• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 963 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

TLV320AIC3204 distortion

R11
Expert 1485 points
Other Parts Discussed in Thread: TLV320AIC3204

When using the TLV320AIC3204, and if the ADC input is overdriven ( IN1_L/R) > .5VRMS, distortion/noise feeds through to the digital output drive for the speaker.  Is this expected behavior and is there is a way to configure the part to eliminate this effect?

I don't believe I'm exceeding the Absolute Maximum Voltage on those pins from the table in the data sheet but If I am it is a momentary transient.  I think it takes just a little more that .5VRMS.

  • 0
    TI__Guru**** 189401 points

    Hi, R11,

    I have asked my colleague to work with you on this issue, but he is out of the office today. He'll be back on Monday.Sorry for the delay.

    -d2

  • 0
    TI__Mastermind 46195 points

    Hi R11,

    Have you met R2-D2?

    The inputs are DC-biased to 0.9VDC by default. The AVDD supply is 1.8V nominal. This means that there is 0.9 Vpk available of headroom before exceeding the recommended operating conditions. A 500 mVrms input is a 0.707 Vpk signal. A 600 mVrms signal will reach the max allowed signal at the inputs (although the inputs to the ADC would be distorted before that depending on the input settings). Once you exceed the recommended conditions (i.e. 600mVrms), current may start flowing from the input pin into the AVDD supply. This may cause issues with input-output crosstalk in part because the AVDD supply may be affected by this current being injected by the input source through the internal ESD protection circuitry.

    To solve this system-level problem, You can add two Schottky diodes at the inputs. Cathode from signal to AVDD and cathode from ground to signal. Try experimenting using through hole diodes on the EVM.

    Regards,

    J-

     

  • 0 in reply to J-
    TI__Mastermind 46195 points

    Also, you can add some resistors in series (e.g. 1k-ohm) with the input. This, in combination with the diodes should help. The diode protects against signals above abs max, and the resistor helps limit the current into the input pin (and consequently into AVDD).

    Regards,

    J-

  • 0 in reply to J-
    Expert 1485 points

    R2-D2... That's funny! 

    Thanks for your help on this one.

  • 0 in reply to R11
    Expert 1485 points

    J Arbona,

    Just to verify...  This should really say 1 diode -   Anode to signal & cathode to AVDD and the other diode -  Cathode to signal & Anode to GND?  See attached..  Can you confirm?

  • 0 in reply to R11
    TI__Mastermind 46195 points

    Hi R11,

    Yes, you are right about the polarity. Sorry for the confusion.

    You should add series resistors as well between the connector and LIN1_L and also for LIN1_R. 1k is fine. It will help reduce current injected into AVDD and GND and will also help prevent the external music source from seeing a short to AVDD and GND at high amplitudes.

    Please let us know how this works. I would also advise to modifiy the current PCB with these modifications to double check that it indeed solves the problem before fabbing the new PCB.

    Regards,

    J-

  • 0 in reply to J-
    Expert 1485 points

    J, 

    Here is the feedback I received after testing was completed:

    Using this approach seems to only give about 1.4 dB better margin than an unclamped approach.  I see feedthrough at +1.87 volt clamp level, which is about as good as you would expect with a schottky clamp approach.  Perhaps I need to design a clamp circuit that assures that the signal does not exceed .6Vrms.  I could use AVDD as a reference for the clamp circuit.    It also appears that this is only an issue when the signal is overdriven on the IN1_L since LOL is the only line used to drive a speaker.  Does that make sense?  I don't seem to see a feedthrough problem from IN1_R to LOL.

    If I do not clamp the other channels, and they are overdriven, will I cause damage to the part?  Max signal wold be 0 - + 3.3 volts.  

    Even though I still hear some feedthrough, it is considerably quieter than the unclamped approach.  So I'm wondering if It will be acceptable to just clamp LIN_L and let the other channels see +3.3 peak?  This may be acceptable.

    Summary:

    • Clamps do not eliminate the problem, but the volume of the feedthrough to the speaker may be acceptable now
    • I only measure feedthrough from IN1_L to LOL.  IN1_R does not feedthrough to LOL.
    • Is it acceptable to just apply a clamp to IN1_L.  IN1_R will see 3.3 volts peak maximum.  Is there any risk of damaging the part if IN1_R is unclamped?
  • 0 in reply to R11
    TI__Mastermind 46195 points

    Hi R11,

    Thank you very much for the feedback!

    You should only provide voltages that are at or below the recommended operating conditions. Exceeding the absolute maximum ratings may damage the part. For INx inputs, 3.3V is too high. If your product specifes a higher range, you can use a resistor divider network (as explained in http://e2e.ti.com/support/data_converters/audio_converters/w/design_notes/2764.how-to-calculate-the-input-resistance-of-an-aic-device.aspx).

    Did you also experiment with a series resistor between the audio connector and the Schottky diodes? This should reduce the input to output crosstalk.

    Regards,

    J-