LME49724: Parts testing out of specification

Hi, Alan,

Welcome to E2E and thank you for your interest in our products!

Could you verify if the power supply in the failing units also has higher noise or ripple level?
In the failing units, is the noise level still high when there's no load connected?
Are the failing units part of a specific lot?

Thank you.

Best regards,
Luis Fernando Rodríguez S.

Hello Luis,

The power supply looks clean. I don't see any difference in noise and ripple on the power supply of a failing units compared to a passing unit. I'll have to check to see if the failing units are from a specific lot. Our test uses a method of running output audio into the input and then analyzing the audio data with audio precision. The load is necessary to check the signal to noise.

Thank you for your response, Alan,

Please let us know if the parts come from the same lot.

In addition, have you checked if the analog inputs have a higher noise level in the failing units? Or in any other pin besides the outputs?

Best regards,
Luis Fernando Rodríguez S.

Is the lot code the series of characters next to the "N" symbol.    The failed units I have seen are labeled 82D5 and 89J2.  

How exactly do I check the analog input noise level?  I did run a 1KHZ sine wave through the part and the output was slightly more noisy.

 

Hi, Alan,

The graphs will be enough to check the noise issue.

Could you try removing the FB14, FB15 and TVS1 components just to verify if they could be associated to this issue? In the past, we have faced some issues with ferrite beads involved in high noise level.

In addition, do you have a percentage number of the failing units? How many units fails compared with the passing units? How many of them are from the lot 82D5 and 89J2? Do you also see passing units with these numbers?

Thank you. And I apologize for all these questions, this is just to ensure if this should be associated to the quality team or not.

Best regards,
Luis Fernando Rodríguez S.

Hi Luis,

I removed the ferrite beads and TVS components with little effect as they were still out spec for SNR.  Every time we replace the failed part with a new one our manufacturing SNR tests pass.

 

As far as the percentage number of failures.  That is difficult for me to give an exact number as this has been an ongoing problem for several years and they would just replace the chip since there were so few.  Probably about less than 1% although it has been over 5% since last November.  Most units are passing and there are many passing ones with the 82D5 and 89J2 lot codes.

Hi, Alan,

Thank you for your details and sorry for my late reply.

Another question from my team, have you tested the bad units with one board or product that has passed the noise test? This is to ensure if the noise is associated to the board or to the LM49724.

Thank you.

Best regards,
Luis Fernando Rodríguez S.

Hi Luis,

I started to notice a trend on the failing units.   I an enclosing a before and after scope capture photo of pin 8 of the LME49724.  The first photo is the circuit as is and the second picture is with everything the same except the LME49724 was replaced and now is good and looks to pass our tests.  

Is there something in the manufacturing process causing a variation on some parts?