TAS2563: Distorted sounds for a few seconds

Hi Tal, 

it could be many things, can you try to share some description, or even better a recording?

Regards,
Arthur

Yes, you can sound something metallic in the sound

I'm not sure I managed to upload the file

Hi Tal, 

I cant seem to play the recordings. I sent you a friend request, perhaps we can share it easier over email

Can you also share the .ppc3 file and any "end system integration" files you generated or may be using?

Are you using the driver files exported from the end system integration panel in PPC3? or is this on the EVM.

Regards,

Arthur

Hey, I got your friend request and sent you the files in a private message

Hi Tal,

thank you, I got the recording via private message.It sounds like some incorrect TDM clock settings to me.  I also check on the EVM with your PPC3 file and I do not find any distortion. 

Is the distorted recording from the EVM using USB, or is it from the end system using the files exported from ppc3? it sounds like there may be an incorrect clock setting

What is the sample rate and BCLK/FSYNC ratio being used? 

Regards,
Arthur

From the end system. The ratio is 64.

I should note that this bug occurs only sometimes. Most of the time the audio sounds good.

I've sent you another private message with MP3 example in case it might help.

Hi Tal, to me it sounds as though the distortion happens anytime audio is played. is it not the case?

Can you please try to disable all DSP processing (ROM mode) this can be selected during the exportation process in PPC3. it can give us a hint if it is something wrong inside the DSP or if there is some issue in the clocks like i suspect.

Regards,
Arthur

So the samples I've sent you are recordings of only the parts where the occurs. Most of the time it sounds good.

I'll try and display dsp processing and will send you a sample.

Hi Tal, 

thank you for testing it will be key in narrowing down the root cause.

Regards,
Arthur

Hey Arthur,

We tried that. As you can see, I've sent you via private messages three files:

First is a recording of sinus wave when there is no bug. As I explained, this bug occurs only sometimes, and this is a recording of the sinus wave playing when it the bug is not there.

Second is the same sinus wave when the bug do occurs. As you can hear, it does sounds distorted.

The third one in an image of the channels. As you can see, it looks like only the right amp is causing problems.

Tal

Hi Tal, 

just to clarify, you have changed the device configuration to ROM mode and you are still finding the distortion? 

Since we are not finding any distortion on the EVM, and we are still finding distortion when the DSP is bypassed (ROM Mode) it leads me to believe that the I2S/TDM clock settings may be getting set incorrectly from PPC3.   

Can you share more details about the I2S TDM clocks in the system. the distortion to me sounds very similar to an incorrect clock.

Can you measure from your system the following

-FSYNC frequency

-SBCLK frequency

Since the ppc3 file is selected as 44.1/48 with a 64 ratio. it can support the following

48kHz/ 3.072MHz or 44.1kHz/2.822MHz. can you check if something different is happening in your end system?

Regards,
Arthur

Yes, we did changed the configuration as you mentioned. The clocks are 48kHz / 3.072MHz. And the bug is not always there. Most of the time the audio works fine. It's occurring only sometimes

I'm sending you how the sinus wave sounds on our preferences with the DSP on when the bug occurs, just for comparison. You can see from previous private messages that while using the ROM option the TDM channel of the right amp was wrong, having the DSP working makes the TDM channel of both amps looks OK, and yet the wave is still distorted

Hi Tal, 

This "TDM Channel of the right amp was wrong" what is the TDM? is it a audio loop back from the device? or is it audio data being sent from SoC->TAS2563?

Regards,
Arthur

The images are the EC output of the amps, I'm guessing this is what you are referring as the loopback. Looking at the images you can see that when bug occurs in ROM state one of the amps channels looks distorted and the other isn't. On DSP state, both channels looks good. Either way, in both cases, ROM or DSP, you can see here that the bug can still be reproduced, and as you can hear in the recordings. I will add that while using our DSP configurations the bug is less frequent

Hi Tal, 

understand that the issue of not dependent on ROM/DSP. And that the clock settings are OK.

I will discuss with my colleagues on this. My original thought is that it could be either clock related or DSP related. It seems it may be something else. 

regards,

arthur 

Hi Arthur, I worked on this project when Ivan Salazar helped us configure the forward and EC paths for I2S.  Did you get feedback about what the failure mechanism might be?  For reference, the same clocks drive both amplifiers' BCLK and FSYNC in our system.  Regards, O.R.

Hi Oren, Tal

Thank you for the additional information. I brought this to Ivan and we discussed. 

What he would like to try is to use your systems I2S clocks to drive the EVM. if the distortion is still present in this case we can know if the issue is in the clocks or it is an issue with the tuning. 

Is it possible to wire out the I2S clocks and data (FSYNC, SBCLK, SDIN, SDOUT) from your hardware to the TAS2563 EVM?

-Set J13 "VCC-I2S Select" to match the IO level of your system.

-Remove J4 "ASI1 SEL" to take I2S input on the J16 header

-using USB from PPC3 write register book 0 page 0 register 0x02 = 0x00. this will place the device in active mode clocks auto detect.

- and then play audio from your system and listen for the distortion on the speaker.

Regards,
Arthur