• State Resolved
  • Locked Locked
  • Replies 19 replies
  • Subscribers 40 subscribers
  • Views 3789 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.

PMP4478: 5V 3A resonant PSU

Andre Strul
Expert 1330 points
Part Number: PMP4478
Other Parts Discussed in Thread: UCC28740

hi to all

We have successfully build our first proto of the PSU...however we have a problem with the feedback circuit
We are using very low esr capacitors for the PI filter and we are observing some problems (and higher noise/ripple  than normal).

How can we modify the feedback circuit for low ESR capacitors (10-15mOhm)

thank you in advance for your support.

Andre

  • 0
    TI__Prodigy 605 points

    Hi

    Shall we know the detailed configuration of output caps? Is there any difference from the PMP4478? The schematic would be helpful for us to understand what is the difference. Thanks!

  • 0 in reply to Yingyang Ou
    Expert 1330 points
    Hello Yingyang (nice name!)
    yes there is a difference, we are using a 820uF low ESR (10mOhm) + inductor of 33uH and another electrolytic 820uF (10mOhm)
    so the PI filter was changed nothing else.
    Thanks
    Andre
  • 0 in reply to Andre Strul
    TI__Prodigy 605 points
    Hi Andre

    Thanks! But I would like to confirm that the components you mentioned is C13, L4, C14 in PMP4478. If so, we believe the system should be easy to be unstable. The L4 and C14 would generate a double pole at 1/(2*pi*(L*C)^1/2). Larger inductance means lower frequency it will locate, as well as more phase loss under its cross frequency. Meanwhile, less ESR cap means the zero introduced by cap should move to a higher frequency. It also means less phase under the cross frequency.

    I would recommend to remove the 33uH inductor or change it to a very small value. Then, we can consider to optimize the feedback loop to compensate more phase for the low ESR.
  • 0 in reply to Yingyang Ou
    Expert 1330 points

    Yes the PI filter that we changed is C13, L4, C14 . What do you suggest is the higest value for L4...can we use up to 4 uH ? We are looking to drop the ripple/noise under 3mV. Right now using we changed the l4 to 2.2uH (as suggested by you) and we got the noise/ripple at 9mV. We are also using c13/C14 as high esr capacitors.

  • 0 in reply to Andre Strul
    TI__Prodigy 605 points
    Hi Andre

    We don't try many values on the L4. I think a good and reliable way is to try to measure the bode plot on your board to know the boundary of the L4 value. Actually, I doubt whether the ripple can be suppressed to be lower than 3mV.

    How about place a LDO after the C13 by replacing L4 and C14? It should be easy to achive such ripple range. But a efficiency loss.
  • 0 in reply to Yingyang Ou
    Expert 1330 points
    Yingyang thanks for the quick reply.
    we are already down to 7mV with a hi ESR capacitor. Can you please advise us what type of change we need to make to the feedback circuit to use low ESR capacitor ?
    We cannot use any LDO since we need 3A...

    thanks
    Andre
  • 0 in reply to Andre Strul
    Expert 1330 points
    hi again, gentle reminder.
    regards
    Andre
  • 0 in reply to Andre Strul
    TI__Prodigy 605 points

    Hi Andre

    We already passed your question to our designer. He will help to clear the guidelines.  But he is in a training this week. So, the response may be slightly delayed. Thanks for your patience.

  • 0 in reply to Yingyang Ou
    Expert 1330 points
    Hello, good Monday.
    Checking in to see if the designer is back.

    thanks
    Andre
  • 0 in reply to Andre Strul
    TI__Prodigy 605 points

    Hi, Andre

    Yes, our designer will respond you directly. Thanks!

  • 0 in reply to Andre Strul
    TI__Intellectual 880 points
    Hey Andre,

    TI has written a really good blog post that goes over the measuring and reducing output ripple voltage here: e2e.ti.com/.../2243.understanding-measuring-and-reducing-output-voltage-ripple

    This document goes over understanding low frequency and high frequency ripple, how to measure them, and how to reduce them including layout tips as well.

    Your goal of 3mV is very aggressive, but if you raise your output voltage slightly and place a series regulator like an LDO it may be possible to reach your goal of 3mV output ripple. Typically, with the UCC28740 and optocoupler feedback network design an output ripple voltage of about 1%, or in this case 50mV, is good.
  • 0 in reply to DavitKhudaverdyan
    Expert 1330 points
    Davit,
    thank you for the reply .
    LDO is not an option. We would like to understand whats the highest value inductor that we can safely use.

    regards
  • 0 in reply to Andre Strul
    TI__Intellectual 880 points
    Hey Andre,

    One thing you can take a look at is reducing the cutoff frequency of the LC filter by a decade.

    You can do this by using the equation L = 1 / (2*pi*fc*(C)^(1/2)) where fc is the cutoff frequency, and C is the filter capacitor C14 in the reference design. To reduce it by a decade, choose fc = fsw/10.

    Davit
  • 0 in reply to DavitKhudaverdyan
    Expert 1330 points
    Davit.

    please excuse the delay, see reply here :
    For the feedback circuit , can we take the feedback point after the C13 (before the LC filter) ? Any drawbacks ?

    regards
    Andre
  • 0 in reply to Andre Strul
    TI__Intellectual 880 points
    Andre,

    There is no inherent drawback to placing the feedback behind the LC filter. The issue that can arise is that the converter is now blind to everything post feedback (meaning the LC filter). So anything that happens after C13 the converter cannot regulate and you're letting the passives try filtering rather than having the converter do its job.


    Another thing to consider about the filter is the Q factor. If your filter's Q factor is high, then it may actually be adding gain to whatever noise/signal is at that frequency and adding to the output ripple that you're trying to suppress.

    Davit
  • 0 in reply to DavitKhudaverdyan
    Expert 1330 points
    Hello Davit and TGIF.
    hope the new year started off well for you.

    We are making good progress with our PSU project but have a small question for you:
    R4 and R5 in your schematics are 1206 package. We dont knwo how to calculate the power that will flow in and we want to know if we can change to 0805 package.

    Thanks for taking a minute to look into this.
    Regards,
    Andre
  • 0 in reply to Andre Strul
    TI__Intellectual 880 points
    Hey Andre,

    I'm glad to hear your PSU project is coming along.

    To calculate the power through a resistor, use P = I^2 * R. In the equation I is the current through the resistor and R is the resistance.

    You can change to an 0805 package as long as it is rated for the same power. I noticed in the BOM of the reference design first mentioned that R4/5 are 1/4W resistors and the other 0805 resistors are 1/5W.

    If you can find 1/4W 0805 resistor then the package change should be fine without you having to calculate the power dissipation through the resistor.

    Regards,
    Davit
  • 0 in reply to DavitKhudaverdyan
    Expert 1330 points

    Hi again Davit,

    We want to replace the output Diode with the below (+ mosfet)

    www.ti.com/.../ucc24630.pdf

    can you help us calculate the values of Rs etc for our 5V / 3A device which is based on the TI design. 
    Also , if we replace the diode with this controller and mosfet... should we expect more EMI noise compared with diode (we will use RC snubbers across mosfet)

    thank you for your time and support.

    Andre

  • 0 in reply to Andre Strul
    TI__Intellectual 880 points
    Hey Andre,

    To calculate values for the VS resistor divider I suggest looking at section 8.2.2.6 in the datasheet (www.ti.com/.../application-and-implementation, which provides the equations for the divider network. Another tool is the UCC28740 design calculator which I will link to here: www.ti.com/.../sluc487

    If you replace the secondary diode with an SR Controller + FET, you can expect to have lower power losses. However, faster rise and fall times generate high frequency noise that can lead to system noise and EMI compliance issues.

    Regards,
    Davit