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.

INA239: Best way to protect it

Part Number: INA239

Hello all,

my goal is to design a very reliable PCB to test a product (motor). The INA239 measures the bus voltage (48V) and current (Imax 20A) going inside the motor. We are using an industrial power supply to simulate the battery, so our use is similar to the Figure 8.2 of the datasheet, but only one direction to the load, no charger needed. We use a second INA239 for the auxiliary input 12V, Imax 2A (LEDs, communication, etc).
My first design works fine, but the INA239 (on the 48V bus) eventually breaks when there is something wrong with the product. We had one board working for around 4 months without problem and in one week 4 chips have broken. The INA239 gets really hot and the resistance between 3v3 and GND are close to 0. The worst case the SPI lines got affected and also burned the MCU. I guess for this an SPI isolator will do.
The main problems about the product I can think are: short-circuit of the 48V and GND lines; 48V and GND lines switched during the motor assembly; bad firmware - short circuiting the H-bridge; motor back-emf.
Many times the board can detect the short or that something is wrong or the MCU watchdog kicks in opening the switch. Other times don't and the chip breaks, which worries me.
As protection there are TVS diodes on the line and the mosfet switch is programmed to open when the current passes 19A. The typical current at max power is around 17A. I have added screenshots of the system.

I know that the pcb and the protection of the INA can be a lot improved, as I am not an experienced PCB designer and is my first time using the INA239, every help is much appreciated.

Thanks,

ps1: the alert lines are open, cause the MCU run out of pins. I don't belive using it, will make a lot of difference with our issue, since the chip is instantly burned.
ps2: the ground lines (product, PCB, power supply) are tied together with a distribution block inside a panel. The external diode is connected directly to the 48V / GND connector of the product.

  • Hi Gabriel,

    the maximum negative input voltage of INA239 is -0.3V. A bidirectional TVS wouldn't help then.

    Why do you switch-off the motor at the full motor current? Wouldn't this cause huge inductive kick-backs? Do you have a decoupling cap at the motor or a motor filter to limit the inductive kick-back?

    And what do you supply with the 12V?

    Another issue is poor grounding. The INAs not allowing big negative input voltages need a state-of-the-art grounding scheme to prevent destryoing ground noise between the motor and the INAs.

    Kai

  • Hi Gabriel,

    It sounds like an issue somewhat more global, since the MCU got damaged sometimes. The fact that INA239 supply is shorted to ground points to over voltage (relative to GND) as one of the potential root causes. Has this been inspected?

    I understand it is not trivial to capture the failure due to its random nature. But is there any commonality in terms operating condition leading up to the failures? The bus voltage, with TVS protection, is far below the capability of INA239. So it should be OK.

    Regards, Guang

  • Hi Kai, thanks for the reply,

    I don't switch-off the motor at full current, and yes, I believe, it would create huge inductive kick-back. I don't control the motor with this PCB, it is not a motor driver. The full product, which has a motor in it, is what we test. I do believe that the people who developed the product put some protections on it. Also the product/motor is controlled via a software (that I didn't write). If there is a short-circuit on the input of the product, the moment I close the mosfet switch, I can detect the high-current and open the switch again. The motor is not moving at this point, since is the very beginning of the test, and the INA239 in this situation does not brake.

    The 12V is there for possible extra components like lights or whatever accessories they have planned. There is on the product an output connector for this 12V, I connect a normal resistor to simulate a current. Inside the unit they have a switch, which is commanded via SW to enable the output. Basically it just test this switch of the product. The INA239 measuring the 12V bus has never shown failures until now and I am not really worried about it.

    I think your input about the negative voltage might be the issue I am having. I've never considered it could be negative and that the back-emf would be a positve voltage spike that the TVS would protect against. The bidirectional TVS would just clamp the negative voltage at -77V, but by then the chip is gone. I've come up with 2 solutions, that I am not sure it would work to protect against reverse polarity.

    The first solution is to use an external fat schottky diode between both TVS diodes. It will probably get very very hot. I have only experience with small diodes to protect USB or 12V inputs but the idea is the same. I've found this big guy here:
    https://www.mouser.de/datasheet/2/389/stps240h100tv1y-1892390.pdf

    The second solution, that i am not sure if it would work, is to put a schottky diode in parallel with the Rshunt. I have to find one that have Vf less than 0,3(?). If this idea works, I could implement directly in a new redesign of the PCB.



    Regards, Gabriel

  • Hi Guang, thanks for the reply

    I haven't done a very proper inspection. I measured only the resistances between the 3V3 and GND lines and also the SPI bus line. I could check that the 3V3 line and, sometimes, the SPI lines are shorted. Once I removed the MCU and the defected INA, and replace them, the board works completely fine.  That is why I don't believe it is a problem at other part of the board. I guess when the INA burns, sometimes the SPI lines also get shorted inside, affecting everything on the bus.
    I believe the input from Kai about the negative voltage might be the issue here. The TVS diode wouldn't protect againd reverse voltage.

    Regards, Gabriel

  • Hi Gabriel,

    Given that the device is on the high side, I would imagine it is not very likely for a negative common mode voltage to be forced upon INA239 inputs. I’m curious to learn the condition/configuration under which this situation happens.

    But in any case I’m glad that you’re able to zoom in on a possible root cause, and came up with solutions.

    Regards, Guang

  • Hi Gabriel,

    when you have a short circuit in the product and the PROFET opens at 19A, this can happen:

    gabriel_ina239.TSC

    Kai

  • Or with a normal ohmic load:

    Kai