics of captures of the two low side gate drives and two high side gate drives. These are not at the same time period (I only have two channels!)

    [​IMG]
    [​IMG]

    [img=http://img407.imageshack.us/img407/5559/ablow160lin.th.jpg]
    [img=http://img13.imageshack.us/img13/2852/abhigh160.th.jpg]

    I am using the phase correct PWM method described here (http://web.mit.edu/orca/www/2004_motorController.shtml) and should be avoiding any shoot through.

    If anyone has any thoughts I'd appreciate it.

    Colin
     
  2. CDRIVE

    Senior Member

    Jul 1, 2008
    2,223
    99
    Don't make us troll for the schematic. Please post it.
     
  3. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Please post also a picture of your layout or a drawing of how the bridge is wired physically (not just the schematic).

    I suggest reading this document: http://www.irf.com/technical-info/appnotes/an-978.pdf

    The amount of ringing will depend on stray inductances/capacitances in your circuit.
     
  4. Deadstar1312

    Thread Starter New Member

    Apr 2, 2012
    16
    0
    I've made a quick schematic:

    [​IMG]

    I've taken a pic of the board below.

    [​IMG]

    The pic includes a snubber I tried to add (33 ohm resistor and .1 uF cap). To the left is the motor. The stripboard is a breakout for the integrated feedback pot on for the actuator. The dev board schematic is in the HIP4081 datasheet (http://www.hvlabs.com/files/HIP4081application.pdf figure 11).

    Colin
     
    Last edited: Apr 25, 2012
  5. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    In a real board the big power supply caps should be near the bridge, but Maybe it doesn't matter here.

    Have you monitored voltage from BHB to BHS and AHB to AHS when this happens? How do they look like?
     
  6. Deadstar1312

    Thread Starter New Member

    Apr 2, 2012
    16
    0
    I've included captures of BHB (red) and BHO (yellow) and AHB (red) and AHO (red) below:

    [​IMG]

    [​IMG]

    From hunting around I can't find anything similar to this. Most examples of ringing appear to be at a lot lower levels. Also my ringing isn't constant, just at the start of movement and only really apparent when retracting my linear actuator. Is this likely to be due to the characteristics of my motor rather than the bridge?

    Thanks again,

    Colin
     
    Last edited: Apr 26, 2012
  7. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Why do the bootstrap voltages have the same waveform as the gate signals?

    They should look like here in figure 6: http://www.intersil.com/data/an/an9405.pdf

    You say "at the start of the movement", so may be only be happening at higher current levels. Can you measure the motor current with a scope (HALL sensor, current clamp etc)?
     
  8. Deadstar1312

    Thread Starter New Member

    Apr 2, 2012
    16
    0
    I think I might have posted the wrong images!

    Here is AHO (red), AHB (yellow) and BHO (red), BHB (yellow). It looks like the bootstrap voltage is fairly constant apart from when it rings. Would you expect to see much of a dip like in the figure you pointed out? I am running everything at 12 V.

    [​IMG]

    [​IMG]

    Colin
     
  9. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Mmh, Let's verify this again. Your bootstrap voltage does not look at all like the second waveform in figure 6 of the datasheet.

    When the gate pulse starts, there should only be a small dip.

    Maybe you are not measuring BHB to BHS / AHB to AHS?

    I meant measure with one channel pin 10 to 12 (AHB/AHS) and the other channel pin 11 to pin 12 (AHO/AHS).

    Then the same thing for B (pin 1 to pin 19 and pin 20 to pin 19).
     
  10. Deadstar1312

    Thread Starter New Member

    Apr 2, 2012
    16
    0
    You are right, those plots are between BHB/AHB and GND and BHS/AHS and GND. I used the math function on the scope to diff them but the plot didn't come out too well to capture. The diff is pretty much a straight line with some slight bumps on it. There isn't a profound dip when the gate turns on as in figure 6.

    Colin
     
  11. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    Can you confirm that all power supplies (both bootstrap and low side) are clean without oscillation?

    If you measure current (with a scope) while starting the actautor you will ahve an idea what current you are dealing with. I would then do the test with a resistive load that equals the actuator load when starting.
     
  12. Deadstar1312

    Thread Starter New Member

    Apr 2, 2012
    16
    0
    I do see oscillation in the upper and lower power supplies (XHS and XLS) which corresponds to the oscillation on the gates and oscillation of the power supply for the system.

    I will not be able to do any current measurements for a week or so but will come back to this.

    Do you have any thoughts as to what would be causing this?

    Thanks again,

    Colin
     
  13. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    No, I have no clear idea yet.

    Looking at your pictures again I wonder what these peaks are:
    [​IMG]

    They get bigger each pulse, until they go down to zero and the ringing occurs. Since these are gate signals and the peak obviously does not come from the driver itself (and therefore is not subjected to the deadtime generation of the 4081) my guess is that you have some sort of very short cross-conduction occuring, i.e. both transistors of one leg are ON at the same time for just a few tens of nanoseconds. Causing increased current flow through traces and parasitic capacitances, which leads to oscillation.

    Please measure both gate signals of one leg at the same time (both legs). Since you don't have an isolated scope just use as reference the source of the upper or lower FET. (I'd prefer the upper)

    Then measure the two lower gate signals and post the pictures.
     
  14. Deadstar1312

    Thread Starter New Member

    Apr 2, 2012
    16
    0
    I have a snap of ALO (red ) and AHO (yellow) wrt ground:

    [​IMG]

    When the high side is turning off and the low side turning on there seems to be a knee on the high side which is delaying turn off. I suppose this could be causing some shoot-through. This is shown in the snap. When the motor is not stationary and I have perfectly locked antiphase PWM the knee is present. When the motor moves the knee wobbles around, from its maximum shown in the pic to not existing.

    When the low side turns off and high side turns off there is no knee on either signal when the motor is stationary and you can clearly see the deadtime which should prevent the shoot-through. When the motor moves the knee becomes apparent and wobbles although it does not appear to overlap with the low FET turning on. This is shown below:

    [​IMG]


    Would you expect to see this knee on the high side gate? I presume this is a function of the bootstrap.

    As I am using a phase correct PWM system those spikes do correspond in time to the turning off and turning on of the other side of the bridge.
     
  15. praondevou

    AAC Fanatic!

    Jul 9, 2011
    2,936
    488
    You are measuring the high-side FET's gate to the low-side FET's source. That gives you an idea wether the timing is correct but not how the actual Vgs of the high-side FET really looks like. You will have to measure it separately with the reference being the upper FET's source terminal.

    Any way to trigger on the peak I indicated in the other post? Maybe triggering single shot on the first negative pulse width less than ... (whatever the peak is wide). If the scope has this function...
     
  16. Deadstar1312

    Thread Starter New Member

    Apr 2, 2012
    16
    0
    OK I understand. I am away from the rig for the next week but will investigate when I'm back.

    Thanks,

    Colin
     
  17. Deadstar1312

    Thread Starter New Member

    Apr 2, 2012
    16
    0
    I do not have an isolated scope and I only have two channels so I can only do a diff between the gate and source voltages for the upper and lower sides.

    Here is Alo (red) Als (yellow) and the gate-source voltage (green) for turn on and turn-off:

    [​IMG]

    [​IMG]

    Here is Aho (red) Ahs (yellow) and the gate-source voltage (green) for turn on and turn-off:

    [​IMG]

    [​IMG]

    Also I dont have a scope capable of triggering on the peak you are taking about. What I do know is this overlaps with the turning on or off of the other side of the bridge as I have the phase correct setup.

    Colin

    Colin
     
  18. Mark_T

    Member

    Feb 7, 2012
    47
    8
    Have you looked at the power supply rails, the linear actuator may be making them colapse, also C1 and C2 are missing, these will help as right next to bridge transistors and are necessary.
     
  19. Deadstar1312

    Thread Starter New Member

    Apr 2, 2012
    16
    0
    Thanks for the reply Mark.

    The ripple appears on the supply lines too. When you say the lines collapse do you mean the voltage drops with the initial current draw? I have a couple of large caps across the supply which seem to substantially reduce the voltage dip, I see the ringing after the dip.

    What is the role of C1 and C2?
     
  20. Mark_T

    Member

    Feb 7, 2012
    47
    8
    i suspect you need either large cap across the bridge, re c1 and c2 are not in circit. This cap should be large >10uF low ESR 2 x voltage on rail min. Or a bigger power supply, is it colapsing?
     
Loading...