LT4356 : easy way to remove auto recovery timer on an already existing hardware..

Hi,

I'm using a LT4356-1 to control the power of motor controllers. Everything seems to work as expected. We are able to soft the start to limit the inrush current of the controller's input capacitors, and current limitation of 50A (shunt resistor of 1mOhm) works fine.

However we are facing a MOSFET (SQM120N10-3m8) SOA problem. Currents of about 20-50A cannot last too long when MOSFET is in its linear region. So I would like to limit the time the MOSFET stays under the max current. To do that I have to decrease the Ctimer value, but the problem is that it also will decrease the Auto recovery timing. Too fast recovery time causes bad reboot of my motor controller that get damaged. I do not find smart to recover so fast when overcurrent occurs. So I would like to have a high recovery time (even infinite) and a small turn off timeout in case of overcurrent.

What could be the best way to do so ? I had the following ideas:

  • Add a diode in serie withe Ctimer to only allow charging it. A second diode/Ctimer2 in parallel could allow 2 different timings. Could that work ?
  • Switch to the LTC7000 which allows a much bigger factor between recovery and shutdown events. I would like to turn off in about 100us and turn it back on in about 200ms. Moreover the LTC7000 offers 2 lines to control the Gate allowing fast turn off while turning on smoothly to prevent inrush currents from being to high. LT4356 has not been thought to do that (without adding a diode and RC filer at the Gate. LTC7000 also allows a very fast turn off mode to protect against very fast and hard short circuit conditions.
  • Adding a resistor in // with the Ctimer has the LTC7000 suggests to prevent from recovering. But would that work on the LT4356-1 too ? How does this resistor do the trick ?

My problem is that I already have assembly PCB of the LT4356 version and I have to find an easy way to patch my current boards. I already have redesigned a new version with a LTC7000 which is a little bit more expensive but I do not have tested it yet.

I feel that my application (50V / 30A) that control power of 2 motor controllers needs a LTC7000 which is more than a surge stopper. Tell me if I'm wrong but the LT4356 is made for preventing high transient voltage and current peaks. It is not specially intended to work as a e-fuse for example. It's an easy to use chip but with basic features. However the LTC7000 is a really high side power MOSFET that is intended to control high capacitance applications, being still able to turn off fastly a power MOSFET in case of a short circuit for example. Are my feelings right ?

Here is the schematics of the LT4356-1 :

(C93 has been decreased to 10nF)

Any suggestion or pieces of advice that could help me making my hardware work safely would be a great help !

Thank you in advance.

Best regards,

MG.

  • Hi MG,

    Since you are fine with latch off after the fault , I recommend using LTC4356-3. As soon as the timer times out the part will latch off. This latch can be cleared by two ways:

    i) powering down the part for atleast 100usec and powering it back up

    ii) Pull SHDN# below 0.4V for 100usec and then pull it back higher than 2.1V with slew rate > 10V/msec

    You can keep the timer very small to shut off as soon as there is any current and reset by controlling the SHDN# pin via a controller.

    LTC4356-3 datasheet can be found here : https://www.analog.com/media/en/technical-documentation/data-sheets/LT4356-3.pdf

    Let me know if there are any further questions.

    Best Regards,
    Ritesh

  • Hi riteshl,

    Thank you for your answer. Your suggestion is more than interesting because LT4356-3 is pin to pin compatible with LT4356-1 (same name..^^). Indeed I'm fine with latch-off because a rise in current to its maximum value means that something is going wrong on my motors (stall or too high load). So it's a safe way to proceed.

    I will replace the LT4356-1 by a LT4356-3, set the Ctimer to its minimum value (10nF) and see if I am able to switch off quickly enough to remain in the SOA region of my MOSFET.

    In the mean time, could you tell me what could make me use a LTC7000 instead of the LT4356-3 in the long term ?

    Thank you.