This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DRV8305: DRV83053 fails quite often

Part Number: DRV8305
Other Parts Discussed in Thread: TPS54040, DRV8312

Hi,

Have been using the DRV83053 for a while; I've had successes with it, failures too. The board what I am using is my own, but is very similar to the BOOSTXL-DRV8305.

I've posted a picture of the board in this thread.

I am attaching the schematic here:

Previously, it was not quite evident what caused the failures, but recently I've been able to identify a bit more as to what's probably happening. Maybe the issue could be sorted out with some help and hence the post.

After fixing the issue as in my previous post. outlined here:

the board has been functioning and has been running quite well, even with motor being braked physically. It looked quite pretty.

Some days back

DRV8305 latest failure situation - Situation log
---------------------------------------
* Motor got stuck due to a mechanical fault, while running.

* SMPS tripped and shut down likely due to OCP

* Smell of cashews being roasted.

* Switched off Power

* Switched ON after a few minutes, SMPS powered ON. Motor does not run.

* No 3.3V LDO output on the DRV8305 - have a Green LED with 330 ohms connected to the LDO output. Likely draws about 10mA. The LED is not lit.


* Checking and looking deeper,
 - no visible burns/cracks:
 - Initial checks showed L1 (a 0.47Ohms resistor is connected in that position) went open

 - Looking deeper, C21, C22 (47nF) capacitors are shorted internally

 - Looking further, R5 100ohms resistor is open

 - C4 appears to be shorted internally.

 - MOSFET's appear okay, checking with a meter.

* Removed C21, C22 (47nF) capacitors
 - C21 is really shorted internally

 - C22 appears okay, checked with a LCR meter, shows 47nF

 - Checking impedance between DRV8305 Pins 43 & 42, it appears they are shorted internally.

 - Checking impedance between DRV8305 Pins 39 & 40, it appears they are shorted internally.

 - Removed C4 (2.2uF) capacitor, it appears okay, LCR meter shows 2.2uF

 - Checking impedance between DRV8305 Pins 38 & 41, it appears they are shorted internally.


So, it appears that the DRV8305 is dead too, without diving any further deep.

What could cause the charge pump capacitors to fail thus. This has been the standard way the failures has been happening each and every time. Once the bootstrap capacitor fails, likely the DRV8305 also fails. Thus the resistor too. This one problem is holding me back from proceeding on this project, for a while. Sigh!

I wonder what could be causing this issue ? Anything that could be changed in the design to avoid this issue over and over again ? (I have tried looking into each and everything that I could read about it, but cant seem to find what I am missing)

HELP! HELP! Can someone please help ?

Thanks,

Manu

  • Hi Manu, thank you for the detailed post. We will try and figure this one out. The smell of cashews being roasted when describing TI parts is obviously disturbing.

    I have assigned the lead apps engineer on this part to help you but let me start getting some additional information from you :

    - You mention that L1 in your schematic is replaced with a 0.47Ohm resistor which was damaged. Is that correct ?

    - Is CON1 connected to the SMPS or is VDD connected to the SMPS ?

    - R5 being damaged is a usually a sign that there was a high negative voltage seen at that node. Can you check if the VDRAIN pin has any damage ( under a magnification) or you can apply a small voltage to it and see if the current clamps ?

    - When you brake the motor suddenly , obviously there will be large transient voltages being applied to the motor pins. Are you able to measure these in a controlled fashion ?

    Regards

    Anuj

  • Hi Arjun,

    Thanks for the help.

    Actually, I am quite happy when the DRV8305 works, most of the times.

    It is those times after it blows up, replacing the DRV8305 with the thermal pad is a bit of a hassle. Also, I must mention that, I have a heatsink also connected to the thermal pad of the DRV8305 to keep the DRV8305 cool. In the previous cases, with no heatsink, nFAULT would be flagged, a RED LED connected to the nFAULT pin would be lit before things go out of control. In this case with the heatsink, I have not noticed the RED LED going ON, maybe it did for an ultra short period before it died probably.

    - L1 was replaced with a 0.47 Ohm resistor. Right. This Resistor also blew up and became open.

    - CON1 is VDD, CON5 is GND. CON1 is connected to a 2200uF/25V cap in parallel to a 100n/100V MKT, fed through an Common Mode Ferrite choke, the other end also having a 2200uF/25V cap and a 100n/100V MKT. Much like a PI filter, except that it is a common mode choke. VDD=21.6Vdc powered from the SMPS. On the SMPS output have a Pi filter with 4700uF, 2uH choke and 4700uF electrolytics. Just mentioned for the sake of completion.

    - VDD on the board is connected on the board in schematic to a TPS54040, for powering the microcontroller.

    - With R5 removed from the board, checking continuity between the R5 pad and GND, the DRV8305 appears to have an internal short between the VDRAIN pin and GND. If I replace R5 and apply power, R5 will blow again. (In two different failed cases of the same situation; Applying power caused extreme heating with the DRV8305. After a sufficiently large heatsink was deployed, replacing R5 in the same situation and then powering up, caused an explosion and a crater on top of the DRV8305. In another same failed case, the track to the MOSFET gate burned up.)

    - Yes, I agree that there will be a transient, when the motor is abruptly braked. But usually, I haven't been able to measure a significant spike. But when it blew up the DRV8305, I wasn't able to measure it. I must mention an important aspect here. I have braked the motor many times. There were no issues, or failures or not even SMPS OCP shutdown when I intentionally braked the motor completely. It is always those unintentional/accidental braking due to a mechanical failure in the mechanical assembly the failure occurs. Maybe with the accidental brake situation, the spike is probably more significant/and or sharper. I cant think a way how I can really see the spike/surge in that situation though; I mean to measure it.

    Have been consuming the DRV8305 like cookies to identify what's going on.

    Thanks,

    Manu

  • Sorry, typo. I meant to type Anuj. :-( Was too much focused at the DRV8305.

    Best Regards,

    Manu

  • Hi Manu, thanks for the information. My first guess would be that the transients being generated during the mechanical failure are either directly violating the DRV8305 specifications or there is a cascading effect causing multiple failures.

    - VDRAIN showing a short to ground is certainly damage from a transient event. There is a -20mA limit on this pin.  

    Can you try a couple of experiments to help isolate the issue in case a measurement is not possible :

    - Place a diode between VDRAIN and VM ( this will absorb any positive spikes on the phase and ensure that Phase to VM is limited to 2 diodes ( 1 body FET + 1 diode). This will ensure that no internal high current paths are opened from SHx to VM

     - Place a diode on VDRAIN (anode on the IC pin) to protect it from a negative transient. This will mess up your VDS monitoring though.

  • Hi Anuj,

    Sure, I can try out some stuff. Would be extremely happy to sort out this problem.

    Was quite lucky, that I have one more spare DRV8305 at hand.

    Diode between VDRAIN and VDD. The easiest way would be to replace R5 (100 ohms with this diode, is that acceptable ? Otherwise, I need to cut the track)

    I have 3 diodes at hand.

    #1. RF081MM2STR

    #2. PMEG6010

    #3. 1N5819

    Which one of these would you prefer to be used between VDRAIN and VM ?

    The diode anode on VDRAIN, I am slightly confused. Diode between VDRAIN and VDD, Anode at VDRAIN and Cathode on VDD, was what you meant ?

    (If so, diode across R5 would be easy)

    Is this what you meant ? The modification in RED

    The 2nd diode that you mention for protection against negative transients, would it be better across the MOTOR, U,V & W windings to GND, 3 diodes in all ? Similar to a diode across a solenoid, but 1 each on each of the windings, which makes 3 of them ?

    In this case, will a TVS diode be better suited ? Can you make a recommendation on the TVS diode, if so ?

    The DRV8312 Eval board has a LC filter at the output 4.7uH and 1nF at it's driver output, just before the motor.

    Something like that would be good to try ? (Or would that mess up the Curent Limiting/protection stuff ? Or even worser, the Instaspin FOC/Motion sensing ?)

    Thanks,

    Manu

  • Hi Anuj,

    Replaced DRV83053, C21,

    Replaced L1 with a RF081MM2STR to protect the Charge pump from the negative undershoots you mentioned

    Replaced R5 (100 Ohms) with a RF081MM2 Diode and a 100 Ohm Resistor in Series.

    In addition (with the thought of reducing noise), at the Motor terminals CON2, 3, 4, Pulled up to VDD with a 100n capacitor and pulled down to GND with another 100n capacitor on each motor winding. 6 capacitors in all. Is this a good thing to do ?

    Currently, the motor runs as expected, the 6 100n capacitors do get a bit warm. The MOSFET's too get a bit warm in this situation.

    I need to wait till the next issue occurs, probably ? Or do you have any other suggestion  ?

    Thanks,

    Manu

  • Hi Manu, 

    Thanks.  Some comments below on the two threads . I ama adding some tags belew  so we can keep track 

    (8/7/19_1) Replaced DRV83053, C21 : Yes, this is a good idea, especially if there was damage the last time

    (8/7/19_2) Replaced L1 with a RF081MM2STR to protect the Charge pump from the negative undershoots you mentioned : Replaced L1 with a  diode from VDRAIN to PVDD 

    (8/7/19_3)Replaced R5 (100 Ohms) with a RF081MM2 Diode and a 100 Ohm Resistor in Series.: I think I may have made a mistake here . You want to simply protect VDRAIN from negative transients, so cathode of the diode to the IC pin . You will need to adjust your VDS monitoring threshold accordingly since you will always have the diode drop in the high side VDS measurement path. 

    In addition (with the thought of reducing noise), at the Motor terminals CON2, 3, 4, Pulled up to VDD with a 100n capacitor and pulled down to GND with another 100n capacitor on each motor winding. 6 capacitors in all. Is this a good thing to do ?

    (8/7/19_4) I am not sure this would really help.

    To reduce switching transients, you could try a schottky diode in paralell to the low side FET body diodes to reduce negative spikes.

    (8/7/19_5)Another thing to try is to adjust your IDRIVE settings so that your turn off time is much slower than what you currently have. The though here is that if the mechanical failure is causing the driver to turn the FET's off too hard and at high currents, the transient spikes would be reduced by a "softer" turn off .

    Currently, the motor runs as expected, the 6 100n capacitors do get a bit warm. The MOSFET's too get a bit warm in this situation.

    I need to wait till the next issue occurs, probably ? Or do you have any other suggestion  ?

    (8/7/19_6) Not the best strategy to wait for failure. One thing you could try is try to simulate a failure ( high current, sudden shutdown) and see if the above improvements help. This would indicate a step in the positive direction. 

  • Hi Anuj,

    (8/7/19_1) Ok

    (8/7/19_2) Ok

    (8/7/19_3) Ok, but the VDS drop shouldn't be a big issue for the time being, temporarily, I guess ?

    (8/7/19_4) The thought was something like this. (Remembering the old times, where capacitors put  across motors on old audio tape players, to reduce the motor hum)

    Hence the capacitors. But the capacitors in this scenario are getting sufficiently warm. Capacitors getting warm, is a sign that they are not going to last very long; So, I guess that strategy was short lived. A schottky diode as you suggested sounds better. Any specific suggestions you have (Voltage, Current etc) ?

    (8/7/19_6) Today, I tried running the motor at max speed and suddenly clamping the motor spindle with pliers to simulate the hard failure. This resulted in immediate SMPS shutdown due to OCP, I guess. No odour from the board. The capacitors across the motor and the MOSFET's were quite HOT; So was the DRV8305 with more than adequate heatsink provided. Switching Power OFF and ON after a minute, the motor sprang to life. The process was repeatable twice, by that time the MOSFETs and the DRV8305 was very hot and had to stop further experimenting. (I wonder, why the DRV8305 gets so hot. During normal run, the DRV8305 does not get hot, but with the transients coming in, I guess it gets hot.)

    I will try out your new suggestions, do some more tests and see how it goes. I will wait for your feedback on the TVS diode before trying out anything new.

    Thanks,

    Manu

  • Thanks Manu.

    (8/7/19_3) Ok, but the VDS drop shouldn't be a big issue for the time being, temporarily, I guess ?

                     Yes, should not be an issue if you have the trip level set high enough. Eventually you can compensate for the extra diode drop.

    (8/7/19_4) The thought was something like this. (Remembering the old times, where capacitors put  across motors on old audio tape players, to reduce the motor hum)You are probably passing the ripple through the capacitor, I do not feel these capacitors are required for the problem you are seeing. 

    Schottkysc hoices are here  : http://www.vishay.com/diodes/rectifiers/schottky/forward-current-gteq-20-a-lteq-30-a/

    You can choose the reverse voltage levels needed based on your system. This will help reduce some of the negative transints on the phase node. 

    (8/7/19_6) No cashew smell. I like that. One reason that the DRV8305 could be getting hot is that if you have a large negative transient on the SHx pins, there is a possibility that the high VGHx-SHx voltage causes the internal clamp on the DRV8305 to turn on and absorb all of the gate current. The clamps should however, disengage after the transients have subsided and the chip should cool down again. Can you verify this. 

    If this is the case, an external zener from VGHx-SHx can also be added ( this is quite common) . An example is BZX84J. 

  • Hi Anuj,

    (8/7/19_4) Will try your suggestions. A question that I have here, Any advantage in using a Schottky diode, over a TVS diode ?

    (I will wait for your comments whether a TVS or Schottky is better suited, before moving ahead)

    (8/7/19_6) Once the components are cool (starting afresh it does not get hot, while running again), it gets hot when the motor is loaded (The heat builds up, depending on the load). At a certain point, instantaneous braking causes SMPS (TIDA-00367) - OCP shutdown. By that time, DRV8305 with heatsink, MOSFET's, are really HOT, capacitors are HOT.

    Two questions I have in this regard:

    #1. In the given situation my SMPS can provide maximum 5A, so a 5A diode whether Schottky or if TVS would be sufficient ?

    #2.It would be good to have a TVS/schottky from the DRV8305 Power to GND (ie VDD to GND) ?

    I will try the MOSFET gate protection Zener diode you suggested too, (15V/500mW is okay ? )

    Thanks,

    Manu

  • Hi Manu,

    (8/7/19_4) The purpose of this schottky is  to take advantage of its low Qrr and reverse recovery time. This will help reduce the negative spikes. A TVS is better suited as a clamp.

    5A rating would be fine

    (8/7/19_6) I need a little clarification here. When you are running without a "nose plier" event and you progressively increase the load , does the DRV8305 get hot or is it only when you try to mess with it ?

    (8/10/19_7) I don't think a VDD to GND diode would help

    Yes, 15V/ 500mW is sufficient for the gate protection zener.

    Regards

    Anuj  

  • Hi Anuj.

    (8/7/19_6) The DRV8305 & the FET's were previously "getting hotter as the motor was revved up". It got "even more hot when I tried to mess with the spindle" with the braking test. I thought the FET's and the DRV8305 might blow up, being that hot. But it didn't.

    Attaching the waveform (across the motor U,V, W one phase) with the negative spikes, this waveform is with the capacitors on the board. (during Normal run. You can notice that there is noise in the form of ringing, as well as a negative undershoot)

    After the previous mail, I removed all the 6 100n capacitors from motor U, V, W to VDD and GND

    I did not have the Schottky diodes at hand and hence yet to order the same. It would take a few days for delivery. I have a few B140's with me.

    Was a bit impatient to try the B140 for the same use, from motor U, V, W to GND.

    Tried that, the results are:

     - The MOSFET's do not heat that that much, they become warm now.

    - The DRV8305 does not heat up, it is warm now, but it is with that huge heatsink (I wonder, whether the DRV8305 was really intended with a large heatsink. The 3 ounce copper PCB with thermal vias- via stitching does a good job, in other previous use cases.)

    - Retried the "hard brake" test. The SMPS does OCP shutdown. Power OFF & ON, the cycle was repeated a few times. Things do look okay. The DRV8305 is warm. MOSFET's slightly warm. Things look under control.

    (Haven't yet tried the Zener at the MOSFET gate yet. Do you still want me to try it ? Haven't got the Zeners at hand, yet)

    Attaching the waveforms across the motor U, V, W.

    The negative undershoots and the ringing seems to have reduced with the introduction of the Schottky diodes. Heat dissipation in all components seems to have drastically reduced. It looks god, I must say. The fix looks good. Schottky on U, V, W. Diode in series with VDRAIN resistor, Diode instead of the Inductor.

    What do you think about a small output L-C filter at the FET outputs (ahead of the newly fixed Schottky) ?

    Thanks,

    Manu

  • Hi Manu,

    (8/7/19_6) Yes, 5A Shottky is OK for Phase-GND

    VDD to GND will not help

    For the VGS protection, yes this will only handle peak currents so 15V/500mW is perfect

     

  • Hi Manu, 

    I wanted to check in on how your testing is coming along ? 

    Regards

    Anuj 

  • Hi Anuj,

    Got the components only today. It looks great as of now. DRV8305 seems quite okay. No issues it appears to be. Will fiddle with it over the week and see how it goes.

    Will update this thread in a few days time.

    Thanks a lot. Much appreciated.

    Best Regards,

    Manu

  • Manu,

    I hope all is well. Please let us know when you have an update.

    Regards,

    -Adam

  • HI Anuj,

    Did some fair amount of testing. Things do look well. The problem appears fixed.

    The spikes were indeed causing the damage. The DRV8305 internal clamps alone are

    not sufficient to handle the spikes. That was indeed the issue at hand.

    Using external schottky diodes, that problem is fixed.

    Thanks a lot, for the debugging help! Much appreciated.

    Best Regards,

    Manu

  • Hi Adam,

    Done. The problem appears to be fixed.

    Thanks,

    Manu