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.

LM5176: Bootstrap issue, MOSFET selection.

Part Number: LM5176

I made some strange experience with a new design using LM5176PWPR but today I see a behavior I can't explain myself. First of all, I control the output conditions of the DCDC converter with an external (low speed) regulation loop at the SS pin. As an example, it can provide a changeable CC/CV characteristics controlled by a microcontroller. It’s a 55V/500W output design. Input is 48V.

The system is working in buck mode. The HS MOSFET of the boost side is constantly active. In the past I noticed, that the bootstrap circuit switched the boost half bridge one time at a gate voltage of about 3.5V to bootstrap the drive circuit. This behavior us not working today. I see no switching, BUT the gate voltage is constantly on at about 7V. Yesterday I notice the same behavior on the buck MOSFET. Is there an internal circuit that provides power for the bootstrap path? I don’t think so, but the voltage is also available and without ripple using an additional “bleeding resistor” at the gate?

The next thing I can’t explain is, that if the same system switches to boost more (constantly voltage instead of constantly current), the gate of the HS MOSFET stops switching. There is no gate-source voltage and the MOSFET works as a simple Diode. Is there a reason the HS-MOSFET is not switching? Are there safety features I don’t know?

Last but not least, I have some troubles finding the right MOSFETs. I need 80V types to have a good margin but most of them have a threshold voltage higher than the minimum high-side gate voltage of 3.5V. I found one with a minimum of 1.1V but with this MOSFET I have to add a bjt gate discharge circuit. Without it, I see a high shoot-trough current (hiccup triggers as well when active). The 80V TI MOSFETs have a threshold that is critical with the actual bootstrap circuit (3.5V min.)

 

Regards,

  • Hi Horst,

    First, thank you for designing with the LM5176.  What you observed is normal performance.  The IC itself also has an internal circuit to keep the BOOT voltage of the unswitched leg.  It is close to the switched one, namely ~7.5V.  So. your FET with 3.5V FET should work.   

    Best Regards,

    Youhao Xi, Applications Engineering

  • Dear Youhao,

    thank you for your response. 

    I have seen a system where the Low-Side MOSFET of permanent on side (buck side in boost mode) is recently switching to charge the bootstrap path. Noe I have notived, that this behavior only happens when I disassemble the bootstrap path of the opposite side. Looks like there is a safety-feature that switches the bootstrap path if there is a problem with a too low voltage (UVLO 3.4V).

    If both bootstrap path are assembled (we initially safe that because we use a high side Schottky instead of the MOSFET related to current in reverse direction), what will be the minimum voltage available for gate drive of the MOSFETs. Are the 7.5V available and MOSFETs with a higher threshold (~3.5V) can be used? Is there any situation, the voltage fall to about 3.5V again?

    The last thing I don’t understand is, that my high-side boost MOSFET (now switched from Schottky back to MOSFET after placing LM5050 at the DC output to prevent reverse current from battery) is currently not switching. The output is boosting as expected but the MOSFET works only as diode. If I switch back to buck mode, the gate voltage for the high-side boost MOSFET is available again. Maybe this can be explained with the mentioned internal circuit, that use one charge pump to generate also power for the other one…. With other words, maybe the boost charge pump has a problem.

    Thanks for your first response. I have to do some more research to understand why the boost bootstrap is not working and why there is so safety feature triggering a charge cycle, as the other side do.

    Regards,

    Dominik

  • Hi Dominik,

    There is charge pump inside for the boost strap.  There is just charge sharing circuit. Note that the switching leg's BOOT pin swings up to be higher than the BOOT of the non-switching leg, so it can keep the non-switched BOOT pin to be charged. If you use a diode, we still recommend to place the boot strap circuit to keep the charge sharing circuit operating.

    Note that the LM5176 does not support the DCM mode, so the MODE pin should have 93.5K resistor to ground, or tying to VCC.  Please make sure your MODE pin is configured such way.  Then, you can see the boost high side FET will not only use the body diode to conduct. It will be turned on.

    Thanks,

    Youhao Xi, Applications Engineering

  • Hi,

     

    thank you. I understand the MODE-pin to select Hiccup but yes, I used a 91k resistor. 

    Unfortunately the high-side MOSFET of the boost side is still not working, with 91k or 200k at MODE. I see currently that the drive voltage is available (Measured at the pin), the bootstrap circuit is working but there is no signal to switch the boost high-side MOSFET in boost mode. In buck mode, the signal is available, so all connections and the MOSFET should work. Is there another input necessary.

    Can you please confirm, that the voltage of both high-side MOSFET driver will stay at about 7.5V. In the first system I used a MOSFET with a high threshold and if the driver is working as described above, the driving voltage falls to about 3.5V what may be not enough.

    Thank you.

  • Hi,

    Same situation with the second board. Please find a screenshot attached. Blue is the boost switching node and green the gate. It's startup, the SS pin is controlled with an external DAC and rising slowly. As the system enters Buck/Boost Mode, (input~output) both MOSFETs are switching but in "only boost mode", the high side MOSFET didn't switch. Any idea? What will stop the high-side MOSFET in boost mode? I can’t find this information (as well as the information related the internal charge pump) in the current datasheet.

  • Why do you use an external circuit to control the SS?  Can you just leave the SS pin with the CSS alone and see if the boost high side FET can switching?  I am afraid your control of SS makes the circuit think the soft start is not completed, so the boost high side FET is kept off to prevent back flow current, which is to cope with the case of starting up with output pre- bias condition. 

    Let me know the results.

    Thanks,

    Youhao

  • I use an external circuit at SS pin to control the charge current to the battery. That's a recommendation (or idea) from TIs side. 

    ...but you are right, without the circuit, the MOSFET is switching as expected. Thank you, great idea and support! 

    What can I do? Is there a voltage level at SS pin softstart is disabled? Do you see another/better option to control the charge current?

  • Thank you Horst for confirming the SS issue.  The idea of using SS directly to control the SS is not a perfect one, if it came from my colleague.  It might be a misunderstanding about the LM5176's average current limit, which actually pulls down SS.  However, it is different than your externally pull-down because the IC cannot distinguish it from the SS incomplete state.  When the internal pull-down of SS occurs, the IC knows it is in current limit mode.

    It seems you don't have a fixed dc current limit but a dynamically adjustable limit.  This will need some extra circuit.  One idea is as shown in the attached drawing. 

    CC-CV Control.pdf

    Thanks,

    Youhao

  • Hi,

    thank you for your circuit. Mmh, this is an interesting because I started with exactly this circuit ar revision 1 but had a lot of troubles with stability. This issues where not there when using the SS pin in revision 2. 

    Regards,

    Horst

  • You need to make the op-amp for CC control a relatively slow circuit to avoid stability issue.  You can simply set the R=0 and just put a C=100nF and see if it works better. 

    If your revision 2 is the one that you pull down the SS externally, will cause the HDRV2 to stay low vs SW2, so it is not  solution anyway. 

    Thanks,

    Youhao Xi

  • Hi Youhao,

     

    thank you. Unfortunately I don't understand your last sentence. Rev 1 is working at comp and rev 2 at SS. The current patch is working again at comp.

    Currently I have a version more or less stable if working but unfortunately it didn't start if the load is initially connected.

    I measure that SS pin (only 4u7 connected now) didn’t rise and the output voltage stays very small. If I disconnect the load resistor, the voltage at the output rise slowly to about 2V and then the system is starting as expected. Do you have an idea with this issue as well? I have to connect a 2k resistor at VOsns, maybe this can be an issue?

    If I short the current sense pins, the situation is the same. I see sometime gate signals in the buck side so I expect that there is another safety circuit triggering.

  • ...are 4u7 too much at SS?

  • ...with 100n it's starting. Is there a maximum. Again I can't find any information in the datasheet. 

    Some of my concerns are, that there are currently a lot of information I can't find in the documentation. With other words, if you make changes at this circuit, I will not notice them and this can be an issue for mass production. 

    Regards,

    Dominik

  • Hi Youhao

    I'm stucking in the final design phase because there are too much issues with the IC working synchronously as battery charger. I can prevent the reverse direction only with a diode instead of the boost MOSFET. The ideal MOSFET in the output path is not really an option, because in fault condition (short) the peak current provided by the output capacitors will be ways to high for this MOSFET.

    Today I made a quick check with the UCC24612-2 driving the boost MOSFET, but the result was not as expected.

    Maybe there are other ICs that can handle the reverse direction but for me it looks like LM5176 is not the best choice for a battery charger. I will check some other options and in worst case, I come back to LM5176 with a diode.

    Regards,

    Horst

  • Hi Horst,

    I just came back from a business trip.  Regarding the SS capacitor, 4.7uF sounds too much. The SS current is only 5uA and your capacitor may leak as much to prevent starting up.  Regarding the boot strap circuit, the LM5176 has a TI proprietary circuit to keep the capacitor charged even if the leg is not switching.  So what you observed is normal.  The LM5176 has been used in battery charger circuit by other customers, so it should work in your design.

    Thanks,

    Youhao

  • Hi Horst,

    Hope the issue has been resolved, given no further update from your side.  You are welcome to re-open this thread by adding a new post.

    Thanks,

    Youhao