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.

TAS2563: Use of PDM mic in ROM mode

Part Number: TAS2563

Dear,

I am using the TAS2563 (QFN) icw a 3W 5cm speaker mainly for voice signals (application is comparable with a full-duplex speakerphone). Therefore the smart amplifier and speaker protection algorithms do not add much added value for my design. Also I would like to avoid device per device calibration during the production of a device as I read this is required for the temperature protection of the speaker. But the main issue is that my echo cancellation algorithm running on my SoC does not foresee in an echo signal input, so it can't handle non-linear processing of the signal. So therefore I would like to use the ROM mode. However, I cannot find how to define the TX slots for the DMIC signal in the ROM mode configuration. Can you please assist on resolving this problem?

Thanks in advance,

Sander Vandekerckhove

  • Hi Sander,

    the PDM mic is a DSP feature and will require the device to be in tuning mode(not ROM mode).

    You can avoid the attenuation due to the speaker excursion protection by setting the Xmax to a very large value, like 10mm, and you can minimize any attenuation from thermal protection by setting the algorithm speaker Re larger than reality, (if your speaker is 4ohm tell the algorithm it is 8ohm) and then set the Tmax to a large value. 

    Regards,
    Arthur

  • Hello Arthur,

    Thank you for the reply. I will try that! Thanks!

    Best Regards,

    Sander Vandekerckhove

  • Hi sanders,

    I will mark the thread as “TI thinks resolved” . If you find any issue you can post back in this thread and it will reopen.

    regards,

    Arthur 

  • Hello Athur,

     I tested with a configuration with the Re of the speaker of 16R (8R speaker is used), temperature limit of 1000° and 10mm excursion limit. With this configuration, I can’t reach the required level in tuning mode on my end device. I did disable the limiter and BOP as well. The Boost voltage is well above the peak signal level so the ‘low’ output level is not due to clipping. I can see that when increasing the input level, the output level does not increase linearly when approaching the maximum output level (meaning also non-linear processing). The input signal (pink noise) is however full scale (peak) and I do have some equalising enabled. Increasing the output level in the 'tuning and audio processing' pane (to 9dB) does not increase the output level. With a 1kHz sine wave, the output level stops increasing from around -15dBFS onwards (3Vrms output, boost converter set to 8.5V). What could still be the problem?

    Thank you for your response!

    Sander Vandekerckhove

  • Hi Sander, 

    I tested today and could only achieve about 7.5Vpeak by increasing the protection headroom. I will discuss with our algorithm team and get back to you next week. 

    Regards,
    Arthur

  • Hi Sander,

    There are some intermediate thresholds in the thermal protection algorithm that cannot be bypassed. We wont be able to bypass the protection completely except by the method I mentioned previously. 

    In addition, you can Increase the Rtv,Rtm,Rtva thermal model parameters to something very large, like 1000K/W. this should allow larger steady state output voltage.

    Regards,
    Arthur

  • Hello Arthur,

    I have tried increasing the thermal model parameters to 1000K/W, but this looks like having the adverse effect; the output level decreases.

    I have been experimenting a bit further with the TAS2563 eval board because I thought to remember that I was able to reach higher output levels when playing with the eval board. I found that the low output level is due to the limiter, which was set to 5.2V. Although I am quite sure I disabled the limiter on my previous measurements with my device. I assumed this was an RMS value with some attack and release rates, but apparently the limit level is not the RMS output level but more like the peak output level. RMS level at which the output is limited to is 3.2V. The rms value of a sine with 5.2Vpeak is 3.6Vrms.

    Can you give some more info about this limiter function?

     

    With the limiter and BOP disabled, when I make a bin file with this same configuration and use that on my end device, I see that the output level is quite a bit lower than that of the eval board, with the same input signal. With a pink noise file with peak amplitude of 0dBFS, the output of the eval board is 1.35Vrms and on my device it is only 850mVrms. If I do an amplitude sweep, it looks like there still is some kind of limiting from -2dBFS to 0dBFS on my device, not on the eval board. When I change the gain in the tuning parameter window from 6dB to 12dB, the output level changes on the eval board to app. 2Vrms and to only 1V on my device. I can see limiting on the eval board from around -2dBFS (could be explained from the high peak level, reaching the boost voltage so clipping is present), but on my device from around -8dBFS ( when only reaching 4Vpeak at 0dBFS).

    With a 1kHz -20dBFS sine, the level of the eval board is app. 1.2Vrms compared to 650mVrms on my device. Another thing is, I see that the pilot tone of 60Hz is only 25mVrms on my device, where it is 45mV with the eval board.

    How could this difference be explained?

    Best Regards,

    Sander Vandekerckhove

  • Hi Sander,

    Arthur is out of office and he will get back to your question next week. We will check internally to see if we can address your question in the meantime.

    Thank you,
    Jeff McPherson

  • Hi Sander,

    I made a logical error in my previous message. setting the thermal parameters to a large value is incorrect, this will decrease the output. try instead to decrease them to a small value. 

    If you are finding the limiter is affecting the output, as you stated, you can bypass it, If you want to read up further on it it is discussed in section 8.4.3.6 of the datasheet. this would be able to explain the feature better than I can. 

    are there any hardware differences between the Eval board and your system? is the supply voltage the same? is the speaker the same? is the boost voltage the same?

    Regards,
    Arthur