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.

LM25141: LM25141 Does not turn on with light load (Vout=0)

Part Number: LM25141

Vin = 20V, Vout=12V, L = 10uH, Rcurrent sense = 5mOhm, Cout=300uF, Iout - not meaning, 450kHz FPWM, SoftStart = 50ms, Hicuup Fault=60ms. EN pin- button

All values ​​correspond to Exel Ref.des (LM25141-Q1 Quickstart Tool - revA2). Compensation Design - 6kHz is designed for real parameters.

 

With a connected load (0.1A small load and more), it does not turn on. At the midpoint Q1-Q2 we see one narrow pulse with a frequency of 450 kHz. No output voltage. Disconnect the load, connect the load - perfect, everything works.

 

What's the problem?

Thanks,

no load then LM25141 turn on

 no Load

smal load (0.1A) then LM25141 turn on

Real ESR for calc Compensation des

Real ESR Out for calc compens

**********

Tests.

  1. No load

Apply voltage -> Turn on EN pin.

The result is great. The output voltage is stable. Q1, Q2 commutation is correct, without overvoltage. We connect / disconnect the load on the included board - it works. All perfectly.

 

  1. We connect the load to the switched off board. We apply voltage, turn on the EN pin - one narrow pulse with a frequency of 450k.

We remove the load, connect the load - it works.

  • Hello Borodin,

    Looks like you may be hitting current limit during startup, especially as 300uF is a large Cout for a 12V output. Also, an electronic loads can pull a pulse of current at turn on. Can you check the inductor current during startup and also verify current limit is engaging correctly in steady state.

    Regards,

    Tim

  • Thanks for the feedback

    Load - 10W ceramic resistor
    The capacity (Hiccup Mode Current Limiting) has not been set. Hiccup protection is disabled.
    Power supply LM25141- 5V external stable (all time!). 450kHz internal.
    The enable pin (En pin LM25141) is stable (all time!).


    The problem is identified here.
    LM25141 works, then turns off. There is no internal frequency of 450 kHz.
    The running time depends on the input voltage. Less input - less generation time. More time - more generation.

    We have installed a Hiccup Res capacitor.


    1. The inductance current does not exceed 75mV - the current limit comparator should not work (we cannot see it on the oscilloscope).
    Installed RC current differential filter (10Ω + 10nF) - unchanged

    3. Resistor inductance current changed 5mOhm -> 2mOhm. Nothing has changed.

    drain - source Volt (power switches on / off)


    confused ... what can the LM25141 turn off ?

  • Borodin,

    Just note that a 100ns time constant on the current sense signal is excessive.

    Please send the completed quickstart calculator file. Looks like you have some issue at startup - is that correct?

    Regards,

    Tim

  • I guess the Feedback Compensator is unstable

    Photo - the inductance current from the oscilloscope (from the inductance, not from the resistor)

    coil - made, accurately calculated
    L = 10uH (from 10kHz RLC tester)
    Cout = 515uF low ESR capacitance, multiple, in parallel.

    Board tracing (neat) - short lines, AGND / GND different sides, power mosfets are separated, the line from the current sensor is clean ...

    Known Current-Mode Control Theory https://www.ti.com/lit/an/snva555/snva555.pdf?ts=1618828129463&ref_url=https%253A%252F%252Fwww.ti.com%252Flit%252Fsnva555

  • Hi Borodin,

    the plots above are of the SW voltage, not current.

    FOr the Excel file, please ensure that the output cap is correctly entered (use the derated value of ceramic capacitance for applied voltage). Also, 800Hz is really low for the corssover frequency - try 40kHz for good loop response (10% of Fsw).

    If you need additional assistance, please send your schematic.

    Regards,

    Tim

  • We changed the cutoff frequency to 45 kHz + Soft Start 50ms - it became more stable inside the relay pack. Thanks for the advice!

    Problem - the stabilizer is never stable(relay mode) when the load is connected when the stabilizer is turned on and goes into relay mode over time when connected.

    relay mode time (sleep?) is constant:
    does not affect changed L = 5 ... 20uH, С = 100 ... 500uF, Uin = 17..24V, Fcomp = 1k..45kHz

    Inside LM25141 VCC - VCCX are combined according to the scheme. Suppose there is no VCC or VCCX difference for bootstrap capacitor.

    I guess the VCC pin can not connecting anywhere if 5V externaly powered HB pin+bootstrap capacitor. Tried to short-circuit VCC -> VCCX. Nothing changed.

    ps. Not specified in the documentation. All circuits in the documentation - internal stabilizer

    Internal power dissipation in the VCC
    Regulator can be minimized by connecting the VCCX pin to a 5 V output or to an external 5 V supply. If VCCX >
    4.5 V, VCCX is internally connected to VCC and the internal VCC regulator is disabled. If VCCX is unused, it
    should be grounded.

  • Not sure what you mean by relay mode here. When the VCCX voltage is above the changeover threshold, an internal switch connected VCCX to VCC and the LDO bias supply regulator from VIN is disabled. Just tie VCCX to GND to disable that function if needed.

    Regards,

    Tim

  • well ... and so

    I cut off the tracks of the current sensor, made a twisted pair of 2 wires and soldered them to the RiL resistor and the chip leads, ...

    The problem with the unstable circuit may have been solved: when a small load (10W) is connected, the stabilizer starts up normally. All looks like correct.

    Presumably EMI is the problem.

    length of tracks RiL was 5cm

    Output voltage 12V, but dRiL is 50mV (max), so give high DC output+small dVriL.


    Jitter now ~ 50ns (450kHz switching), Fcut = 35kHz, external VCCX.

    ps. the result is positive, I am not closing the question yet - we are checking

  • ok, thanks for the update.

  • so,
    EMI problem confirmed 100%.
    We returned the cut tracks - the circuit is unstable. The circuit is stable (temporarily) only without load.

    The twisted pair of wires was soldered again - the circuit is stable.


    Raised in power up to 35% of the nominal - stable. Connecting and disconnecting the load, turning off / on the chip does not make the relay mode. Jitter - 120ns

    RiL PCB trace - was only 5cm, constant gap differential line in one layer. Didn't help to save.

    The issue has been resolved

  • Thanks, Borodin.

    Regards,

    Tim