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.

INA240: Reverse Voltage Failure Questions

Part Number: INA240

Hi,

At present, I'm looking to develop a current sensor capable of measuring up to 300A of current on a 50V power source, and I'm trying to make my system robust in case of a reverse voltage event (rare, but potentially very costly). I have a circuit designed using the INA240 on a 3.3V supply with the REF pins biased by a voltage divider on the same 3.3V supply. While I've been able to incorporate upstream reverse voltage protection circuits for lower current (20A or less) sensing applications, I don't see this as a viable option for my application due to the very high current and my space constraints. I've also looked for other current sense amplifiers designed to handle -50V, but I haven't really found anything. I'm willing to let the INA240 fail (i.e. fry), provided the power source's reversed 50V doesn't cascade through and create reverse voltages on the output or the 3.3V supply. The 3.3V supply is downstream of the current sensor and does have reverse voltage protection in place, so it will not be supplying any power in a reverse voltage event.

Do I need to be concerned about a sustained -50V common mode voltage creating negative voltages on the VS or OUT pins before the chip fries? If so, what would be the best way to guard against this? I've thought about using TVS diodes to clamp the pins and fuses to break their connections if there's unusually high current (say, 50mA), but I wanted to see if anyone else had any ideals.

Thanks,

Wil

  • Hi Wil,

    Vs and OUT pins will see the negative voltage before the chip fries. A -50V common mode voltage will destroy INA240 certainly, but I’m not sure the propagation of this harmful voltage can be stopped afterwards..

    Your idea should work. The TVS has to be able to pass enough current to blow the fuse. I would add a pair of current limiting resistors before the input pins. These can be the same ones for input filtering as found in datasheet diagrams.

    In addition, a diode between GND pin and system ground will help as well.

    These are the few things I can think of for now, hopefully can be of some help.

    Regards, Guang

  • Guang,

    Thanks for the response. I looked into adding the filtering/current limiting resistors, and I noticed the filtering resistance was recommended to be kept below 10 ohms. That got me wondering what the typical resistance of a low ampacity fuse is; after some searching, I found a 28mA fuse with a maximum resistance of 7.5 ohms.

    Rather than just sacrificing the INA240, what are your thoughts on changing the Rs filtering resistors to these 28mA fuses and putting unidirectional TVS diodes between the fuses and the IN+/IN- pins? That'd clamp the IN+/IN- pins at a safe level (I'm guessing around -0.7V), the resulting 6.5+A each would very rapidly blow the fuses, and the INA240 would be safely disconnected. For a 58V, 600W TVS diode (just to grab one), it looks like the maximum reverse leakage current (even at 85C) would be 1uA, which is a fraction of the INA240's nominal 90uA input bias current, so it seems to me that normal operation would be minimally impacted. I suppose I could see inconsistencies in fuse resistance affecting the results, but for a given system, couldn't the resistance mismatch issue be calibrated out?

    Thanks,
    Wil

  • Hi Wil,

    I think your plan makes sense and should be feasible. The mismatch of fuse resistance is probably not superior, but since the resistance is small be begin with, it shouldn’t be something too worrisome.

    Calibration will remove most of its effect as you proposed. There will be certain 2nd order effect that can’t be completely taken into account with a simple offset calibration. But I think the residue impact is small.

    Regards, Guang

  • Guang,

    The 2nd order effect is good to know about, but I'll tackle that issue if it significantly presents itself. I'll move forward with the fused IN+/IN- method and test its efficacy.

    Thanks for the help!

    Wil