• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 63 subscribers
  • Views 1567 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.

20W Boost 3V->20V LM3478 Webench design

HUGUES MARCEAU
Prodigy 170 points
Other Parts Discussed in Thread: LM3478, CSD16323Q3

Hi all,

I'm designing a boost DCDC with a variable output, where efficiency (and BOM cost, and size ... ) is a concern.
Vinmin = 3V
Vinmax = 4V
Vout = 5-20V
Pout = 20W

I think the LM3478 is the most well suited IC for the application, but I'm struggling at different areas. At first, I guessed 350 kHz would be a good compromise. Here's the proposed schematic.

From the proposed WeBench solution, I noticed the transistor was the part that was toasting the most, reaching 75C.

From my limited knowledge, steady state simulation indicates that most of the transistor losses occur at switching.

What can I do to reduce them? Here's a list of option I've been considering:
1) Change the switch. My understanding is that I need a switch with a lower Vth and a smaller gate charge. I haven't found one. :-( It seems the Rds on is not a big deal. Am I wrong?
2) Use an external gate driver, maybe with a bigger MOS or with another MOS in parallel.
3) Reduce the frequency and increase the inductance.

Are there other options? Can someone advice?

Also, I haven't successfully added parallelised components to the WeBench shematic editor. Where can I find instructions? More specifically, I tried to parallelise the Schottky diode and haven't found the Toshiba CMS06 (as proposed initially) in the proposed list.

Thanks in advance!

P.S. I truly enjoy TI design tools and community. Cheers!

  • 0
    TI__Expert 4100 points
    Hi Marceau,
    You are correct, the MOSFET will get hot.
    With your Vin/Vout combination the average current through the switch will be significant (>7A).
    From your comments above I'm not clear if the circuit you are showing here is working or not.

    In general your thinking is correct, but here are some comments:
    1. Rdson matters, since in your application duty cycle (with nominal efficiency taken in consideration) is relatively high (>80%). In such case it is very important to minimize Rdson.
    2. Switching losses will dominate, making ~2/3 of total loses on the switch.
    3. Regarding Webench try this link for tutorial
    www.ti.com/.../help-overview.page

    Hope this helps
    Regards
    Brani
  • 0 in reply to Brani Dubocanin
    Prodigy 170 points

    Hello Rani.

    Yes it helps. Thank you. Could you help me  selecting a gate driver, please? I would like to keep the inductance small, and would rather have a bigger switch.

    Unless TI would have an IC that supports higher gate current (>1A) or that boosts the gate voltage (>Vin)?

  • 0 in reply to HUGUES MARCEAU
    TI__Expert 4100 points

    Hi Marceau,

    I don"t think you need separate gate driver for load levels required in your application.

    My webench simulation is suggesting you use CSD16323Q3 N channel power MOSFET. And when checking all the datasheet specifications this should be sufficient for load level in your application.

    Best regards

    Brani 

  • 0 in reply to Brani Dubocanin
    Prodigy 170 points

    Hello Brani.

    I thank you.

    I agree. CSD16323 should work. I can lower the switching frequency to reduce operating temperature. Is 60C an acceptable operating temperature?

    I've simulated the same schematic Webench proposed, reducing Rfb2, in order to get a lower Vout (going from 20V to 5V), and obtained unacceptable voltage ripple (going from 1A to 4A)... What would you recommend to improve the "output voltage robustness" of the converter?

  • 0 in reply to HUGUES MARCEAU
    TI__Expert 4100 points
    Hello Marceau,
    There should be no issues with power FET operating at 60degC (even slightly higher), but it is always recommended to keep device temperature as low as possible.
    Regarding output voltage ripple...I would suggest try following:
    1. increase output capacitance. Webench is recommending ~180uF of output capacitance. If you have an option to add some additional ceramic capacitors and possibly some smaller HF ceramic capacitor (for example 100nF close to the IC output) that should help.
    2. try different values for inductor. Using smaller/larger inductor will have direct effect on ripple current through inductor.

    Regards
    Brani
  • 0 in reply to Brani Dubocanin
    TI__Guru**** 191445 points
    Let me just add a comment that Webench typically picks small amounts of output capacitance. I suppose it is for cost and size minimization, but may not always be the best choice for some applications.