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.

OPA2379: Output oscillation (abnormal operation) at Low temp

Guru 21045 points
Part Number: OPA2379

Hi Team,

 

We are evaluating OPA2379 on the actual board.

(The circuit is the following Figure.1.)

Then, we observed the output oscillation (abnormal operation) at Low temp and the incidence is few.

 

Therefore, we have three questions.

 

---------

[Q1]

We guess that this cause is the individual specificity of specs so OPA2379 don’t have enough phase margin.

Is my understanding correct?

---------

 

[Q2]

We guess that key specs are “Open Loop Voltage GAIN” and “Gain Bandwidth Product”.

Could you please let us know if you have any concern spec?

---------

 

[Q3]

If my understanding of [Q2] is correct, we would like to know vs temp Graph

-Open Loop Voltage GAIN vs temp Graph

-Gain Bandwidth Product vs temp Graph

 

Would you send it if you have this data?

---------

 

[Figure.1 actual board Circuit image]


 

 

Regards,

Hide

  • Former Member
    0 Former Member

    Hello Hide,

    Could you please tell me the temperature range of oscillation, the percentage of units/circuits that had oscillation, and an oscilloscope shot of the oscillation if possible?

    I have simulated your circuit for stability.  Let me explain my findings and then I will try to answer some of your questions.  I ran two different types of stability analyses.  First, I looked into the phase margin of the circuit.  As you can see below, the simulated phase margin is about 53 degrees.  This is not bad, but might not be enough margin if there is a lot of temperature shift for the device and/or the components.

    Then, I ran a stability analysis and looked at the rate of closure.  Rate of closure is a technique we use to gain more information about why a circuit may or may not be unstable.  This is done by looking at the open loop gain curve (AOL) and the feedback factor curve (1/Beta).  For an amplifier to remain stable, these two curves must intersect at a difference of 20dB/decade.  To learn more about this analysis, please see this set of slides: https://training.ti.com/system/files/docs/1332%20-%20Stability%202%20-%20slides.pdf

    From my analysis shown below, we can see that the AOL and 1/Beta curves intersect at a rate of -20dB/decade at about 99kHz.  However, the AOL curve begins to descend to -40dB/decade at around 300 to 400kHz.  Thus, while the circuit is stable, there is not too much margin for a shift in component parameters and/or op amp specifications.  Below are the images from my two simulations.

    Let's now consider your questions:

    [Q1] Oscillation will typically occur when the phase margin of the circuit degrades too much.  This is dependent on several factors.  For one, the amplifier's bandwidth will play a part.  However, the surrounding components including the feedback and load are just as important, if not more important.  With the correct component values, you should be able to stabilize the amplifier.

    [Q2] We typically consider the bandwidth of the amplifier when speaking about stability.  Again, the bandwidth, gain setting, and load of your amplifier will determine the component values that can be used.

    Have you considered the possibility that the component values themselves may be shifting with temperature?  These may be shifting more than the amplifier specifications themselves and this could cause instability at lower temperature.

    [Q3] Let me see if I can find some information on this for you.

    Regards,

    Daniel

  • Hi Daniel-san,

     

    Thank you for the detail information and answer for [Q1] to [Q3].

    Unfortunately, we don’t have an oscilloscope shot and Incidence is 1 / 100 units at around -40℃ to 0℃ so very low incidence.

     

    We confirm a behavior of Fail unit at voltage follower circuit.

    Then, the output(7pin) is oscillation at room temp.

    (Good unit isn’t oscillation at room temp.)

    Therefore, I guess that the internal circuit of this unit has damage or this unit’s bandwidth is lack.

    Could you please let us know if you have any estimation?

     

    Regards,

    Hide

  • Former Member
    0 Former Member in reply to Hide

    Hello Hide-san,

    Thank you for the information.

    When the output pin begins to oscillate, does it oscillate without stopping or does is oscillate for some time and then dampen to a final voltage?  In other words, does the ripple eventually dampen out?

    The part is operating within its specified temperature range and, if no specification is being violated, then it is quite unlikely that the unit is internally damaged.

    My colleague is looking into obtaining the plots you requested in [Q3].  I should have an update for you regarding this matter on Monday.

    Regards,

    Daniel

  • Hi Daniel-san,

    Thank you for always kind support.

    Output voltage oscillate without stopping.

    I’d greatly appreciate your verification.

    Regards,

    Hide

  • Former Member
    0 Former Member in reply to Hide

    Hello Hide-san,

    Here are some follow up comments:

    1.  As temperature drops, the Aol of the parts will increase.  This occurs because as temperature drops carrier mobility increases, transconductance increases, and the part's bandwidth would be expected to increase.  As the part's bandwidth increases, the closed loop bandwidth of the circuit will approach the second pole introduced by the feedback network.  You can see this in my first response where the closed loop response goes from -20dB/dec to -40dB/dec.  Finally, as the closed loop bandwidth approaches the second pole, the phase margin of the circuit will decrease until the circuit becomes unstable.  Unfortunately, the plot data is confidential and cannot be shared on e2e.

    2.  Your circuit design is already a bit too marginal as the second pole is probably too close to the closed-loop bandwidth.  When you take into account the normal +/-15% variation in bandwidth across process variation, potential variation in Aol across temperature, and variation in component values with temperature, these factors may combine to give you problems.  I think the solution to you problem is to increase the margin between your closed loop bandwidth and the second pole.  This can be done by increasing your gain or moving out your second pole.

    3.  Do you know why C1 is present in the circuit?  It will provide a feedback path for the output to reach the positive input.  This can lead to positive feedback and is not desired.

    Regards,

    Daniel

  • Hi Daniel-san,

    Thank you for the detail information.

    I will check C1 and cut and try for stable operation.

     

    I greatly appreciate your cooperation.

     

    Regards,

    Hide

  • Former Member
    0 Former Member in reply to Hide

    Hello HIde-san,

    I hope this information is enough for you to find a solution to your problem.

    Please, let me know if you need any further help.

    Regards,

    Daniel