• Locked Locked
  • Replies 6 replies
  • Subscribers 64 subscribers
  • Views 1626 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

LM2676 excessive output over-voltage output for certain conditions of startup

Carl Van Wormer
Prodigy 240 points
Other Parts Discussed in Thread: LM2676S-ADJEVM, LM22676, LM2676

Using the LM2676S-ADJEVM board (and 2 circuit boards that I've designed), there is a certain condition that can cause a significant overshoot in the output voltage.  This is caused by the startup circuit inside of the part, and can go higher than 7.7V in a 5V system.  It occurs if the input power is interrupted for a short time, allowing the output to drop below 5V before the input power is re-applied.  The LM22676 does not seem to have this problem.  An easy way to see the problem is to diode-couple a variable supply into the output to keep the output slightly below 5V when the input power is removed.  When the input power is re-applied, the output will overshoot to higher levels while the startup circuit activates.  There is a burst of switching cycles before the feedback circuit detects the output over-voltage condition.  The overshoot is also dependent on the input voltage.  The EVM on my desk is showing an overshoot peak at 7.72V when 17.3V  is applied to the input while the output is at 4.9V.  This problem does not show in my WebBench simulations.  We had many field failures (over several years) until I finally observed this window of failure conditions.  More information can be provided if needed.

good luck!

Carl

  • TI__Guru 57555 points

    Thank you for bringing this to our attention. 

  • in reply to Frank De Stasi
    Prodigy 240 points
    Let me know if you need any more information to duplicate the problem in your environment.
    Thanks,
    Carl
  • Intellectual 380 points
    How excessive are the over voltage events? As a quick fix could a Transorb be used to clip the spikes to something that won't damage the circuit being powered?
  • in reply to Carl Van Wormer
    TI__Intellectual 2790 points
    Hi Carl,

    1. Slow down the Vin transient slew rate will reduce the overshoot. Add a LC filter at the input side.
    2. Is it allowed to shut down the LM2676 when Vin is lower than 5V. If yes, you can set the system UVLO a bit higher than 5V through ON/OFF pin. When Vin drops below 5V, device shut down, and when the Vin recovers, the device should re-start. So should be no overshoot then.

    BR
    Luke
  • in reply to John Jones
    Prodigy 240 points

    The evaluation board (LM2676s-ADJEVM) on my desk shows a 2.7V overshoot on the 5V output.  I have seen transients raising the 5V output up to 8V on some of my circuits with slightly different component values.  Had the design anticipated this behavior, we could have added (at an extra cost) a TVS on the 5V outputs to minimize this problem, but this behavior was not anticipated.  Our solution was to replace the regulator with the LM22676, which does not exhibit this problem.

  • in reply to Luke Pan
    Prodigy 240 points
    The input transient is the automotive 12V power applied to the system. The initial turn on transient is not the problem. The problem occurs when there is an input power drop-out (maybe 100ms), while the 5V drops to less than the regulated output, and then the input is re-applied, allowing a startup burst of switching activity (about 500us) to pump energy into the output caps before the internal control can take over while the output goes over 7V.

    Since this is a design that has been in the field for years, adding circuits to switch the shut-down cannot be considered. Changing to the LM22676 (which has different startup control methods) eliminated the problem.

    Until I was able to determine the root-cause of our problem, the occasional field failures were blamed on many things. Since the newest technology (smallest geometry) was the memory, it would usually act as the fuse for the system. The cratered memory chips prompted our customer to question our memory control timing as the cause. I spent a lot of time not finding timing problems. I placed several systems on a power cycling system (relay on parallel port and BASIC random time cycling program logging delay time and system function) and finally trapped the failure. The setup to show the problem is quite easy. If you are using this part, you should see if your circuit is vulnerable to this problem.