TPA2016D2: Ticking noise heard after product in field for few months

Hi, PYTsai,

I'm not able to see the past e2e thread. Could you share details about the schematic? Do you know if the problem is solved when the TPA2016D2 unit is replaced with a new one?

Best regards,
Luis Fernando Rodríguez S.

Hi Luis,

Please check the original post here 

TPA2016D2: Consistent Low Volume Ticking Noise After Some Time in the Field

Yes, with replace a new device the issue gone. 

Thanks!

Hi, PYTsai,

Do you know if there are differences on the bad units output when the output filters are removed? Does the issue still persists?

Best regards,
Luis Fernando Rodríguez S.

Hello Luis,

I am the original customer represented by Py who is having this issue.  I work with Ed, who first posted our question about this part in the thread Py linked to.

We first removed the series resistor connected after the speaker and found the issue still persisted.  We then removed the LC filter, leaving L open-circuit (L is actually a ferrite bead in this case) and the issue still persisted.

Below is a capture of TPA2016D2 pin 6 (yellow) and pin 7 (blue) with the L, C, and R components of the output filter removed.  Essentially, the TPA2016D2 output pins see an open-circuit.  You can see a severely skewed duty cycle and the ticking is quite loud.

Thanks,

Dan

Hi, Daniel,

Nice to meet you.

Thank you for providing additional information about this.

In the previous post, you mentioned that the problem follows the part. I mean, when the TPA2016D2 is replaced by a new unit, the problem is fixed. Have you confirmed this behavior on a new board? I mean, if you test the bad unit in a board that have not faced these issues.

In addition, do you know if the part is exposed to some parameters limits? Is the part near of the absolute maximum rating or beyond the recommended limits specified in the datasheet?

Have you sent some units for a failure analysis?

Best regards,
Luis Fernando Rodríguez S.

Hi Luis,

Thank you for the response, nice to meet you too.

Yes, we've confirmed the same behavior for a bad TPA2016D2 on a new board.  We completed TI's A-B-A Swap Method prior to returning parts for failure analysis, so we've confirmed both a new TPA2016D2 on the original board fixes the issue, and a bad TPA2016D2 on a new board recreates the issue.  The issue does follow the part.

We are pushing the TPA2016D2 to its maximum capability as it is powered from 5 V and drives a 4 ohm speaker load.  As the oscilloscope captures show, we can see ringing measured at the TPA2016D2 speaker output pins whose peaks can slightly exceed 6 V, but I don't see an absolute maximum rating for the speaker output pins in the datasheet.

The audio input to the TPA2016D2 is driven by a codec powered from 3.3 V and I have not seen the input exceed the 6 V absolute maximum rating.  I've measured peak-to-peak noise on the 5 V input to the TPA2016D2 and have not seen worse than a bit over 300 mV, so a +5.15 V supply is still under the 6 V absolute maximum rating.

I do wonder if our ferrite bead on the speaker output has any contribution - it is rated for 3 A and has 220 ohm impedance at 100 MHz.  I notice in Table 2 of the Filter-Free Class-D Audio Amplifiers Application Report (SLOA145-February 2010) that the TPA2013D1 with a 4 ohm load suggests a 4 A ferrite bead.  I also see the TPA2016D2EVM User's Guide (SLOU201-June 2008) lists 100 ohm, 4 A ferrite beads.  Otherwise our design matches the reference design for the TPA2016D2EVM evaluation module.

We have returned units for failure analysis under product return # CPR201037144.  The return was approved and the shipment was delivered this past Monday, 3/2.

Thank you,

Dan

Hi, Dan,

Thank you for your information. Could you please share the results of the failure analysis when you receive them, please?

Since the application matches the evaluation module design and no absolute maximum ratings are reached, it is hard to arrive to the root cause. The ferrite bead current rate could be associated, but we cannot confirm it until we get some data.

In case you have any other information about it, please let me know.

Thank you.

Best regards,
Luis Fernando Rodríguez S.

Hi Luis,

I will share the results of the failure analysis as soon as I receive information.  I will also post any new information we find in our testing.  Meanwhile, please let me know if there are any other aspects to investigate or tests to try that may lead to the root cause.

Thank you,

Dan

Hi, Dan,

Do you have any update on this?

Best regards,
Luis Fernando Rodríguez S.

Hi Luis,

Thank you for following up on this.  I was contacted by TI on Tuesday of this week (3/10) and informed that the parts have been received and they are proceeding with initial testing to try to verify the issue.  They expect to have an update and analysis plan by the end of this week (3/13) which we are eagerly awaiting.

Meanwhile we've had a few more failed units return from the field.  I seemed to have one amplifier on the verge of failure - in our application a sound is made when an on-screen button is pressed, and this sound initiated faint ticking that stopped after a few seconds.  This could be repeated with each button press.  I did not see any input signal from the codec to the amplifier after the initial button press and while the ticking was occurring, but otherwise could not observe any new clues.

Two other thoughts - is there any reason to suspect a ferrite bead in series with the 5 V supply to the amplifier could cause the issue over time?  I don't see one in the evaluation board for the amplifier.  What about the fact the 5 V supply is sourced from a buck switching regulator?

Thanks,

Dan

Hi, Dan,

We have faced cases in the past where for some reason the ferrites cause issues on the audio converters during the sampling process. We cannot guarantee that they are the root cause of this problem until we have a complete analysis, but in the meantime I would recommend to remove it and see if the problem persists or not. In addition, it is always recommended to reduce the ripple activity on the power supplies in order to ensure an stable activity of the device. We have used buck regulators in the past and the only issues are associated to ripple noise.

Please let me know if you have any other feedback on this.

Best regards,
Luis Fernando Rodríguez S. 

Hi Luis,

Thanks for the feedback.  I will see about removing the ferrite in series with the supply, that is certainly something we can test in-house.  The issue would be allowing enough time or recreating the conditions under which the part may fail.  Since it seems to be based on time in the field, as in we are not aware of any parts failing relatively early in their field deployment, we either have to just let the test cases run or attempt to accelerate the aging.  I will see if we can better characterize any potential ripple as well.

As is certainly expected due to the COVID-19 outbreak and corresponding restrictions, the Amplifiers group at TI handling our failure analysis will be delayed in testing and collecting the information.  I will provide the results as soon as they are able to complete the analysis.

Thanks,

Dan

Thank you for your feedback, Dan,

We will wait for your response.

Best regards,
Luis Fernando Rodríguez S.