• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 63 subscribers
  • Views 208 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

LMR12020: Persuading it to start up

Patrick Herborn
Intellectual 520 points
Part Number: LMR12020
Other Parts Discussed in Thread: TPS62173, , TPS629210, TPS62901

A design that previously used TPS62173 was updated to use LMR12020 due to supply issues with the original part.

A design feature of the board is that it can be powered either by its own switcher, or via VBUS when connected via USB. This worked fine with the original switcher, but with the LMR12020 the switcher does not want to come online if VBUS is already present. It should be noted that VBUS feeds VCC5 via a diode, so there is "only" 4.6-ish volts on VCC5 when running from VBUS, which is less than the 5V regulation target - in other words the switcher should come online and deliver the requisite current to raise that rail to 5V.

It is my suspicion that this is because it cannot charge the bootstrap capacitor (something that the TPS62173 was devoid of). There is only 0.4V between the ends of the bootstrap capacitor, which won't be enough to turn on the NMOS device - this in turn means that no current can enter the inductor, which means it cannot fly back, which deprives the bootstrap capacitor of charging up to the desired 5V.

I am thinking that a possible way round this would be to use a resistor (plus maybe a diode) from xVIN to the BOOST pin with a Zener across the bootstrap cap to make sure it cannot charge to more than 6V as per datasheet. This would be similar in principle to the datasheet suggestion of a diode feed from a VCC5 in the event that VIN is too low. Hopefully that would then be sufficient to kick-start the bootstrap cap, and of course as soon as there is flyback, the cap can then properly charge.

Question is - am I barking up the right tree here, or is there some other protection or shutdown mechanism which could prevent the LMR12020 from firing up in the event that its output is already near (but not yet at) its desired regulation voltage ? Or could it be some lock-out based on SW pin not being at or near GND to start with ? Or something else entirely ?

Many thanks,


Pat.

  • 0
    TI__Mastermind 47020 points

    Hi Pat,

    I am not quite sure about your question, do you mean your application could not start up to target output voltage 5V(only stop at 4.6V)? or your application could not start up with 4.6V output pre-bias?. Is the boot voltage is 0.4V? What is input voltage? Could you share the schematic?

    Thanks,

    Nancy

  • 0
    Intellectual 520 points

    Hiya Nancy!

    Many thanks for responding to my question. In response to your questions :

    My LMR12020 will not start up if there is a 4.6V output pre-bias. If there is no pre-bias then the LMR12020 will happily start and it will output 5V, no problem.

    Re-reading the datasheet pointed out an error in my original post. I thought the internal LDO was a 5V unit, but it is 3.9V. The 0.4V was based off the difference between 5V LDO on the BOOST pin and 4.6V pre-bias on the SW pin. But if the LDO is only 3.9V then the 4.6V pre-bias means that the boot is -0.7V, not 0.4V as I suggested previously. Apologies for this error. It is likely impossible to turn on the NMOS transistor with this boostrap voltage.

    The input voltage, when present, is 12V.

    With regard to the schematic I don't believe there is anything interesting there - it will be the same as the datasheet or WEBENCH with the notable addition of a diode from USB VBUS to the output (5V VBUS minus a diode drop gives 4.6V). I can likely export it tomorrow if this description is insufficiently detailed.

    Another quick look at the LMR12020 equivalent circuit shows the BOOST pin LDO has a diode, so it looks like it should be safe to raise the BOOST pin's voltage up to (say) 8.6V by means of a resistor (plus maybe a diode) from VIN... but of course if the problem is something other than the lack of adequate voltage across the bootstrap capacitor, then this won't help.

    Many thank again,

    Pat.

  • 0 in reply to Patrick Herborn
    TI__Mastermind 47020 points

    Hi Pat,

    Understand your application and no need to export schematic. Yes, the bootstrap capacitor can not be charged with 4.6V pre-bias output voltage. You could not use 12Vin to charge the bootstrap capacitor either since it violates the absolute max rating of Boost to SW voltage. It is -0.5V to 6V in the datasheet. SW voltage is low during high side MOSFET off time. Because you are still in design phase, are you open to use other devices? We have TPS62901/2/3 family. We also have TPS629210 family which is latest generation buck converter with very good performance (pre-bias startup is ok). The parts and EVM are available now. Please have a look.

    Hope this helps,

    Thanks,

    Nancy

  • 0 in reply to Nancy Zhang
    Intellectual 520 points

    Hiya Nancy!

    Many thanks again for coming back to me on this, and confirming the suspicion that it is indeed the bootstrap that is causing the issue.

    I refer you back to my original post with regard to the use of the 12V to pre-charge the bootstrap capacitor. Notably I stipulated a Zener in parallel to the capacitor in order to remain within the 6V max BOOST-to-SW voltage limit. I only need to charge the capacitor sufficiently in order to barely turn on the NMOS - that will in turn cause current to build up in the inductor and that will cause flyback which can then further charge the capacitor. I should be able to get away with a reasonably high value resistor to accomplish this goal - which will in turn reduce the work that the zener has to do during the flyback phase. This will undoubtedly reduce the efficiency, but that is not super critical in this application.

    Whilst changing to a different switcher long term is a viable option, in the short term I have to make this work. There is already one production batch which I need to devise a simple yet effective mod fix for, then there can be a new board spin to use up the remaining LMR12020 stock (which we have in quantity, as necessitated by the current state of the supply chain) - and then finally we could do another board spin to accommodate a different switcher... by which time there should be decent availability. I think we may have looked at some of the suggested alternatives, but at the time there was a Max Order Quantity of 5 pieces, which was no good to us... we needed quantity and we needed it fast.

    Many thanks again,

    Pat.

  • 0 in reply to Patrick Herborn
    TI__Mastermind 47020 points

    Hi Pat,

    Understand. Thanks for sharing. I will close this post.

    Thanks,

    Nancy