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.

LMV822-N power supply rise-time causing latch-up

Other Parts Discussed in Thread: LMV822-N, OPA2377, LMV652, OPA2316, LM6142, TLV2422

Hi,

I'm having an issue with the LMV822MMX/NOPB (formerly Nat Semi version) where the output will latch up to about 1.8V until power is removed. This will happen if the 5V supply rise-time is less than about 3-4us.  There seems to be some variation from part to part.

The LMV822DGKR (which has gone EOL) won't latch up in the same circuit with supply rise-times down to 0.86us. I believe that I have ruled out most other possible causes such as inputs being outside the supply rails and insufficient decoupling.

For our application there is room to slow down the rise time to about 5us. However, the supply rise-time or dv/dt parameter is not specified and I presume that there will be some process variation given the variation I have seen - but how much variation?  So it is difficult to know what a reasonable rise-time is for this part that will reliably prevent the latch-up condition.

Can anyone please recommend a good approach here or offer some information or rule-of-thumb on what is an acceptable supply rise-time for this part?  Our only other option is to switch to a different op-amp.

Regards

Robert

  • Hello Robert,

    Very fast rise times on the supplies can trigger the ESD protection cell across the supply pins.

    Fast slews above 1V/us may cause the structure to trigger. When it does trigger, it shorts the supplies for about 5us with a few ohms then releases (for the ex-NSC devices). It is designed to take the brunt of an ESD event.

    Usually this is faster then any normal power supply can rise, but can cause issues when the device is switched from a fast, low-Rds on electronic switch (or a gate output). Usually the regulator output resistance and the bulk/distributed  bypass caps on the supply line slow the edge rate down.

    Under the hood, the TI ('RAC' LMV822) and the National ('V822' LMV822-N) are different devices built on different processes with different structures (and by different companies at the time). Only the specs are identical. So they will not act the same under extreme conditions. Two different animals...

    Is it a continuous latch? Or are you just seeing the clamping for a few microseconds? Are you seeing this in your system? Or during reliability testing?

    Are you seeing any "ringing" on the supply line? If it does clamp, a poorly bypassed supply line (or poor bypass grounding) may "ring" when the clamp kicks-in.

    It's not a bug...it's a feature...I would keep the supply slew rate slower than 10V/us.

    Regards,

  • Hi Paul,

    Thank-you very much for your reply.

    The latch is continuous until the power is removed.  I do see something that could be the ESD protection where the o/p is about 2V for 4us.  At a slew rate of about 1.7V/us I see the 2V/4us o/p spike and then the continuous latch-up afterwards at about 1.8V o/p.  Lowering the slew rate (<1.2V/us) stops the latch-up for the opamps I have tested but still the 2V/4us o/p is present.  Also at <1.2V/us the output sometimes spikes for only about 1us presumably when the ESD protection does not kick in at all.

    We are seeing this in our system under normal conditions with a 1.3V/us slew rate in about 1 out of 8 cases.

    There looks to be no ringing or overshoot on the supply as it comes up - the Rds on of the FET is 0.5ohms and it is not being driven very hard with a 10k resistor.

    There are 4 dual opamps in the power domain are bypassed with 10nF per package and there is another 50nF of ceramics totalling 90nF after the FET.  All the 4 opamp positions can exhibit latch-up at different slew rates from 1.2-1.7V/us.

    I will look into bypassing and grounding to see if there is any improvement to be made to handle the ESD protection shorting the supply.

    We may be able to limit the slew rate to 1V/us as you suggest.  What is the risk that process or temperature variation will cause the latch-up to resurface below 1V/us slew rate?  I realise this may be a rather difficult question to answer but would appreciate any further advice you can give.

    Regards

    Robert

  • Perhaps I can re-phrase the question slightly:

    What is the risk that process or temperature variation will cause the ESD protection to engage at 1V/us power supply slew rate?

  • Paul,

    I am having a latch-up issue similar to Robert. We were using the LMV822DGKR with no lock-up issue. When this part went EOL we replaced it with LMV822MM/NOPB. The amp is used in an analog phone circuit. When we plug in the phone line and very high freq ringing couples onto the 5v power supply, which sometimes caused the amp to latch-up. The amp stays in this condition until the power is cycled.

    To verify this problem we took a board with the MM amp that had the latch-up issue and replaced it with the DGKR. We plugged the phone line in numerous times and no latch-up.

    Can you recommend a replacement amp that will not latch-up.
  • Hi Frank,

    Can you elaborate on: "When we plug in the phone line and very high freq ringing couples onto the 5v power supply".

    Does the supply peak above 5.5V? What is the "high frequency ringing"? Is this the ~20Hz phone line ring freq coming through? Any scope photos you can share? Can you share a schematic?

    The LMV82x is part of the original core family, so some of the LMV32x cores will be similar (assuming it is a core issue).

    Try a LMV652, OPA2377 or OPA2316. The OPA's are completely different beasts and the 652 is a different core.

    Regards,
  • Paul,

    Thanks for the quick reply.  Sorry there about the typo, what I meant was:

    When the phone line is plugged into the phone jack on the circuit board, a high freq voltage couples onto the system 5v.  (attached pic)

    Channel 1 is the amplifier output 2v/div.  Normally it is at 1.5vdc.  

    Channel 2 is the 5v supply to the board.  The ringing voltage occurs when the telephone line is plugged into the phone jack.

    I have seen the amplifier output latch to the 5v rail as well as slightly above ground.  When it latches to 5v the power to the board has to be cycled to get it out of this state.  When latched to gnd, the amplifier will recover and operate normally.

  • Hi Frank,

    The Abs Max supply voltage of the LMV822 is 5.5V.

    The transient looks to be peaking at 9V and stays well above 6V for several cycles - this is grossly violating the Absolute Maximum spec and will eventually cause damage..

    You are seeing a classic latch. Even violating the Abs Max for a nanosecond can cause damage. I am not surprised you are seeing an issue.

    You really need to tame the supply transient. Is this a linear regulator or switcher? If so, the regulator seems a little unstable.

    Do you have adequate bulk supply bypassing? Are you using an inductor for supply filtering?

    In the mean time, try adding a 5.1V Zener across the power supply pins to clamp the voltage and see if the latching goes away.

    If you cannot fix the supply, you should use a higher voltage device, like the LM6142 or TLV2422.

    Regards,