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.

  • Resolved

TPA3255: Better LC output filter?

Intellectual 370 points

Replies: 10

Views: 254

Part Number: TPA3255

Has anyone tried a 4th-order LC filter for the output of this amplifier?  (4th-order as in: inductor, capacitor, inductor, capacitor)

The switching frequency is 600kHz;  a 2nd-order LC filter with cutoff frequency around 60kHz to 80kHz produces a very high output ripple.   Though this is not audible, I worry that it may not play nice with the magnetic circuit of the speaker, possibly causing hysteresis distortion, leading to intermodulation distortion and other horribillities.  (notice that I'm using an active crossover, so each amplifier output connects directly to an individual speaker/driver)

A quick LTSpice simulation with L1=6.8uH in series, followed by 1uF to GND, followed by 15uH in series, followed by 0.47uF to GND, with the load being 4-ohm, shows an 80kHz cutoff frequency, and approx. 80dB attenuation at 600kHz --- I will definitely sleep better knowing that there's only  a few mV ripple at the audio output !!

Any comments?  Any experience with trying this?

If I decide to try it, should the snubber circuit 1nF || {10nF - 3R3} be after the whole LC network, right before the speaker?

Thanks,
Carlos
--

  • Hi Carlos,

    I have rarely seen 4th order LC filters being used. This is likely due to the costs associated with it, i.e. twice as many components to the filter.

    This is not to say that you can't do it. As you mentioned, there are benefits to using a fourth order filter! I wouldn't see any issue using a fourth order filter as long as you are choosing the right cutoff frequency and overshoot.

    A snubber circuit isn't necessary, depending on the application and desired performance so you're specs are what decide the placement of the snubber network.

    Best Regards,
    Robert Clifton
  • In reply to Robert Clifton56:

    Robert Clifton56
    A snubber circuit isn't necessary, depending on the application and desired performance so you're specs are what decide the placement of the snubber network.

    Thanks Robert, for your reply.

    Could you explain (or point me to some literature / application notes) the purpose/rationale of the snubber circuit in this case?

    As far as I can see, the RC part has cutoff frequency at around 5MHz, and the 1nF is an open circuit below 50MHz or even more.

    If I understood what its role is, then I would be better equipped to decide if it should be present, and where.

    Thanks again!
    Carlos
    --

  • In reply to Carlos Moreno1:

    Hi Carlos,

    We do! Try the Class-D Output Snubber Design Guide

    Best Regards,

    Robert Clifton

  • In reply to Robert Clifton56:

    Thanks again Robert!

    I now have a new concern --- the snubber circuit, as shown in the TPA3255 datasheet (in all typical application schematics) contradicts the Snubber Design Guide (SLOA201).

    If I'm understanding correctly, the snubber circuit should go directly at the amplifier's output pins, as close as possible to avoid the added trace inductance (this is consistent with the remark in section 2 of SLOA201: "The overshoot and ringing at the output are present in the feedback signal to the amplifier" --- the outputs connect through the 33nF capacitors back to the chip, to the BST_x pins).

    However, in the TPA3255, the snubber circuit is placed *after* the LC low-pass filter, which I suppose would render it useless, no?

    EDIT / ADDED: In the TPA3251 evaluation board, the schematic shows snubber circuits with a Do-Not-Populate label  (1nF in parallel with {1uF in series with 3.3Ω} --- 1uF???)

    Also, the snubber circuit has a 1nF capacitor in parallel with {10nF in series with 3.3Ω} --- the Snubber Design Guide prescribes just a capacitor in series with a resistor (between the output pin and GND).

    Could someone shed some light on this?

    One last comment/question: the Snubber Design Guide says nothing about the type of capacitor. I used film capacitors in my current design; but I wonder whether NP0 ceramics might perform better?

    Thanks,
    Carlos
    --

  • In reply to Carlos Moreno1:

    Hi Carlos,

    I'm glad you are doing your research.  Let's try to break it down.

    What you see on the TPA3255 datasheet is actually a zobel network rather than a snubber network that you are thinking of.  Both are very similar so I see where the confusion is coming from.  These networks have different purposes.

    We used placeholder values for the Evaluation board schematics since depending on the customer's configuration, it would require different values.

    We wanted to provide a way for customers to be able to have some sort of option for any feedback that they would want to possibly implement. 

    I have only really come across ceramic capacitors being used in the snubber network. 

    Hope that clarifies things.

    Best Regards,

    Robert Clifton

  • In reply to Robert Clifton56:

    is actually a zobel network rather than a snubber network that you are thinking of

    Huh --- that's curious;  the Zobel network is speaker-specific, and not amplifier-specific;  it also (from what I read) tends to have much larger capacitance values, and resistors with somewhat high power rating.

    In any case, this clarifies my doubts, so now I'm better equipped to make a decision about what my design needs.

    Best regards, and thanks again for your help!
    Carlos
    --

  • Where did you get a spice model of the TPA3255?
  • In reply to Daniel Zinner:

    Hi Daniel,

    There is no Spice model for the TPA3255.

    Regards,
    Robert Clifton
  • In reply to Robert Clifton56:

    Do you know if there is a Class-D model similar to the TPA3255?
    I just want to test different configurations in spice with regards to reducing the ripples on the output.
    I am using the design in a none Audio Application where I need a fixed frequency and need to reduce noise on the output.
    Is TI planning on making a spice model?

    Kind Regards

    Daniel
  • In reply to Daniel Zinner:

    Hi Daniel,

    Unfortunately all of the devices similar to the TPA3255 don't have a Spice model.

    Regards,
    Robert Clifton

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.