• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Subscribers 63 subscribers
  • Views 763 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.

LM3421 SEPIC Design Equations

R11
Expert 1485 points
Other Parts Discussed in Thread: LM3421

For the LM3421,  I see there is an eval board for a SEPIC design, but equations in the datasheet are only for buck, boost, and buck-boost topologies.  So, 2 questions:

1)      Would the LM3421 be a recommended part to use in SEPIC configuration?

2)      Does TI have equations that can be used for SEPIC on the LM3421?

  • 0
    Prodigy 5 points

    Hello Rustin,

    In the LM3421 spec, page 58, it shows a sepic configuration.
    Being a PWM controller for a ground referenced mosfet, it can drive boost and any boost derived topology.
    Flyback and SEPIC are 'boost derived' topologies.

    AN-2009 is an Eval Board LM3421 for SEPIC, automotive LED applications.

    There is a National app note AN-1484 that describes sepic equations.
    This is a basic voltage output sepic, but you can adapt it for current output for LED drive.

  • 0 in reply to Ed Walker
    Expert 1485 points

    Thanks Ed.

    I assume AN-1484 should be accurate for almost all equations; however, the equations for compensation assume current mode control.  The LM3421 appears to be a hybrid of current mode control and predictive off time.  What are the proper equations to use for SEPIC compensation with the LM342 (e.g. LM3421 datasheet as buck-boost, AN-1484, or something else?

  • 0 in reply to R11
    Prodigy 5 points

    The power stage components will still be calculated the same way Rustin.
    Still Peak Current Mode Control with the PRO helps to cancel some undersirable loop effects.
    Switching frequency will be calculated by Equation (6)  Fsw=25/(Rt Ct)

    Compensation is interesting.
    The SEPIC EVM SNVA414b has a .22uF and 100ohm series.
    That's basically just a cap to ground.
    You can try to calculate it with the equations in the data sheet.
    OR notice that the Ccmp ranges from 0.1uF to 1uF in typical designs, just start with the 1uF and see how it works.

    Predictive Off-time PRO:

    Even though the off-time control is quasi-hysteretic, the input voltage proportionality in the off-timer creates an essentially constant switching frequency over the entire operating range for boost and buck-boost topologies.

    The buck topology can be designed to give constant ripple over either input voltage or output voltage, however switching frequency is only constant at a specific operating point. This type of control minimizes the control loop compensation necessary in many switching regulators, simplifying the design process. The averaging mechanism in the peak detection control loop provides extremely accurate LED current regulation over the entire operating range.

    PRO control was designed to mitigate “current mode instability” (also called “sub-harmonic oscillation”) found in standard peak current mode control when operating near or above 50% duty cycles. When using standard peak current mode control with a fixed switching frequency, this condition is present, regardless of the topology.

    However, using a constant off-time approach, current mode instability cannot occur, enabling easier design and control.

    Predictive off-time advantages:

    • There is no current mode instability at any duty cycle.

    • Higher duty cycles / voltage transformation ratios are possible, especially in the boost regulator.

    The only disadvantage is that synchronization to an external reference frequency is generally not available.

  • 0 in reply to Ed Walker
    Expert 1485 points

    Awesome!  Thanks Ed.

  • 0 in reply to R11
    Expert 1485 points

    What is the best method for determining min and max duty cycle for the PWM capabilities (dimming the LEDs) if using digital PWM method (e.g. some factor of loop bandwidth)?

    Also, what is the drive current of the LM3421?

  • 0 in reply to R11
    Prodigy 5 points

    Digital dimming can be done with the DIMFET, a fet in series with the LED's.
    Min duty would depend on the propagation delay through the part, not at all specified.
    Probably a few 100 nano seconds for the signal to make it from the nDIM to the DDRV output.

    Max Duty, I suppose you could go to 100% but the LEDs would be totally off, with a series pass element.
    So going from full off to full on would take some time for the inductor current to ramp up.

    With the DIMFET in parallel with the LEDS the power stage is always supplying current, the DIMFET just bypasses the LEDs so it is more versatile but wastes more power since it just bypasses the LEDs but the power stage is still putting out current.

    There are many variables involved, Vin, LED voltage stack, led current.

    Drive Current, the GATE drive is +-1Amp peak, from page one of the data sheet.
    DDRV is +-200mA.

  • 0 in reply to Ed Walker
    Expert 1485 points

    OVLO and UVLO are showing Vhys = 23uA x ROV2 and 23uA x RUV2, respectively.  Shouldn’t it be ROV1 or RUV1 instead? 

     

    For UVLO and PWM dimming, if we are OK using the UVLO level of VCC, do we simply do the following?  This would allow PWM dimming but not instantiate a separate UVLO function, correct?

     

    • Populate RUV2 (value not terribly important)
    • Not populate RUV1
    • Zero ohm for RUVH
  • 0 in reply to R11
    Prodigy 5 points

    The current shows up at 23uA LESS that the upper resistor must provide, so that's why the upper resistor is used Rustin.

    For the Dimming, if UVLO or OVLO is not needed then you can use the upper reistor as a pullup and pull down on the nDIM pin as you state. That will work.