VS1053 - Noise when connected to a Class D amplifier

[fgomes]

Hi, I'm using a class D amp (PAM8403) connected to the VS1053 L and R outputs using the 'signal reconstruction circuit' suggested (470 ohm in series with a 4n7 cap). I have the GBUF output floating. The PAM8403 breakout board that I'm using has a resistor divider in the input (10k/10k) and a blocking capacitor (1uF) connected between that resistor divider and its audio input.

What i notice is that connecting any other audio output to the PAM8403 the sound is good, without noise. Listening to headphones connected directly to the VS1053 outputs (L, R and GBUF) the sound is also good, I hear some low level noise when there is no music playing, but it is ok. The problem is that when I connect the VS1053 to the PAM8403 I have lots of random noise in the speakers.

The noise appears immediately as I power the modules, so it is there even before starting streaming data to the VS1053 (during the connection to the server to get audio stream, for example). With the scope I've noticed some high frequency noise in the connection between the VS1053 and the PAM8403, that should be related to the class D amplifier inside the VS1053, and that noise pattern varies if I connect the headphones at that point, but the noise in the speakers is the same.

Do you recommend any extra circuitry besides the signal reconstruction circuit? Is it necessary to make any GBUF connection when not using the headphones? Any other idea why this happens?

Thanks in advance!

Best regards

Fernando

[Panu]

Hi!

Hmm, with the reconstruction filter, the sound should be good. Hmm, can you compare your D class amp connection to this schematic, which is known to have very good sound quality: http://www.vlsi.fi/fileadmin/software/V ... der_03.pdf

Can you check the component values in the reconstruction filter, perhaps add another cap to gnd, and see if it helps?

-Panu

[pasi]

Does the noise disappear after you have played something?

[fgomes]

Hi

Panu, I've already ordered a TPA2012 to test and see if anything changes, I'll let you know as soon as I have it.

Pasi, the noise appears as soon as I have both (VS1053 and PAM8403) powered and connected. If I remove the audio connection between them the noise disappears, and if I add another sound source to the amplifier it plays well without noise. But with the VS1053 connected to it (through the suggested audio reconstruction filter), I have always the noise even during the VS1053 initialization, and while the main CPU is connecting to the server to get the audio stream (so before streaming audio data to the VS1053). I've also added a pause at the end of the audio stream before muting the amplifier and during that pause the noise is still there, with no audio playing. But I've also observed that during that time I have some digital signal present on the VS1053 audio output, I think that it is because it has a built in class D amplifier, right?

Best regards

Fernando

[pasi]

The vs1053 analog outputs contain out-of-audible-band modulator noise at 6.144MHz, which gets automatically suppressed in headphones - they cannot reproduce it.

If you connect to external amplifiers (especially D-class), the amplifier can alias that out-of-band noise to audible frequencies. That is the reason why line output connection and connection to an amplifier requires an additional low-pass filter. The quality of the ground plane and the DC block for the signal also affect the sound.

[fgomes]

Hi Pasi, thanks for your reply. I have discussed this issue in the past and that is why I've added the signal reconstruction filter (a low pass filter, basically, a series of a 470 ohm resistor and a 4.7nF capacitor), but even then the noise is noticeable. I've found that the amplifier I'm using has a resistor divider in the input (a series of two 10k resistors), dividing the signal to half, and then a capacitor to remove the DC component, so it should be ok to put directly in front of the low pass filter. The main difference I found when comparing to the amplifier suggested by Panu is that the one I'm using hasn't differential inputs, and the schematic suggested by Panu has an amplifier with differential inputs, with both negative inputs connected to GBUF. So if the noise is present on both channel outputs and GBUF, it will be canceled because of the amplifier differential inputs. I've ordered one TPA2012 to see if it changes the behavior, if it does I might choose another amplifier with differential inputs because this TPA2012 is a bit more expensive in volume production, but to make this test I prefer to use it in order to avoid introducing more variables in the test.

Just as a side note, I notice some noise even with the headphones connected directly to the VS1053, but it is very low when compared to the sound level, I'm only able to listen it in quiet passages. With the amplifier connected to the VS1053 it seems that the noises are much more amplified than the sound, making these noises much more difficult to tolerate.

Best regards

Fernando

[pasi]

Also check out viewtopic.php?f=10&t=2016 , maybe the cause of your issue is similar or at least gives you some ideas.

[fgomes]

Hi Pasi, thanks again for your help. I've already read the post you refer, the kind of noise I have is different, is more random clicking, not a constant sound, nevertheless I think it is worth to try some changes in the way I'm making the tests to see if it is possible to go to the root of the problem. I'll look at the AVDD and DVDD quality after the regulators (AMS1117 3.3V and 1.8V, I've already tried to replace the 3.3V one by an MCP1700 with similar results), eventually increasing the capacity before / after the regulators. I'll also check the influence of the low pass filter cut-off frequency, the noise level while streaming zeros vs no SPI activity, the noise level stopping the MCU, etc.

I'll keep you posted of the results, might be useful to others.

Best regards

Fernando

[Henrik]

Hello Fernando,

[...] the kind of noise I have is different, is more random clicking, not a constant sound, nevertheless I think it is worth to try some changes in the way I'm making the tests to see if it is possible to go to the root of the problem.[...]

Random clicking? Then it does not really sound like the sigma-delta reconstruction filter problem that is the most common issue, because that tends to create constant and pretty pure white noise. I am almost certain you issue lies somewhere else.

I'll keep you posted of the results, might be useful to others.

That is always a good thing, thanks!

Kind regards,
- Henrik

[CarlaTuck]

I had similar issues connecting a bluetooth audio receive into a Kenwood 50W stereo amplifier.
The only thing that worked was to use a DC-DC converter, but I agree its not that cheap, perhaps $5 for the low current which I needed
Even with DC-DC I still seem to get some noise but I think its just the cheap chipset in the bluetooth receiver, as they only cost $2.

printed circuit board