• State Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 1 answer
  • Subscribers 49 subscribers
  • Views 441 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.

INA321: Input polarity caused a stability problem.

yunsik kim
Prodigy 140 points
Part Number: INA321
Other Parts Discussed in Thread: LMP7721, REF2030, , REF70

Hi, TI.

I designed a trans-impedance amplifier circuit using a LMP7721, INA321 and REF2030.

The INA321 was used to amplify the fluctuating voltage signal based on 1.5 volts.

The 1.5V bias was generated by the REF2030.
The schematic image is below.
The problem occurred when the input polarity was reversed like below.
When C1 was 10pF, both circuits worked well during transient simulations.
However, C1 was below 3pF, the second circuit started oscillating like below.
The first circuit worked well for the lower capacitance of C1.
 
Why do the stability problem arise when polarity is reversed?
  • 0
    Guru 140200 points

    Hi Yunsik,

    even without running a phase stability analysis, I would say that 300pF is a very high detector capacitance and that a 3pF phase-lead capacitance will be way too small to recover the phase margin to a sane value. So both circuits appear to me to be unstable. A 10pF phase-lead capacitance looks way better to me.

    Please carry out a phase stability analysis!

    Kai

  • 0
    Guru 140200 points

    Hhm, I see another issue: Keep in mind that the REF2030 generates a lot of noise! 0.25ppm / SQRT(Hz) equals a spectral noise density of 313nV / SQRT(Hz). This is 40 times the noise level of LMP7721 Scream

    Therefore you would urgently need some noise filtering at the output of REF2030 Relaxed

    Kai

  • 0 in reply to kai klaas69
    Prodigy 140 points

    Hi Kai,

    I carried out a phase stability analysis on the first stage.

    Because the feedback resistor R4 had very large resistance, I guess the stability of first stage is not a problem.

    Or should I try a multiple feedback loop analysis? 

    And, here I have another question.

    I chose the largest resistor to boost the gain.

    However, I found the bandwidth of the TIA was too low ( < 100 Hz).

    Is it correct to simulate as below to check the gain value according to frequency and the bandwidth?
    Thank you.
    Yunsik.
  • 0 in reply to yunsik kim
    TI__Genius 10850 points

    Hi Yunsik,

    Kai is correct, high C in the feedback may lead to gain peaking and stability issues. To learn more about compensating transimpedance amplifiers, please refer to the following note: https://www.ti.com/lit/an/sboa055a/sboa055a.pdf

    The bandwidth of your op amp should be chosen based on the source capacitance and feedback components. Equation 4 in the following article shows how to calculate the minimum bandwidth required for a stable TIA design: https://www.ti.com/lit/an/sboa354/sboa354.pdf 

  • 0 in reply to yunsik kim
    Guru 140200 points

    Hi Yunsik,

    yes, you are right, your circuit is stable Relaxed

    To achieve a bandwidth of 100Hz I would modify your circuit this way:

    yunsik_lmp7721.TSC

    By the way, modify the "Analysis Parameters" this way to get a proper transient analysis:

    Kai

  • 0 in reply to kai klaas69
    Prodigy 140 points

    Hi Kai,

    Please let me know how to convert ppm / SQRT(Hz) to V / SQRT(Hz)?

    I want to redesign the circuit using REF70 of TI or voltage reference from other companies.

     

    Thank you.

  • 0 in reply to yunsik kim
    Guru 140200 points

    Hi Yunsik,

    multiply 0.25ppm / SQRT(Hz) with the output voltage of 1.5V, which gives

    1.5V x 0.25 x 10^-6 / SQRT(Hz) = 375nV / SQRT(Hz).

    See also figure 35 of datasheet of REF2030.

    There's no need to do much redesign of the reference voltage. Just add a heavy RC low pass filter. Something like 1k and 100µF and suited current limiting resistors to protect the inputs of LMP7721 and INA321.

    Kai

  • 0 in reply to kai klaas69
    Prodigy 140 points

    Hi Kai,

    I tried adding an RC low pass filter like below.

    However, the gain curve became weird by unexpected poles and zeros.

    Before adding the RC filter is like this. I see gain peaking Frowning2

    After adding the RC filter is like this. Pole and zero below 1Hz lower the gain.

    I modified the circuit design to handle these problems. How about this?

    The circuit simulation showed a normal gain curve.

    Thank you.

  • +1 in reply to yunsik kim
    Guru 140200 points

    Hi Yunsik,

    I meant it this way:

    yunsik_lmp7721_1.TSC

    Only the +input of LMP7721 and the +input of INA321 should get the filtered bias voltage, but neither the REF pin of INA321 nor the base point of R5 Relaxed

    Kai