• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 64 subscribers
  • Views 761 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.

LM3478 SEPIC doesn't work as expected on the whole range

Laurent Guillier
Prodigy 90 points
Other Parts Discussed in Thread: LM3478

Hello,

I have designed a SEPIC with LM3478, with the following specs :

Vin : 8V .. 19V

Vout : 12V

Imax : 6A

The design is a mix between Webench (see webench_design_744428_44.pdf) 4760.webench_design_744428_44.pdf and my own calculus from AN1484 and LM3478 datasheet. The final schematics is here attached (ugbCV12V.pdf).4062.ugbCV12V.pdf

L201 is Vishay IHLP6767GZER220M01 (26.5mOhm max DCR)

L202 is Vishay IHLP5050FDER100M01 (17.2mOhm max DCR)

The SEPIC behaves as :

- with no load, Vout is 12V on 8V .. 19V Vin range ; audible noise is heard (MLCC or self ?) : OK

- with 20 Ohm load (0.6A), Vout is 12V on 8V .. 19V Vin range : OK

- With 10 Ohm load (1.2A), Vout is 12V from 19V down to around 10~11V ; under this limit, the T203 gate is not driven correctly any more, a sharp noise is produced, and Vout is not kept at 12V any more. L201 gets rather HOT ... NOK

Hence I wonder what is the reason for that behaviour, and more over if this design will be able to provide 6 amps on the expected input voltage range. Can you, power experts, provide a clue ?

Many thanks

Laurent

  • 0
    TI__Genius 15810 points

    Hello Laurent,

    I took a quick glance at the documents you have provided. You  mentioned that you hear audible noise with increasing load and the inductor heating up. It would be helpful if you could take some screenshots of the voltages and currents while the board is in operation. The current through the inductors, output voltage, voltage across sepic cap and switch node voltage are examples of such. I believe your circuit at higher loads is not switching but is infact oscillating and that could be a reason why your inductor is heating up a lot.

    There were a couple of things I noticed in the schematic. First was the placement of the ramp capacitor. This capacitor is acting as a noise bypass cap at the ISEN pin. Therefore it should be placed right next to the ISEN pin. Refer to the Webench schematic as an example. This was an error in the datasheet and we are in the process of rectifying it. If there is noise at the ISEN pin, the device could hit current limit erroneously.

    Secondly, for a separate inductor design, the choice of inductors largely influences the stability of the circuit. The two inductors and the sepic cap form a dominant double pole that can cause instabilities if not damped properly. Webench will pick inductors based on a certain calculated ratio of Lout (output inductor connected from sepic cap to ground) to Lin (input inductor from input to sepic cap). If you notice on the webench schematic, the Lout is higher than twice Lin. This provides for the proper damping of the voltage across the SEPIC cap. Unfortunately for separate inductors, there is no closed form equation available for the calculation of the inductors. Webench goes through an iterative process to determine the values based on stability. I would suggest following those recommendations if you can. Alternatively/additionally you could also use snubber circuits across the sepic cap to damp the noise across the sepic cap.

    I hope this information is helpful.

    Regards,
    Akshay Mehta
    Applications Engineer
    TI-SVA 

  • 0 in reply to Akshay Mehta
    Prodigy 90 points

    Hi Akshay,

    thanks for your answer. Here are scope views of MOS gate, Switch point and Vout voltages for no load 8585.ugbCV12V-noload.pdf, 20Ω load 0636.ugbCV12V-20ohm.pdf and 10Ω load 4456.ugbCV12V-10ohm.pdf; I didn't add the plot for Vin=8V @ 10Ω load, it seems to be like 12V, but more messy, with Vout not kept at 12V, and system going crazy and wheezing. Note that the 2.2nF C209 cap has been removed, and a 4.7nF has been added between ISEN end GND before getting this measures. Hope this may help you to confirm your felling on this behavior.

    I will investigate my design according to your recommendations regarding inductors value, trying to reverse ratio within the actual footprints.

    Regards

    Laurent

  • 0 in reply to Laurent Guillier
    Prodigy 90 points

    Hi,

    After reversing inductors value ratio, I can confirm that your recommendations did the trick !

    For information, compared to initial schematics :

    - L201 is now 10uH (Vishay IHLP6767GZER100M11)

    - L202 is now 22uH (Coiltronics HC9-220-R)

    - C209 is 2.2nF betwwen ISEN pin and GND

    - Compensation filter components have been adapted : R207 is 1K, C210 is 470nF, C211 is 4.7nF

    Tests coverage is not yet 100%, but till now everything works as expected. Thank you for your support !

    Regards

    Laurent

  • 0 in reply to Laurent Guillier
    TI__Genius 15810 points

    You are most welcome, Laurent. If possible send me your LM3478 schematic.

    Regards,
    Akshay

  • 0 in reply to Akshay Mehta
    Prodigy 90 points

    Hello Akshay,

    Here it is ...1513.ugbCV12V-OK.pdf

    Regards

    laurent