• TI Thinks Resolved

OPA2322: Output Ripple issue

Part Number: OPA2322


We used OPA2322 for Scanner application detect paper function.

We get problem about OPA2322 U3A pin 1 output ripple too big. Some OPA2322 is no problem. But some OPA2322 is problematic.

I think maybe is design issues. Please kindly reference on below information and need your help to provide suggestions to us. Thanks~

All schematic

OP 1

OP 2

OP 3

Waveform report

Daifuku ultrasonic Fail report_053018.xlsx

  • Hi Aj,

    what do we see on the scope plots? Is it uncorrelated noise? Or is it the 300kHz signal from the ultra sonic sensor?

    The OPAmps: Is it the OPA2322 or OPA4322?

  • In reply to kai klaas69:

    Dear Kai.

    Thank you for your reply.

    Below is not detect paper and detect paper waveform for you reference.

    I think the noise is not from ultra sonic and we used two OPA2322 not OPA4322. Thanks for kindly help~

    Good OPA2322 no put paper waveform

    Fail OPA2322 no put paper waveform

    Good OPA2322 put paper waveform

    Fail OPA2322 put paper waveform

  • In reply to AJ Cheng:

    Hi Aj,

    I think your circuit is a bit instable. The "Fail OPA2322 no put paper waveform" is typical for an instable OPAmp circuit which mutates to a radio receiver demodulating AM modulated radio broadcast. There are several issues which should be discussed.

    The OPA2322 is a 20MHz OPAmp. It needs to be treated like a HF-OPAmp. So, I would recommed the following points:

    -Paralleling ultra low ESR ceramic decoupling caps can cause nasty resonances. At the resonance frequency the decoupling network is high-impedant and the decoupling performance can be impaired. So, I would use only one decoupling cap per OPAmp.

    -The OPAmps should be "isolated" from each other a bit. Insert a small resistor (10...22R) in each supply line which forms a RC-low pass filter in combination with the decoupling cap. Mount the resistor close to the decoupling cap.

    -All +inputs of OPAmps are biased from the same voltage divider. Here also some isolation should be provided: Connect a 100nF cap from each +input to GND, directly behind the 1k resistor.

    -GND_B and GND_C should be connected to each other at each chip. Is this possible? If not, mount a cap between these pins.

    -The PCB of this circuit should contain a solid ground plane.

    -GND_EARTH should be bypassed to signal ground (GND_B / GND_C) by the help of a cap.

    -The bandpass has a Q of 6. This seems to be small enough. But the phase stability analysis shows that in the feedback loop a phase lag of 78° develops which dramatically erodes the phase margin. I would suggest to use two bandpasses with a Q of 3 each. This would double the phase margin and stabilize the circuit. From the schematic can be seen that originally three bandpasses were intended. So, this change would be very simple.

    Another hint: Do not connect the scope probe directly to the output of OPA2322 but insert a small resistor of 50...100R. Omitting this "isolation" resistor can make the OPA2322 become instable.

  • In reply to kai klaas69:

    Dear Kai.
    Good day.
    I am worried that there will be a cognitive error.
    Can you please kindly provide the reference schematic for us. Thanks for kindly help.
  • In reply to AJ Cheng:

    Hi Aj,

    where exactly do you need help?

  • In reply to kai klaas69:

    Dear Kai.

    I am currently trying other ways now. Thanks for kindly help~