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.

TPS7A33: How am I damaging internal reference?

Part Number: TPS7A33

I have a board using a TPS7A33 and a TPS7A4701 to power some opamps. The input voltage is +/-34V and the output voltage is +/-32V. Somehow the TPS7A33 keeps getting damaged. The symptoms are the voltage at the FB pin changes (sometimes out of spec to -1.215V) which causes the output voltage to change. The output also begins drifting around by up to 10 mV (see scope image below). 

I thought this: https://e2e.ti.com/support/power-management/f/196/p/762824/2819981 was the problem so I put a schottky diode in parallel with R52, with cathode at ground. The forum post I linked says anode to ground but I don't think this can be right because it would short out R2 due to the fact that the reference voltage is negative. After adding this diode to a working board however, I still managed to kill the TPS7A33 after a few power cycles.

Here is a screenshot of what the output looks like (AC coupled) after the damage event:

  • PS the diode I used for protection was a PMEG10010ELR

  • Hi Justin, 

    Thanks for providing the after-damage plot but to identify the problem in this application, will you help to provide your power-on cycle plots including Vin, Vout, FB, NR/SS in one plot? If the part is damaged, it's likely there is some voltage violation outside of absolute max. 

    Since you are concerned about the Cff, will you help to remove the Cff and retake the plot of the same four signals?

    Regards, 
    Jason Song

  • I don't want to remove Cff because the datasheet recommends it for power supply rejection, which I want. If I have to sacrifice it to make the board work though, I will.

    I don't have a good board anymore to test with, so these plots are made from board that is already exhibiting the problem. I hope that doesn't invalidate the results.

    I have another set of power supplies on this board which I have been turning on/off manually, so just to be safe I did a measurement of TPS7A33 power up and power down with the the other supplies off and with them on. All measurements are below.

    The only abs max violation I can see is that on one of the power up plots, Vout jumps to around 1.5V above ground, which is higher than the 0.3V abs max in the datasheet. I don't know why this is or how to prevent it, but you can see the same jump in the Vfb plot (although the Vfb is not violating abs max).

    In all plots, chan 1 (yellow) is Vin, chan 2 (green) is Vfb, chan 3 (blue) is Vout, chan 4 (pink) is Vnr/ss.

    Other supplies on:

    Other supplies off:

  • Hi Justin, 

    I posted a reply earlier, but for some reason, it did not go through. I have reviewed your plots and for the one with "other supplies on", the output that was jumping above 1.5V was an indication that the ESD structure on the output pin is damaged. I have also noticed the NR pin voltage only goes to around 0.5V which also means the internal NR circuit may also have been damaged. In the plots, the Vin is still at negative during the Vout jumping to 1.5V, what do you have on the load side? Is there anything pulling output pin higher? 

    The other plots with "other supplies off" have the NR voltage goes above 1V, and I am trying to understand what's the difference between the other supplies "on" and "off" in your application, will you help to clarify? 

    When you have a new set of good boards, I would recommend removing Cff to rule out any possibility of damage due to Cff. It does not seem to be a problem, but I would like to make sure. 

    Regards, 

    Jason

  • Thanks for that info, Jason. I was able to repair one of my boards and remove Cff. I power cycled it 50 times and so far it has not been damaged.

    I tried putting in a CUS10S30,H3F diode to protect the reference but the reverse current on this diode was so high it messed up the output voltage. Do you have a schottky you recommend for protecting against this problem?

    Another option would be to make Cff lower than 10nF, rather than totally eliminating it. When I attempt to measure the voltage on Cff, the impedance of my probe messes up the measurement. Do you have a suggestion for how to measure the discharge of Cout and Cff so I can determine what capacitor I should use for Cff?

    In answer to your question, there are other supplies that power different parts of this board, so I took measurements with those supplies on and off, because when I have been powering up the board I have not always followed the same sequence. Unfortunately I cannot post the rest of the circuit for IP reasons.

    Thanks!

  • Hi Justin, 

    Thanks for providing the updated test results. Due to the size of the Cff, it would difficult to measure without bringing in additional capacitance and the RC constant forming by the top resistor and the Cff may be too small for you to capture. For most application, we would suggest having a 1nF to 100nF Cff, but some application also recommends having 100pF; for this part, we don't have an equation to calculate the optimal size of the Cff. I would recommend trying to reduce the Cff to 1uF and repeat the test, and if there is an issue, try to reduce it to 100pF. The Cff will help with AC performance, like noise and PSRR, and it is really up to your application to decide if you really need it. 

    I would be really cautious to recommend the solution with the protected diode for the reference if you already experience an issue with the Cff; the diode may not be fast enough to protect the reference and we don't have a lot of experience with different components as our focus would only be on our products. 

    It's definitely understandable you can not share certain things for IP reasons; I hope this reply provides enough information for you to solve your issue, if not, please let me know. 

    Regards, 
    Jason