Post Go back to editing

LTC3894 stops working and heats after short circuit


I use an LTC3894 as a current limiter. You can find the circuit below. Without load and with low load the output voltage follows the input voltage. The nominal working voltage is 48Vin = Vout. When the load increases the output voltage follows the input voltage minus a voltage drop due to the Rdson of the Pmos, Inductor resistor and shunt . Once the current reaches approx. 1.2A the LTC3894 starts limiting. We can increase gradually the load up to the short circuit and everything works well (1.4A limitation).

Problems start when output is suddenly short circuited... 98% of the time, the LTC3894 will work perfectly and limit the current at 1.4A.

The other 2%:

- During output short circuited: there is no current (normally 1.4A), the switch is not enabled and the LTC3894 starts heating up to 100°C (The rest of circuit elements are cold)

- Once the short circuit is removed (no load): the output voltage stays at 2.8V (normally it should jump to 48V), the Mosfet is still not activated, and the LTC3894 is still heating at 100°C...

There is no way to jump out of this state, even with other short circuits, even by reducing the input voltage below the undervoltage threshold...  the LTC is blocked.

The only way is to cut the input voltage and supply 48V again. 

As I said, the circuit works very well most of the time it also fits the simulations. 

I hope that you have an Idea to help...

Parents Reply Children
  • Hi Fil,

    Thank you for your answer. What the circuit should do:

    • The output should follow the input voltage (48V)
    • Limit the current at 1.4A in case of overload or short-circuit


    I do not understand why this IC could not do the job. The datasheet shows multiple examples of circuits where the input voltage starts below the desired output voltage and, thus the output follows the input (ex. Typical Application Figure 13).

    Concerning the current limitation, the IC provides short-circuit faults protection…


    I repeat, the circuit works most of the time. Can this be a latch-up of the LTC3894 due to short-circuit current pulse?

  • Hi HotaR,

    I'm not sure if Figure 13 typical application is designed at output short. This product is still intended for step down converter. The problem that I see is the RUN pin (or enable pin). When you short the output, you are bringing VIN also close to zero. And since RUN is tied to VIN, will shutdown the chip and also run out of control of the switch and the current limit protect.

    If your application is just to put a switch and over-current protection between VOUT and VIN, you should check at our power path products in link that I gave earlier. Please check LTC7000-1

  • In my second post there are captures of signals during short-circuit. VIN is measured (at LTC terminals) it seams that the voltage do not drop as significantly as you propose. The 220uH inductor also prevents VIN to drop quickly.

    Thank you for the proposition (LTC7000-1). Unfortunately, this product will not fulfill the requirements, there is no current limitation (hold the current), it only cuts the output once the current limit is reached (if I understood well)…

  • There is just a resistor between VIN and RUN pin of IC. If RUN has dropped so must VIN of IC too.

    When output is shorted, it is recommended to shutdown the operation. This is the purpose of the LTC7000 and most of our short circuit protection chips. What you are suggesting may not be possible.  If you short the output and want to keep 1.4A running, then it means the converter or whatever it is must dissipate the 48V x 1.4A = 67W of power. That is very big power and must be very hard to handle. 

  • Please check LT8210: 100V VIN and VOUT Synchronous 4-Switch Buck‑Boost DC/DC Controller with Pass-Thru

  • Yes, there is a resistor, 470kΩ between VIN and RUN. VIN voltage stays constant at the terminal and RUN  drops, there is no reason for that. RUN is supposed to be high impedance. Even if internally the VIN would drop there is no reason for the chip to jump to a self-destructive blocked state.

    For information, I also did the test with RUN pin directly tied to VIN. The chip burned instantly after a short-circuit… I think that RUN pin is not high impedance (as it should) there is something that ties to the ground briefly during a short-circuit (see screen shot). This explains why the chip burned when RUN is directly connected to VIN, it makes an internal short-circuit.

    Your dissipation calculation is true for a linear regulator not for the buck (switching regulator) converter I intended to use (LTC3894).

  • Thanks, but I will not try a more complex chip. LTC3894 should do the job, in fact it does, but I think there is an internal issue that occurs during some short-circuits. My application needs to be reliable and I cannot trust this product. Without an explanation of what happen I will leave a side this circuit (and certainly most of Analog regulators).

  • Hi HortaR,

    Apology if expectations were not met. I believe that this is more about understanding fully your application and requirement. ADI has a wide range of product and there's more likely something for every power need. Are you able to connect to a local FAE to discuss your application further? If not, I'd be glad to find you one.