LT4256 hot swap controller enable/disable

Has anybody used the LT4256 hot swap controller as an electronic circuit breaker? I have simulated and designed a board that is essentially identical to the evaluation module (resistor scales are different due to different source voltage). I am trying to enable (turn on or off) the LT4256 by pulling UVLO pin low using an external MOSFET circuit, similar to Fig 11 of the datasheet (see link below). 

Test Setup:

I have a 36V source (battery) connected to the input, and a 32 Ohm power resistor load. When closed, the resistor draws a little over 1A. An MCU toggles the external MOSFET gate low to attempt to open the larger series NMOS. 

Problem:

When I attempt to connect the MCU GPIO to the MOSFET gate, the LT4256 IC heats up to over 100C (verified using a thermal imaging camera). When I disconnect the MCU control signal, the LT4256 cools down and seems to operate properly. 

Any thoughts or suggestions? I will test again soon, but with a manual short across UVLO. I realized may not be controlling the FET with a high enough gate voltage. Thank you!

Useful links:

lt4256 datasheet

lt4256 evaluation module

    •  Analog Employees 
    on Feb 2, 2018 2:35 AM

    The UV pin current of the LT4256 is less than 3uA, so you should not see the LT4256 getting hot in the configuration you have described.

    Can you provide your complete schematic? (An LTspice schematic is fine.) The image from the thermal camera might help also.

    Is there any chance that the heat you see is from the 32ohm resistor that is dissipating >36W in your circuit?

  • Thanks for your prompt response! Here's a snapshot of my sim file, also attached.
    The 32 Ohm resistors are off board, and I confirmed (touch) that it is indeed the LT4256 that gets hot. I also tried manually jumping the UVLO pin to GND instead of using a small discrete nmos, but the same hot IC issue occurs. It is definitely related to pulling the UVLO pin low.

    For the series MOSFETs, I am using dual IRFS7530-7PP to block current in both directions instead of what's shown in the simulation. It is worth noting that I measure V_gs = 0.0V on the big series MOSFETs even though I see pass-through current to the load from the power supply. 

    Thanks a bunch!

    e_ckt_brkr.asc.zip
    •  Analog Employees 
    on Feb 2, 2018 11:59 PM

    The input series MOSFET (M4 in your circuit) is damaged. The IRFS7530-7PP's SOA plot shows that it can only support 1A for 1ms when the drain-to-source voltage is 36V.

    I recommend that you review this article concerning safe operating area (SOA) of MOSFETs:

    http://cds.linear.com/docs/en/lt-journal/LTJournal-V27N1-05-di-SOA-DanEddleman.pdf

    Your MOSFET has failed, and now the drain and source of the MOSFET are shorted. When you pull the UV pin low, the LT4256 attempts to turn off the MOSFET by pulling the GATE pin low. In your circuit, the failed MOSFET M4 is now shorting the 36V input supply to the anode of protection diode D1, and that diode holds the voltage of the GATE pin near 36V. The LT4256 is pulling down the GATE pin with roughly 62mA. The resulting power burned in the LT4256 is P=36V*62mA=2.2W. That is the reason the LT4256 is hot when you pull the UV pin low. In your simulation, you can add a wire from the MOSFET drain to source to see the behavior caused by the failed MOSFET.

    Good luck! You might consider repeating the experiment with a more suitable MOSFET like the Nexperia PSMN4R8-100BSE. You might also wish to simulate future designs using the SOAtherm model distributed with LTspice.

    http://www.linear.com/solutions/5239

  • Wow, your diagnosis was spot on! The M4 MOSFET was indeed shorted, and upon replacement the circuit works as expected. I can toggle 'on' and 'off' the circuit breaker function.

    Concerning the SOA of the FET, I had designed for steady state conditions, where Vds << 36V, rather than the transients as your article points out. Perhaps the UV and FB resistors were designed improperly that lead to more transient behavior than expected. The minimal operating voltage was right around the 'low' thresholds. Would you say this contributed to the demise of that shorted FET?

    Also, given your expertise in LT's hot swap controllers, how tolerant would you say the LT4256 is to bidirectional current?  I am using this for a battery application, and negative current is required for charging the battery. I don't need to implement an overcurrent trip for the negative current, but is there a safe threshold for a negative voltage on that Vcc minus sense line?

    Thanks!

    •  Analog Employees 
    on Feb 6, 2018 4:32 AM

    Steechung,

    I'm glad to hear the circuit works now. I recommend using a different MOSFET in your design, since the SOA of the one you chose is very small. An IPB017N10N5LF should fit the same footprint (7-pin D2PAK), but it will have much better SOA.

    Starting up into a resistive load is often the worst case SOA event in hot swap applications. As you mentioned, the choice of foldback resistors can help to reduce the SOA. The tradeoff is that if you foldback too much, the circuit might fail to startup into your load. Reducing the TIMER capacitor value also helps to reduce the SOA that occurs during a fault, i.e. a short-circuit at the output. The third variable that affects the SOA is the GATE capacitance. In your design, a smaller GATE capacitor would allow the MOSFET to turn on faster, which in turn allows the TIMER to activate sooner and shutoff faster in a fault condition.   

    I've attached your LTspice schematic where I've added the SOAtherm model for the PSMN4R8-100BSE. I have also stepped the output load resistor values and applied a short-circuit fault at 0.5sec.

    Allowing reverse current from the output to the input should be fine. The LT4256 pins can be safely taken to 0.3V below ground. There is no need to limit the voltage between the VCC and SENSE pins, but it is reassuring that the 3mohm resistor will hold the voltages fairly close together. Be aware that there will not be any current limiting in the reverse direction. Also, you might consider if the back-to-back MOSFET configuration in your design is necessary. A single MOSFET will allow reverse current through its body diode, which is desirable in your application. With the back-to-back configuration, if your input battery voltage is too low, the LT4256 will not be powered and you won't be able to charge the battery.

    Regards,

    Dan 

    e_ckt_brkr_05FEB2018.asc.zip