• Resolved

ISO5500's going up in smoke

Hi All,

I recently designed in the new ISO5500's into our latest inverter, with four driving an H bridge, and one for a buck battery charger. They all drive juicy STGW35HF60WD IGBT's that like a +15V for clean turn on, and -15V for a fast solid turn off. I implemented the design to fig 62 of page 27 of the ISO5500 datasheet.

The problem I have is the ISO5500's (all of them) randomly latch up, gets hot, kills itself and then promptly toasts its Isolated power supply through excessive current draw. My conclusion is that I am hitting the ISO5500 with a short burst of power across VEE and VCC2 that exceed the recommended absolute maximum of 35V at turn on.

Like everyone else, we use unregulated +/- power modules (in our case the VDS series by CUI) these $7 modules are supplied by a switch mode. Both the switch mode and the isolated module supplies have a small amount of over shoot. Further more the isolated supplies are unregulated with a +/- 10% spec, so it is very easy to go over the 35V limit during power on. We need to use 15V for IGBT turn on, we need isolated supplies, and we need low cost. Im not using zeners and TVS to "hopefully" catch the overshoot . However even for normal operation, a 2.5V tolerance on each supply is dangerously close to going poof as it is, unless you use regulators, and thats a lot of circuitry for each supply of each channel (10 in my case).

Does TI have a version of this chip that has a more forgiving tolerance for "Positive output supply Voltage, VOUT+ (VE – VEE-P)" specified on page 4 of the ISO5500 datasheet. As +/-15V IGBT gate drive is very popular? But a mere 5V wont do it for common isolated power supplies, including the DCH01 series, which is all over the place - from 15.2 to 16.2 without taking into consideration of overshoot.

Or does TI have a silver bullet - a DCH01 series with +15V and -9V output - that would perfect. But, I dont think they have.

steve

  • Steve,

    the DCH010515 is specified with 15V +/- 0.75V. please explain what's there "all over the place" ?

    Cheers, Thomas

    Best Regards,

    Thomas

  • In reply to Thomas Kugelstadt:

    Thomas,

    According to Texas Instruments data sheet DCH010515D page 7 figure 16, the DCH010515D at 3mA can be as high as 16.5V. An error on each rail to this degree = 33V applied to a circuit that can handle a maximum of 35V. This exceeds that recommended maximum of 30V and leaves a 2V safety margin to the absolute maximum on a circuit controlling IGBT's with up to 600V at 200A.

    thanks,

    steve

  • In reply to Steven Howell:

    Steve,

    I'm so sorry. Your are right and our parametric search table is wrong as it indicates Vout-min and max with 14.25V and 15.75V respectively.

    I suggest adding regulation to your 15V outputs with a couple of cheap TL431 shunt regulators.

    regards, Thomas

    Best Regards,

    Thomas

  • In reply to Thomas Kugelstadt:

    I think I may have misunderstood this website. I got the impression this is where TI's technical support is now provided.

    My question now is, why was my post tagged as "Answered"?. My original post had questions, which are clearly not answered.

    It seems the only person to produce any answers to anything here was me, and that was just to educate those who knew little about the discussion.

    Ill hold back making any further comments about TI, marketing, clouds, communities and ridiculous self collecting point systems.

  • In reply to Steven Howell:

    Sorry that you feel like this. Unfortunately no one took over while I am on vacation. Neither has TI a more forgiving device with wider tolerances. Hence my short suggestion from vacation to use an unregulated DC/DC converter and use shunts to keep overshoots from occuring.

    Sorry for being not clear enough. Thomas

     

    Best Regards,

    Thomas

  • In reply to Thomas Kugelstadt:

    No worries.

    thanks,

     

    rgds,

    Steve

  • In reply to Steven Howell:

    Hi Steve,

    Have you found the problem with the iso5500?  I am also considering using this driver.


    Thank you,

    Stefan