• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 48 subscribers
  • Views 617 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.

Stability problem for non-unity gain stable amplifier. Ask method to improve stability for LM3886

yiting.wang
Prodigy 25 points
Other Parts Discussed in Thread: LM3886

Hi TI

 I only use LM3886 as a non-inverting/inverting amplifier not audio application. Otherwise LM3886 will turn to audio forum.

Here goes the problem:

Due to their special need. The customer need to design LM3886 at non-inverting gain less than 3. Also, The LM3886 needs to output 1A to drive an 8 ohm load

Normally LM3886 is non-invering 10 gain stable in datasheet, and gain 5 is stable in my several PCB tesing. But turns to strong unstable when gain<3

I read carefully the article about some stability compensation method for unity-gain stable amplifier from TI bruce  https://www.edn.com/op-amps-g1-stable-decompensated/   In the website , he talks about this three method are not suitable for solving non-unity gain stable amplifier. But I can't see compensation method for non-unity gain stable amplifier, only increasing the gain

 and post from TI collin   https://e2e.ti.com/blogs_/archives/b/thesignal/posts/taming-oscillations-the-capacitive-load-problem

Previously  and   talk about stability in this thread, which is helpful https://e2e.ti.com/support/amplifiers-group/amplifiers/f/amplifiers-forum/1217871/opa548-whether-opa548-are-suitable-for-driving-inductive-coil

But I think these method are suitable for capacitive load, I know them well, but for non-unity gain stable (decompensated )  such as LM3886, and this 1A output current. I didn't see much info and don't know whether there are other methods rather than increasing gain, would you give some suggestion?

I attach the simulation result in the attachment using parallel capacitor LM3886 stability.pptx

May I ask other method (and some related post), rather than increasing gain for  non-unity gain stable (decompensated)  op-amp ? 

  • +1
    TI__Mastermind 40410 points

    Hi Yiting

       I'm really sorry that, this device assigned to our team due to organization restructure. The Thread relate to LM3886 will always comes back to us. But I'm afraid we has very limited resources. And usually the datasheet mentioned the Gain value can't be set lower than 10, means it's inner control loop can't guarantee stable if too small outer loop be used. The simulation model also can't give full details of the devices.

  • +1 in reply to Shadow He
    TI__Guru 62965 points

    Hello,

    You will have to keep the AC gain above the minimum stable-gain value of 10V/V.  There is no way to get around that requirement. That’s why you use unity-gain stable amplifiers for lower gains.

     The only option is to use a noise-gain circuit to increase the AC gain (noise-gain) to above 10V/V while the lower frequency gain is his desired 3-5 V/V.

     

  • 0 in reply to Shadow He
    Prodigy 25 points

    Hi He

    Thanks for your reply~  I will try on LM3886 and consider the IC you suggested

  • 0 in reply to Collin Wells
    Prodigy 25 points

    Hi Collin 

    Thanks for your reply and your precious ppt post 

    Since some LM3886 PCB has been manufactured.. I try to remedy them..

    Here is the summary of the LM3886 stability.pptx 

    May I ask whether  method3---Cf are suitable for this case (I see in your slides, and it is a little helpful in reality), why you say only method 4 noise gain are are the only choice ?

    Best Regards

    Yiting

    I forget to record the oscillation frequency in my oscilliscope. I think roughly ---- huge oscillation ,hundreds hertz when non-inverting gain =2 and 2MHz~2.4MHz when ltlle oscillation when non-inverting gain =4

          

          

  • +1 in reply to yiting.wang
    TI__Guru 62965 points

    Hello,

    The answer lies in the high-frequency 1/Beta in the "Cf" results.  1/Beta is equal to noise-gain in nearly all cases and you can see the high-frequency noise-gain is unity-gain for this circuit because the Cf capacitor shorts out the feedback resistor. 

  • 0 in reply to Collin Wells
    Prodigy 25 points

    Hi Collin

    Thanks for your reply, now I understood noise gain difference between method 3(cf) and method 4 (noise gain circuite)

    But I am little confused because the application is LM3886 output <10kHz 1A to the 8 ohm load. So the bandwidth of this simple amplifier circuit does not require much. 

    So I think Cf method and Low pass filter or bandpass filter method are suitable for this case, right?  I am still quite confused why only noise gain, method 4 are the only option.

    Because as you can see, Cf method, although 1/β goes to 0 dB, but it increase the phase margin 

    Regards

    Yiting

  • 0 in reply to yiting.wang
    TI__Guru 62965 points

    If you're able to place the Cf pole/zero in exactly the correct location then it will work.  I would caution that I would be hesitant to recommend a "CF" only compensation for an amplifier that is not unity-gain stable. If the circuit open-loop gain or unity-gain bandwidth shifts over process you can easily lose 10-30 degrees of phase-margin.  Anyways, do what you think is best for your application and good luck!