• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 47 subscribers
  • Views 729 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

TLV3402 latch-up

Alexander Scheuermann
Genius 14890 points
Other Parts Discussed in Thread: TLV3402, TLV3401

Dynamics Inc. is having an issue with the TLV3402 output going high and then never changing states again even when the input should cause it to go low. They are unwilling to give a schematic but they did give me a description of what was happening. Their comments are in blue:

"I’m wondering this because we’ve been seeing a situation where it appears that when the comparator turns on and the output goes high, then it never changes state even though the inputs to it should cause the output to be low.  It’s a very rare situation, but I happen to have a couple of scopes hooked up and monitoring different points on the comparator as well as other points on the card when I saw it happen.  Is there some case that could cause this chip to latch up?"

I asked him for a description of how the device is being used:

"Basically I have the comparator voltage at 2.7V and I have the output voltage pulled up to roughly 3.5V.  Its connected to 4.1V through a diode and a 100k in series.  Can you double check if there are protection diodes on the output and that we can only go to approximately 0.5V above Vcc?  I thought one of the benefits of an open drain output on a comparator was that you could use a different voltage there instead of needing to level shift?  Also, the comparator doesn’t normally stay on so we’re constantly turning it off and on.  Normally it appears to work without a problem, but once in a while when we’re turning it on and off faster than normal it looks as though the output goes high on start-up and stays high with the inputs set so that the output should be low."

I  talked with Thomas Kuehl about the protection diodes, and we decided that they wouldn't be causing an issue like this. Gave that feedback to the customer and got this reply:

"It looks like the latching up may have something to do with timing.  When I watch the two inputs and the Vcc of the chip rise it appears that they rise from 0V to their respective voltages in about the same time.  Vcc is rising to 2.7V in about the same time that the plus and minus inputs rise to their voltages, which are about 2V and 1.5V.  The latch-up problem seems to become less frequent if I ground the output of the comparator for approx. a 1ms.  Since this seemed to improve things I also decided to ground (the input is not actually at ground, but maybe it’s held to approx. 0.5V) one of the inputs to the comparator as well as the output.  If I ground the output and the an input for approximately 500us when powering up the chip it appears to get rid of the latch up problem.  Is there a spec. for how long I would need to wait to apply voltages to the inputs after power for the chip has come up?  Again, this doesn’t happen every time, just every so often."

 

I was hoping you could help me answer the last question specifically and any additional assistance in determining the cause of this problem would be great as well. Thanks!

  • 0
    TI__Guru* 93105 points

    Alexander,

    I tested a TLV3401 (the single version of the TLV3402) comparator on the bench with VCC set to +2.7 V and a 100k + diode to +4.1 V. The non-inverting input was set to +1.5 V via a resistive divider connected to the VCC supply. The inverting input was driven by a generator set to VCC/2 with a square wave riding on the dc level. It operated flawlessly. The arrangement was used to verify that it is acceptable to bias the TLV3402 open-drain at a higher level than VCC and operation was as expected.

    All analog circuits do require some amount of time to become fully operational after power is applied. That time is completely dependent on the internal circuit design and often isn't characterized. The amount of time can be anywhere from microseconds to tens-of-milliseconds depending on the device. When I increased the square wave frequency in the circuit mentioned above I did reach a point at about 33 kHz where the output stayed high. The output transistor could was off at that point. I suspect that is a different condition than what the user is observing.

    When the user is making the start-up measurement is the bias for each of the inputs, the VCC pin and the open-drain pull-up being switched on and off together? I need to know what is, or isn't, being switched during the test.

    Regards, Thomas

    PA - Linear Applications Engineering

  • 0 in reply to Thomas Kuehl
    TI__Genius 14890 points

    Customer's response:

     

    To answer the question below.  In the initial setup nothing was being pulled low when I powered up the comparator.  I just let them come up naturally.  This is where we noticed the latch up conditions.  When I pulled the output low during power up I noticed less latching up.  When I pull the output low as well as one or both inputs low (the (+) actually sits at 500mV when I talk about it being pulled low) while powering up Vcc, I don’t appear to see the latch-up conditions.  The voltage on the (+) pin is static, while the voltage on the (-) pin is the signal.  The signal being sent into the (-) input is at roughly 2V, but it is actually a small 500kHz signal, roughly 80mV pk-pk.  Right now I have roughly 70mV of hysteresis, but I experienced this same condition before when I had roughly 190mV of hysteresis.  

     

    When I read what you sent to me before I thought maybe when powering up the device the two input signals cross as they rise to their respective voltages.  From what I can tell the frequency would be too high to cause the comparator’s output to change, but wouldn’t cause its output to latch-up.  To test this scenario I made a small 80mV pk-pk square wave (I also tested a triangle wave) and started its offset lower than the voltage on the (+) pin.  Then I raised the offset to allow the signal to cross the (+) pin voltage and it worked as I would expect.  I also set the signal so that it was directly centered on the (+) pin voltage and the output of the comparator didn’t change, but as soon as I increased or decreased the offset of the signal then the comparator’s output responded as I would expect.  The signals I’ve seen with the latch up condition, the two inputs are clearly not close to each other.

     

    I hope this is helping.  I’m doing my best to give you information that you need, but without giving more than the my company would like to give out.  It’s a fine line.

  • 0 in reply to Alexander Scheuermann
    TI__Genius 14890 points

    Thomas,

     

    Any comments on this?

  • 0 in reply to Alexander Scheuermann
    TI__Guru* 93105 points

    Hi Alexander,

    I was traveling this past week and I am still catching up on email.

    It is difficult to know what the problem could be without having all the circuit details. They mention the inputs being about 500 mV apart and coming up with Vcc, but then mention a 500 kHz, 80 mVp-p signal on the inverting input. The TLV3401 is a nanopower comparator and is really only intended for switching frequencies <10 kHz, and I am not sure what applying a 500 kHz signal to an input does to the device's operation.

    Is there any way we can work towards getting the schematic and details of this application? We need to see the circuit details, have information about the supply start-up time, how the output/input is being grounded for milliseconds and other details.

    We can take this off the external forum and handle it in a more confidential manner. Also, check and see if we have an NDA in place with the customer.

    Regards, Thomas

    PA - Linear Applications Engineering

     

     

     

  • 0 in reply to Thomas Kuehl
    TI__Genius 14890 points

    Thomas,

     

    The customer has been sharing only what he could under an NDA. So he will not be able to tell us very much more. He said that he is not seeing the latch-up anymore after doing some changes based on your suggestions, so I've marked this as answered. If he comes back with more issues I'll let you know. Thanks a lot for your help!

  • 0 in reply to Alexander Scheuermann
    TI__Guru* 93105 points

    Hi Alexander,

    Thank you for your feedback regarding the TLV3402 resolution.

    Regards, Thomas

    PA - Linear Applications Engineering