LM4670: Output Noise without any Input

Hi Rushi,

Let me take a look at your questions and I'll provide some feedback shortly.

Best regards,
-Ivan Salazar
Applications Engineer - Low Power Audio & Actuators

Rushi,

Have you tried to connect pin 1 (shutdown) directly to VCC_5V instead of VCC_1V8?
Perhaps this pin is not stable and its resetting the device periodically.

From your scope captures of VO1 and VO2, I don't think the device can be oscillating, but you can also try by removing C208 and C209.

Basically three different things that I think you can try are as follows:

• Short pin 1 to VCC_5V
• Short the inputs to GND to make sure there is no high frequency signal coupling in at the inputs. Connect the input side of C124 and C125 to GND.
• If you're not using inductors or ferrites at the output, remove the capacitors. Capacitive loading is not good for the amplifiers as this could cause oscillations.

Best regards,
-Ivan Salazar
Applications Engineer - Low Power Audio & Actuators

Hi Ivan,

Thanks for the suggestions. 

1. By removing the C208 and C209, It helped a bit to reduce the audible noise. But the same noise was still audible.

2. Shorting the Pin 1 to VCC_5V did not help.

3. Connecting the input side Capacitor to the ground (Basically, shorting the inputs to the ground) did not change any response. The Audible noise went low but it was still not neglectable. Here is how the output waveforms look like with short and without shorting.

Input without Short (Connection to the host processor, as per schematics shared and C208 and C209 removed, Pin1=5V when the processor is not playing anything).

Inputs Shorted to Ground (Pin1=5V, C208 and C209 removed)

Hi Rushi,

Thanks for sharing your test results.

Those changes in the output switching phase are not good, this is present both on your first capture and in the latest one. Perhaps there is still something restarting the device or making it trigger some protection.

Can you share your layout as well? You can send it through a private e2e message or to my email: ivan.salazar@ti.com
Have you tried by disconnecting the speaker load? Does the output waveform looks the same without a load?

Best regards,
-Ivan Salazar
Applications Engineer - Low Power Audio & Actuators

Hi Ivan,

I can help sharing the VCC_5V, Pin1, Vo1, Vo2 in the one capture to help you understand better. In second capture I can put Vin+ , Vin-, Vo1, Vo2 without the speaker connected. I changed speakers from (1W,8R) to (3W,4R) just to check if my speakers have a problem with THD+N from device, but audible noise is the same as we see in the capture itself.

I shall connect you on your email. Thanks.

Hi Rushi,

Those captures could help.
However the two options that I thought could be happening are:

• There is some high frequency oscillation, and that's the source of the noise
• The device is being restarted periodically, causing the phase displacement seen on the scope captures

If there is high frequency oscillation, the output Class-D should be modulating, but from the captures you have shared so far the output signals seem to be at 50% duty cycle.

Best regards,
-Ivan Salazar
Applications Engineer - Low Power Audio & Actuators

Hi Ivan,

Thanks for that.

I checked the device power and shutdown pins and observed that through the scope; the power and shutdown pin seems to be stable. I checked from device to device by swapping the boards but seems to be the same case with the noise. Removing the capacitor at the Vo1, Vo2 lines certainly helped but noise is still annoying. I tried changing the speaker loads from (3W, 4R) to (1W, 8R) but not much change here. Do you think that the power supply noise could be causing this noise? I supplied power from the battery, but the noise is still the same.

Placement Layer:

Immediate Layer:

Hi Rushi,

Did you get some captures of the output when no load is connected?
Do you have spec or data sheet of your speakers?

Is there a chance you can connect the speaker on the right side of R327 and R328? I would like to check if the traces can affect the behavior.

Best regards,
-Ivan Salazar
Applications Engineer - Low Power Audio & Actuators

Hi Ivan,

1. The capture response is same as shared earlier.

2. The speaker I am connecting is AS04004PO-2-LW152-R

3. I removed those resistors and connected the speaker right infront of the Vo1 and Vo2 lines. It doesn't help.

Changing the Length of the wires also not making a difference.

Could you check if this is observed on the EVK as well?

Should we shift to a new part altogether? Any recommendations for 1-ch mono class D, 3W? Our primary reasons for choosing this part was the WSON package to reduce the assembly complexity.

Rushi,

There is no evaluation board I can use for test, I'll check if we can use some kind of breakout board for this device.

As for alternative devices, we have plenty of alternatives although most of them are BGA packages.
LM48511 and TPA2013D1 could be QFN options.

Best regards,
-Ivan Salazar
Applications Engineer - Low Power Audio & Actuators

Thanks, Ivan.

I will look forward to your results from breakout board experient. For the alternatives, I do have a limited choice due to package limitation. I am looking towards LM4675, but little afraid now to choose the part from the same old family.

Regards,

Rushi.