Because of the holidays, TI E2E™ design support forum responses will be delayed from Dec. 25 through Jan. 2. Thank you for your patience.

  • State Resolved
  • Locked Locked
  • Replies 12 replies
  • Subscribers 49 subscribers
  • Views 2314 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.

OPA2365 oscillation at 8.9MHz

David Fishman
Intellectual 475 points
Other Parts Discussed in Thread: OPA2365, OPA2350, OPA365-EP, OPA365, TINA-TI, OPA320, OPA369, OPA364, OPA350

Hi,

I have a TIA design that uses a OPA2365. The schematic is shown below.

The TIA function performs well over the bandwidth of interest, 1-500kHz.

However, there is a -62dbm 8.9MHz oscillation peak (25dB above the spectrum analyzer noise floor).

The bottom layer of the 2-layer PWB is ground.

Another board was tested with identical results.

Adding capacitance to the bypass caps have no effect. The only thing that removes the oscillation is placing at least 100pF across the 60k feedback resistor.

I replaced the OPA2365 with a OPA2350 and the problem goes away. However, I would like to use the OPA365-EP for a COTS application.

Thanks,

Dave

 

The photo-diode is a EPM605 from JDSU with a body capacitance of 0.5pF.

  • 0
    TI__Guru* 94285 points
    Dave,


    I've moved this post from the Precision Data Converter Forum, to the Precision Amplifiers Forum. I think you'll find an answer here.

    If you want you can search for OPA2365 (or OPA365) and oscillation, and you might find some answers, otherwise, an apps engineer will likely jump on this question soon.


    Joseph Wu
  • 0
    TI__Guru 62915 points

    Hi Dave,

    Can you post your schematic so we can review it?  All transimpedance amplifier circuits require a feedback capacitor to cancel the 1/Beta zero formed from the op-amp + diode input capacitance and the transimpedance feedback resistor.  This is covered in detail in the following blogs:

  • 0 in reply to Collin Wells
    Intellectual 475 points

    I guess it didn't "stick" the first time.

  • 0 in reply to David Fishman
    Intellectual 475 points

    Still having a problem with uploading a pdf.

    Trying again.

    SCHEMATIC1 _ Tunable Filter Tes.pdf

  • 0 in reply to David Fishman
    TI__Guru 62915 points

    Hi David,

    I looked through your circuit and there will definitely be a stability issue if a feedback capacitor is not placed across the 60k feedback resistor of the OPA365. This is shown in the simulation results below.  Notice that with the feedback capacitor set to 0pF, the transient results show heavy overshoot and ringing and the phase margin is only 3 degrees.  Adding a 100pF capacitor improves phase margin to >70degrees and as expected the transient results show a nicely damped response.  The simulation files below can be used to perform simulations on this circuit using the free TINA-TI circuit simulation software.  

    Transient results.TSC

    Open-Loop_Results.TSC

     

  • 0 in reply to Collin Wells
    Intellectual 475 points
    Thank you Collin.
    The simulation results agree with the lab measurements when placing 100pF across the 60k, however the simulations show a >50 degree phase margin with a 10pF cap. Nevertheless the oscillation at 8.9MHz does not change.
    Thanks,
    Dave
  • 0 in reply to David Fishman
    TI__Guru 62915 points

    Hi David,


    Thanks for the update.  I didn't mention it in my last post because the circuit appeared to be unstable, but the other aspect of this design is that the OPA365 includes an internal charge-pump to achieve the rail-to-rail input stage.  This is described on page 9 of the product datasheet and I've included a simplified schematic from the datasheet below.  The charge-pump operates between 8-12MHz which is possibly what you're seeing.  Generally the charge-pump ripple is below the broadband noise of the amplifier so it is hard to view in the transient (o-scope) domain, but in a spectral response evidence of the charge-pump will show up.

    When you add the larger feedback capacitor you also limit the closed-loop bandwidth of the circuit providing a filtering effect.  Perhaps the 100pF/60k combination reduces the charge-pump ripple below a level that you can identify it at.

  • 0 in reply to Collin Wells
    Intellectual 475 points

    Hi Collin,

    My application bandwidth is only 300kHz, so I can adequately filter the charge pump ripple.

    However, I'm surprised by the revelation that a 50MHZ GBW OpAmp has a charge pump with a ripple in the pass-band.

    Supposed someone wanted to implement a narrow band-pass filter application between 8-12MHz. 

    Is there anything in the datasheet that indicates that the charge pump ripple is <50MHz?

    Thanks,

    Dave

  • 0 in reply to David Fishman
    TI__Guru 62915 points

    Hi David,

    If a bandwidth limit at 300kHz is not sufficient to remove the ripple, then consider using the OPA320 devise.  These are the next generation products in the zero-crossover distortion line of products (OPA364, OPA365, OPA369, OPA320/322) which feature technologies that further reduce the charge-pump ripple.  The ripple-frequency is still near 10MHz +/-20% for that product.

    The charge-pump ripple is not an issue in most applications because the ripple amplitude of the newer OPA365 and OPA32x devices is smaller than the unfiltered broadband noise of the device, so without filtering (averaging) or spectral analysis, most people don't even realize it's there.  Furthermore, it only takes a closed-loop gain of ~5V/V to limit the closed-loop bandwidth of these devices down to the frequency of the charge-pump ripple, which will then start to attenuate it. Therefore, applications with higher levels of gain also usually do not notice the ripple.

    We've had customers try to use these products in the exact type of applications you suggested and have had to recommend other products that don't feature the charge-pump such as the OPA34x or OPA35x products. 

  • 0 in reply to Collin Wells
    Intellectual 475 points

    Thanks Collin.

    The attraction to the OPA365-EP was the Enhanced Product status and the RRIO high unity GBW specs. 

    The OPA320 looks a little limited for 40dB gain at 200kHz (I misspoke when I said 300kHz earlier.)

    The choices are not that many for EP with those specs.

    thanks,

    dave

  • 0 in reply to David Fishman
    TI__Guru 64140 points
    Dave,
    As Collin suggested, please consider OPA350 with GBP of 38MHz which means 380kHz usable bandwidth for the gain of 100. Similarly to OPA365 it can swing within 10mV to either rail with 10kohm load but it does not use charge pump in its front-end and it can directly drive tens of uF loads.
  • 0 in reply to Marek Lis
    Intellectual 475 points
    Thank you Marek.
    Is the OPA350 available as an Enhanced Product?
    dave