• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 2 answers
  • Subscribers 65 subscribers
  • Views 5273 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

High Efficiency Buck Converter Design Guidance

Chris Minerva1
Prodigy 100 points
Other Parts Discussed in Thread: LM25116, LM25149, LM25148-Q1, LM25149-Q1EVM-2100, LM5085

Hello,

I am seeking some guidance in regard to a custom buck controller design I am working on. My design goal is >=97% efficiency and so in order to accomplish this I am going with a discrete implementation that uses a synchronous buck controller with gate driver and external MOSFETs. I am specifically having trouble determining if there is any additional value in separating the PWM controller and gate drivers into separate chips as well. I'm not convinced doing so will further contribute to increased efficiency and this is the first time I'm designing a supply where I'm breaking things down into the individual components like this. It seems like I'd only need to split the PWM controller from the gate driver in the case where I can not get high enough gate drive current from an integrated solution. This is a non-isolated DC/DC converter, so I'm also trying to keep the implementation as simple as possible. The requirements for my design are as follows:

Non-isolated DC/DC buck converter

Vin(nom)= 24V (so some safety margin above that, say 27V is warranted)

Vout = variable 12V to 24V. I plan on using an MCU to PWM a bias on the FB pin to set the output voltage dynamically.

For the Vout = 24V case I need 100% duty support for a high-side N-FET. If this is prohibitively difficult to implement I can probably trim adjust the input rail to say 24.5V instead, or whatever the minimum dropout is for the selected part.

Iout = 1.5A to 5A. The load is always resistive and fairly constant, so the current will change only when the MCU changes the output voltage.

I have room to grow the inductor and capacitor sizes considerable and I plan on doing this in conjunction with a lower switching frequency (up to several 100kHz) to maximize the efficiency.

If you would provide me with some general guidance on this and perhaps a few part recommendations to get started I would greatly appreciate it. Right now I'm feeling a bit overwhelmed with too many options/choices.

I am also not sure about how fast I need to transition (i.e. ton, toff) the MOSFETs. My understanding is that parasitics in the drive circuit will limit the max speed I can switch the gate without ringing and that I will need to provision gate drive resistors (and parallel R-D) to tune the transition times accordingly on the bench. Therefore, it seems necessary to over-specify the minimum gate drive current support needed and then reel it in with the resistors rather than risk underspecifying the gate driver current required. Even with knowing MOSFET Qg and driver Igate, I still don't have a good feel for how to determine what my initial target should be for MOSFET ton, toff. I'm just not sure how to estimate that.

Thank you!

Chris

  • 0
    TI__Genius 13451 points

    Hi Chris!

    You mentioned you want a 97% efficiency. Do you mean full load eff or peak eff?

    Shuai

  • 0 in reply to Shuai Fan
    Prodigy 100 points

    Hi Shuai,

    For this application, there will be a load current operating point somewhere in the range of 1.5A to 4A. This operating point will correspond to where the load operates most efficiently and I will not know what this operating point is until I perform system-level characterization on the actual hardware. So, for the time being, I really just need to maximize the efficiency across the full load range as much as possible.

    Perhaps I need to really get into some of the nuts and bolts of this in order to best communicate the challenge on my hands here. I am redesigning an existing system that uses an OTS thermoelectric controller module currently capable of providing up to 4A current at 21V. This existing circuit is not always capable of providing the necessary load current, and so the design also implements a solid-state relay (SSR) switch in parallel with the controller. When the controller is incapable of providing the adequate current, the MCU switches the controller out of the load circuit and switches the SSR in circuit, the later of which just directly passes the 24V rail to the load. My goal is to replace both the controller and the SSR switch with a single buck supply capable of 100% duty cycle so that it can emulate the SSR direct connection.

    The SSR path has an efficiency of about 99%, and so the most important goal is to meet or exceed the SSR performance for the 24V in to 24V out case. I have found a MOSFET that is capable of exceeding the performance of the existing SSR in saturation (SIR184DP-T1-RE3) provided I can get the buck controller with an auxiliary bootstrap/charge pump supply input to supply 100% duty cycle (i.e. LS FET always off). In absence of a part with this capability, I do have some leeway in terms of making a hot adjust to the trim pot on the 24V input rail coming from the AC/DC brick to cover the minimum dropout to guarantee 24V output.

    The efficiency requirement to meet or exceed the performance of the existing OTS thermoelectric cooler at Vout <=21V and Iout <=4A is about 96%. It is important that I am able to exceed this a bit (hence my 97% goal) because we do not have any good data on the thermal performance margin of the existing system. I am in a difficult design position because just a couple degrees C of addition heat rise above the existing design performance could cause thermal runaway. I'm therefore in a position in which I need to just set the efficiency bar as high as possible.

    The final complexity, to which I alluded to above, is that I'm not convinced the current design is operating the load at the most efficiency operating point, so I can't definitively say, for example, that I need peak efficiency at 3.6A load current of the current design. I will not know the operating point until we perform thermal characterization of the entire closed system using the new power supply.

    I hope this wasn't too confusing. Thank you for listening and if you have any follow up questions please let me know!

    Best,

    Chris

  • 0 in reply to Chris Minerva1
    TI__Genius 13451 points

    Hi 

    We don't have a device can satisfy your requirement.

    But we can assign you question to the WV team.

    BSR-MV

    Shuai

  • 0 in reply to Shuai Fan
    Prodigy 100 points

    Hi Shuai,

    What requirements can you specifically not meet? Is it just the efficiency? What can you offer that is the closest you can get? I am looking at other suppliers as well but I am clearly going to need to make some compromises if my options are zero. The one part I've found that seems to check most of the boxes is the LM25116.

    Thanks,

    Chris

  • 0 in reply to Chris Minerva1
    TI__Mastermind 22301 points

    You can try using LM25149 or LM25148-Q1, these device can fold-back frequency during dropout conditions and achieve close to 100% duty cycle.

    Also LM25149 and 8 have lower Input current IQ compared to LM25116, which will result in higher efficiency.

    You can create WEBENCH designs to estimate efficiency.

    In dropout the efficiency will be determined by the resistive losses, so minimize RDSON of high-side FET and DCR of inductor.

    As you go down in frequency the inductance goes up, so the DCR will also go up. I'd say stick with 300-400kHz for a good compromise.

    I recommend ordering a LM25149-Q1EVM-2100 and modifying it to meet your design requirements for the best measure of efficiency.

    -Orlando

  • +1 in reply to Orlando Murray
    Prodigy 100 points

    Hello, Orlando.

    Thank you for your suggestion. However, there is no stock on the LM25116 and the lead time is over one year. I have recently found a suitable and available part from ADI (LTC7801) that satisfies all of my requirements along with inductor and MOSFETs to support 150kHz.

    Thanks,

    Chris

  • 0 in reply to Chris Minerva1
    TI__Mastermind 22301 points

    Hi Chris,

    Glad you found an alternative.

    Also consider LM5085 which is able to achieve 100% duty cycle.

    Hope this helps,

    -Orlando