• State Answer Suggested
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 64 subscribers
  • Views 1609 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.

LM25085: Much Longer Than Expected Off Time Following Over-Current

Daniel Cianfrocco
Prodigy 130 points
Part Number: LM25085
Other Parts Discussed in Thread: TL431, LM3489

I am using the LM25085 in a 29V->24V Sealed Lead Acid Battery Charger. I use an external current sense circuit feeding a TL431 to set the maximum charging current to 4A. The compensation required on the TL431 is slower than the time it takes for the inductor current to reach the built-in Over-Current protection inside the LM25085.

What I expect to happen: On startup it bangs off of the LM25085 over-current a few times until the TL431 compensation catches up and stabilizes it below the over-current threshold.

What actually happens: When it reaches the over-current threshold it shuts off the switch for one cycle for the correct amount (4.7us), resumes switching for a few cycles, and then shuts off for over 800us.
This is a problem for me because now the average current going out to the batteries is below the TL431 threshold and so the slower (and stable) TL431 feedback loop never kicks in and it just goes from 0-7A every milisecond.

More details: I chose to inject the TL431 into the V_FB pin instead of ISEN (like TI recommends) is because hitting the ISEN limit forces minimum off times around 3-4us which would push my switching frequency too low for my desired inductor. With the setup I have if I bypass the ISEN overcurrent-protection the charger shoots up to 13A and settles at the desired 4.2A charging current at my desired 650kHz switching frequency. Using ISEN would produce a much lower average frequency.

The schematic, scope capture, and simulation are attached. The simulation (using a hand-built model) exhibits the behavior defined in the datasheet in Equation 10. The scope plot shows how it latches off for much longer than Equation 10 suggests.

  • 0
    TI__Guru 64987 points

    Hi Daniel,

    During these overcurrent events, the slower average current loop is likely interacting with the LM25085's current limit circuit by pulling up the FB voltage and thus causing the longer than expected off-times. Perhaps you can increase the bandwidth of this outer loop so that it engages as expected during startup.

    Also note that the ripple injection circuit's coupling capacitor looks somewhat low based on the FB divider impedance -- see the LM(2)5085 reference designs and quickstart calculator tool available here: http://www.ti.com/product/LM5085/toolssoftware. From a layout perspective, the FB node is noise sensitive so its trace length should be as short as possible.

    Regards,

    Tim

  • 0 in reply to Timothy Hegarty
    Prodigy 130 points

    Hi Tim,

    Thank you for the quick reply. The issue is that the slow loop isn't charged up yet. In the attached screenshot you can see that V_fb is held at a constant 1.11V during the off-time and that the TL431 has not started conducting. I have tried to increase the slow loop bandwith to get it to catch the charging current before it hits the hard LM25085 OCP but oscillations creep in if I go any faster. Is it possible that the internal soft start is getting fully discharged? That's the only way I can think of that would prevent the converter from running.

    You're right that ripple network is out of date. My apologies, what I'm actually using is R25=66k, C13=3300pF, and C19=0.1uF.

  • 0 in reply to Daniel Cianfrocco
    Prodigy 130 points

    The only solution I found was to change to the LM3489 since it is functionally very similar to the LM25085 so that a complete redesign wasn't necessary. The benefit of the LM3489 is that it does not have any internal soft start behavior, instead it simply bangs off of the overcurrent limiter which is exactly what I needed. Now the average output current is high enough for my slow external loop to take control and it remains stable through the whole charging cycle.