TPA2026D2: Follow up the register reset issue here

Hi Rannie,

TPA2026D2 is an analog input amplifier with I2C control, can we get more details on how Windows is controlling this device? I assume there's a codec or some processor that handles I2C transactions to TPA2026D2.

Please keep the experiments simple, I got more questions to understand the problem:

• Can you test always with a specific signal at a specific amplitude? At some point it was mentioned that "there will be no issues with increasing volume".
  • Is this problem occurring only when signal goes from low volume to high volume?
    
  • Is it different if you play a silence audio file? or a file with a constant amplitude?
  • Is this "volume" the system volume or just the input signal amplitude?
• Is it possible to probe the I2C signals during this test? Make sure there are no additional I2C commands when audio starts.

Comments on each question:

1. Do you have any insights on the root cause?
   1. At this moment the root cause is unknown, there are many factors that we still have to understand as described above.
2. Can you help to reproduce in TI Lab? Customer operation system is Windows.
   1. I can check on TPA2016D2 EVM (similar to device to TPA2026D2. I have requested a sample of this board. However, the I2C controller will not be the same as customer implementation.
3. Do we have such a FW to modify the default register?
   1. Default register changes are complicated, and most likely not available anymore. Usually this involves creating a custom process for the device which seems not feasible for an older device like TPA2026D2.

Best regards,
-Ivan Salazar
Applications Engineer

Hi Ivan,

One more thing to clarify:

“There is no problem with volume increases” means there is no “gradual increase in volume” problem, it does not mean that the problem disappears with volume increases. Sorry for the misunderstanding.

Should customer still implement the test you mentioned above?

How about the EVM results? Why the register set to fault when opening audio file?

BRs,

Rannie

Hi Rannie,

Yes, I think it still helps if customer can test as suggested.

I got the EVM but haven't bring it up, I'll check it out later this week and share any updates after the test.

Best regards,
-Ivan Salazar
Applications Engineer

Hi Ivan,

Thanks. Looking forward to your reply.

BRs,

Rannie

Hi Rannie,

Can you please share the whole register configuration used on customer implementation? I'll try with the same configuration.

By the way, it was mentioned before that register 0x03 is set to 0x00, however I don't think this is correct. The values mentioned in Table 8 on page 27 of data sheet start at 0x01 only; this is the same for ATK_time as well.
Please make sure only supported values are written to the device registers.

Best regards,
-Ivan Salazar
Applications Engineer

Hi Ivan,

The only register customer modified is 0x0B.

Ivan Salazar said:

Please make sure only supported values are written to the device registers.

Got it, I will ask customer to modify it accordingly.

BRs,

Rannie

Hi Rannie,

I assume you mean register 0x03 is set to 0x00 from original value of 0x0b, correct?

On the original thread it was mentioned that limiter was disabled and compression set to 1:1 so I thought there were more register writes. I understand that for now the test is only setting one register.

Please let me know if the result is different by writing a value other than 0x00 into register 0x03.

I'll try in a similar way writing a supported value to some register and check.

Best regards,
-Ivan Salazar
Applications Engineer

Hi Ivan,

Please find below feedback from customer:

1. Currently using 0x01 to fill in the value of offset 0x03. When customer open the audio file, there is also a reset problem.

2.Try to grab a command on I2C when starting audio with a logic analyzer, no I2C command transmission was found, but re-reading the register found that it was reset.

3.Customer cannot find a completely muted audio file. On this side, the volume of the OS (the small speaker in the lower right corner of Windows) is turned off. Click on the audio file to play, there is a reset problem.

BRs,

Rannie

Hi Rannie,

Thanks for running more tests on customer side. It would seem the device is not receiving any different signal through I2C or even at the analog input during this test but somehow is being reset. There must be something else missing here.

I test on my side with a 1kHz sinewave, and the device is still operating normally. I'll test with different signals but since customer already tested with volume muted, I don't expect it to have different results.

Best regards,
-Ivan Salazar
Applications Engineer

Hi Ivan, 

Thanks for your reply. Can you test in Windows, which is the same testing environment with customer?

To configure the TPA2026D2, customer using external I2C configuration tools. But they didn't capture any CMD on I2C line when the register is reset. 

BTW, can you help to elaborate the details of your setup in TI Lab? Such as the VEM setup, the register configuration and the input signal, and how to verify the register is not reset.

BRs,

Rannie

Hi Rannie,

This is an I2C controlled device, the tool I use to generate I2C commands is also based on Windows, although is not the same as the one from customer. I don't think I can use the same as they do.

I'll elaborate a report for the test and share later.

Best regards,
-Ivan Salazar
Applications Engineer

Hi Ivan,

Based on the test up to now, customer proposed following ideas, can we support?

1. Whether there are Lock registers? That is, the register can be locked and not changed anymore.

2. Can BU provide a Driver under the OS to configure these registers? Customer can't configure I2C in OS.

BRs,

Rannie

Hi Rannie,

I'll reach out to you over email to continue these discussions.

Best regards,
-Ivan Salazar
Applications Engineer