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.

LM25069-2 when there is overcurrent it disconnect and doesn't turn on again

Other Parts Discussed in Thread: LM25069, LM25066

Problem:
When we put overload, IC turns off and doesn't want to turn on again.
 

VIn=5 V (from 2x30A 5V ORing). Imax is defined as 33,3 A. If we provide more, MOSFET shut off. 
Vgate is 0V (before it was 12V).
Once it did turn on again after disconnecting power for long time, now it does not want to start again at all. I'm not sure if IC is still OK,
but I do not see reason why it would get destroyed (IC or perhaps MOSFET).

Any idea if there is failure in design or what we could check?





1-Vout, 2-power good , 3-timer , 4- gate

  • Hi Ales,

    When there is that much current being turned on/off, it is good practice to include a TVS diode on the input and a schottky diode on the output. You can check out the presentation below, slide 5 for information. These protection devices help clamp inductive voltage spikes.

    6712.2_4_16_Hot Swap Presentation.pdf

    From the waveform above, if the output voltage is up while the GATE is off, then that indicates the MOSFET is likely damaged. A MOSFET can usually get damaged due to exceeding its transient power capability (it's SOA curve - or Safe Operating Area curve from its datasheet). However I went through the SOA calculations and I believe the MOSFET has adequate transient power capability for this design.

    I would recommend running experiments to determine if it is the MOSFET, the IC, or both. Adding a TVS and schottky diode should help clamp inductive voltage kicks when shutting off the high current. It could be that these voltage kicks exceeded the ABS_MAX of the IC or FET.

    For reference, we have a design calculator tool for the LM25069 here:

    I got one started, attached below based on the schematic. Using the tool I see the MOSFET power limit is roughly 24W for a fault time of roughly 12ms. The MOSFET can handle up to 32A at 5V Vds (160W of power) for a duration of 10ms. So this design should be operating within the SOA curve of the MOSFET (safe) even under a short circuit event on the output, and at elevated ambient temperatures:

    LM25069 E2E.xls

    Thanks!

    Alex

  • Thank you for comment. I confirmed that MOSFET is OK. Very likely Lm25069 is destroyed. I Will order new samples and test with additional protection as proposed.

  • Hi,

    LM25069 was destroyed. I replaced it and add all proposed protection, now it works stable. thanx.

  • Hi Ales,

    I am glad to hear it is working well now. I will close this E2E thread, but please post a new thread if you have any future questions about our products.

    Thanks!
    Alex
  • Hi,

     module work stable now by using suppressor diodes. But, I did found one other issue:

    When I turn on power supply, Lm25069 doesn't start if load >=4 A. I do not use PGD to drive additional MOSFET to conenct load when Vout is stable. Can this be the reason or is there any other catch here?

    I attach oscillogram what happen at startup:

    1- VIn

    2-V GATE MOSFET

    3- V TIMER ( C=680nF)

    4-VOUT

    As you can see V gate doesn not go above 3V so MSOFET does not turn on at all. if load is belo 4 A, everything is OK, Vgate goes to 11 V.

  • Hi Ales,

    The timer running indicates the part is in regulation mode (either power limit or current limit).

    In this case, looking at previous discussion I believe you had the power limit set to 24W while VIN = 5V.

    This means that when the device starts up, there is initially 5V across the MOSFET and the part will initially regulate current to around 24W / 5V = ~4.8A This isn't much design margin since the power limit may vary quite a bit, and if the load will quickly draw 4A of current.

    Overall the simplest solution would probably be to double the power limit by changing Rpwr to about a 18k. This should give you more headroom such that 4A of load current shouldn't be affected.

    Thanks,

    Alex

  • Hi,

    I checked with new resistor value and now it start  with current around 8 A, which I think is much better and I think it will be acceptable. Thank you for support.

  • Hi,

    after extensive tests we noticed that we can still get LM25069 destroyed when we do short circuit nearby. we use current limit 34 A with 1,5 m ohm resistor.

    Description:

    we connect load 30 A (dark yellow line), which has peak around 50 A when we do short circuit. Red is 5V Vout. Blue and green are voltages on sense resistor 1,5 m ohm and light yellow is voltage over sense resistor, which is a bit over 40 mV before short circuit and it jumps to negative -450 mV when we do short circuit and MOSFET turn off!

    Question:

    according to datasheet is voltage VIN-SENSE < +- 300mV. Is it possible that we destroy IC with this short overvoltage pulse. If yes, do you have solution how to prevent this  - can we add R_C filtering of sense resistor, like, for example, you propose for INA 199?

  • Hi Ales,

    Check out the layout guidelines in the LM25066 datasheet (which is a similar device, but with PMBus):

    Section 11.1 on page 55.

    - Basically it is not intuitive, but it helps to ensure there is not capacitance close to the pins, which allows the pins to spike up together rather than separate (debiased).

    However I would warn that that particular spike is probably not the problem. Although the spec is 300mV, it is due to a PN junction diode which will typically have a higher threshold and will probably not conduct much at -0.45V.

    From the waveform, I would be more concerned about Vout. The Vout will spike negative due to the current being shut off (large di/dt) and any inductance on the output (even just trace). The part has an ABS_MAX spec of -0.3V but in truth, it comes from an ESD diode (like most IC's) which is a PN junction diode and will have a forward voltage threshold closer to -0.6V to -0.7V. So the goal is to prevent this ESD diode from conducting and causing too much current through the IC.

    Capacitance on the output will help absorb this negative spike. So if you add capacitance or add some closer to the IC, it will help. But in some cases (especially with high current levels like 30A) you will end up needing a schottky diode on the output (pointing from GND to the OUT pin_ to absorb this transient. The schottky diode should at least clamp below -0.7V (preferably below -0.6V) to prevent the internal ESD diode from conducting much current.

    Additionally, this app-note may be of interest (although it may not directly be the problem). The app note discusses system level transients for hot swap and why the sense resistor inductance can cause problems with a large di/dt.

    Finally, while an external schottky diode on the OUT pin may be the most reliable method to help - in a pinch you could (at your own risk) simply put a resistor in series with the OUT pin. This will limit the current through the ESD diode and help clamp below -0.7V. Theoretically, the OUT can never fall below the threshold voltage of the MOSFET (because GATE is 0V and if OUT/SOURCE goes below a Vth, then the FET will turn on and VIN will start passing to VOUT such that you don't go below Vth). So if the output cannot fall below -2V to -4V (FET threshold) then a 1kohm resistor in series with the OUT pin will limit current to <4mA which is probably safe.

    Thanks!

    Alex

  • Hi,

    thank you for feedback. I attached latest scehmatics, as you can see suppressor diodes and schottky diode are already in.

    As you proposed I Will perform 3 tests:

    -position of 1nF capacitor in design is too close to VIN, I Will change that

    - output schottky diode: perhaps voltage drop is too high, i Will test with diode with lower Vf

    - 1k resistor: this I do not get it but I Will try . Can you confirm position of 1k resistor (marked red)?

  • Hi Ales,

    That is the correct position for a resistor on OUT. I would have guessed that Schottky + caps would be good enough to clamp the output to a safe voltage level but then again, the waveform showed it exceeding -1V.

    Basically what I was suggesting is the output negative spike can never fall below -2V to -4V or otherwise T7 will turn on (since the GATE will be held to IC_GND which is 0V - so if OUT went below -4V due to an inductive spike, then there would be a gate to source voltage for T7 to turn on).

    If the VOUT rail can be at worst case up to -4V below GND, then current will flow from IC_GND to OUT through an internal diode (ESD protection). This current will cause the IC to heat up and be destroyed, which is why many IC pins will have around a -0.3V rating. In the end, it is power through the diode which is a concern. So if it is difficult to reliably limit/clamp voltage, then another potential option is to limit the current. In this case since OUT is not a high current path, you can put a resistor in series to limit the current. A 1kohm resistor will limit the current below (Vout - Vdiode / Rout) = 3.3mA.

    Thanks!

    Alex

  • Hi,

    by using 1k resistor or power schottky diode with low Vf  I can reduce negative voltage transient from 1,5 to below 0,5 V. Looks now all Works stable.