• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 62 subscribers
  • Views 503 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.

LM3481 design with resistance on input

ofirbo
Prodigy 230 points
Other Parts Discussed in Thread: LM3481

Hello,

We have a design for boost converter with input from 9-21[V], Output of 21[V] and load of about 1.5[A].

When we connect Vin directly to supply, the boost converter works fine with all the different duty cycles at different voltages.

However we need the design to work also from a POE Ethernet cable of about 100[m], which have about 12[Ohm] resistance.

This is a problem because even when we input 36[V], the voltage drop over the wire is about 18[V] which means the LM3481 will see 18[V] input which should be enough - but still it doesn't work, It tries to draw more current and drops the input voltage.

We added more capacitors at the Vin input but it doesn't solve the problem. I suspect that the maybe the current sense / feedback doesn't know how to compensate for the voltage drop from the cable?

Attached below is the schematic for the converter which was generated using the Webench tool.

Help will be much appreciated.

Thanks,

Ofir

  • 0
    TI__Expert 6860 points
    Hello Ofir ,

    The boost solution used by you in this application is load dependent ie only when you draw a certain current , then device will boost to 21V and in rest of the cases should follow the input voltage ( minus the diode drop). The best solution would be to design a SEPIC based on LM3481 .You may use Webench for the solution .

    Now coming back to boost , the design shared by you suffer from loop instability . Did you completely follow the Webench design ?
    Can you please change Inductor to 56uH and current sense resistor to 15m (instead of 3m).Also change the RC compensation components to R7:1.3K and C11:330nF (you may remove C8)and re do the test .

    If you face further issues , please share the switch node waveform(FET)'s Drain) as well as output and input waveforms .

    ----Ambreesh
  • 0 in reply to Ambreesh Tripathi
    Prodigy 230 points

    Hi Ambreesh,

    Thanks for your suggestion. in order to analyze the situation I've tried to build a TINA simulation from the reference design.

    I'm not sure that I'm modeling the 100m cable correctly as a RWire=12 [Ohm] resistance at the input.

    I've tried 2 parameters current sense resistor (R5), 15m and 100m. As can be seen at the transient outputs attached, with 15m there is a large voltage drop, but with 100m it starts to look stable, although not outputting the full 21[V] output as designed.

    Do you have any further suggestion or maybe the simulation is totally wrong? When we tested it in the real world it worked pretty consistent with the simulation results.

    Regards,

    Ofir

    snvm418b_grisha.TSC

  • 0 in reply to ofirbo
    TI__Expert 6860 points
    Hi Ofir ,

    we are working on this and will reply to you soon.

    --Ambreesh
  • 0 in reply to Ambreesh Tripathi
    TI__Expert 6860 points

    Hi Ofir,

    To ensure the stability of dc-dc converters, the input source impedance should be small compared to the converter's input impedance.

    Below is the explaination :

    Now Pout Max with 12 ohm of load is 25.92W .Hence Iout(max)=1.23A 

    Also Vin=Vs/2  (Source impedance same as DC/DC impedance)

    where the load line is tangent to I-V curve and has one solution only .FOr larger Rs or larger Pout , Eq1 has no real solution and the DC/DC converter no longer fucntions properly .

    Now when you load 1.5A at 21V output , the solution will always be unstable because the eq1 will have no real roots .Effectively what will happen is , when the boost is operational the input current will increase and hence the supply voltage to DC/DC ocnverter will dip which will further increase the supply current and this will repeat and hence the instability .

    The solution here is to limit the input current such that the the input to DC/DC converter is higher than 18V which amount to INPUT current(Not OUTPUT) lesser than 1.5A .

    I would suggest to use two stage solution :

    1. Use buck converter to get 15V@ 1.5A output

    2. Use boost to get 21V @ 0.95A output

    --Ambreesh

  • 0 in reply to Ambreesh Tripathi
    Prodigy 230 points
    Hi Ambreesh,

    Thanks for the elaborated answer.

    Ofir