The PAM8403 amplifier is very noisy

i can see you have 3 seperate capacitor groups, Are you going to have 3 seperate amp boards,?

where is the 5v coming from is it from a switch mode regulator or battery controller,?

does this regulator have a biger size cap like 220uf where you are getting your feed from?

can you make a second regulator to supply the leds seperatly ?

It's a single PAM8403 (the little circuit you can buy anywhere).

It's a single power line, 5V via USB (so lots of options). I put in several capacitors to support the power.

The old board contained the amplifier and a controller for the LEDs, but it is damaged.

Most likely this is just digital noise coming through - you need to make sure you've got good routing to avoid it being picked up, good ceramic bypass capacitors nearby, ideally a separate analog ground plane that joins the main ground away from the digital and perhaps even consider some kind of filtering on the power (e.g. inductor choke) if it's really bad.

Ensuring the bypass on the LEDs is good is also important, especially if you have a bad ground, the noise could come around due to ground bounce. The addressable LEDs are PWM so they switch on and off "hard" and that modulates your supply and power distribution network.

Personally, I'd say using a differential input audio amplifier, rather than a single-ended type like you have, would be a better choice and routing pairs of audio lines to that amp, that way if the noise is induced, it is induced into both legs of the audio signal and thus cancels out.

I have some PC speakers powered by 5V [USB], the original board had a digitally controlled amplifier [via push buttons] and an unlabeled IC [for LEDs], all on the same board. The PAM8403 is a module soldered onto a PCB, and for the connections I had to use JST connectors.
I don't know much, but I'm still experimenting.

can you measure the current the complete led circuitry is drawing?

and try an inline choke to power the leds 

maybe a photo of the old board and complete unit ? 

It's difficult to completely remove noise when a lot is outside of your control; even some so-called hi-fi products get noise artifacts when things are interchanged in setups the manufacturer didn't expect.

Are you sure you're not expecting too much; what is the noise level in comparison to the audio level, and what does the noise sound like? Could you record it?

I happen to have one of those PAM8403 boards here, but I've never used it. I've buzzed out pin 8 which needs to be connected to a bypass capacitor according to the IC datasheet, and confirmed it's present, so the basic minimal circuit is at least implemented on that module.

I will try to power it up sometime, but no guarantee I can do that this week. Not sure it would be all that useful to you anyway, since I don't have the same setup as you.

If you can power-down everything else (LEDs/microcontroller), and remove the input (and maybe even short the inputs) then check to see if the noise is reduced a lot. Then try to add in the audio cables, and feed in audio from something completely separate and battery-powered to begin with, such as a portable MP3 player. Just take baby steps, and see what affects the noise.

I suspect that you may eventually reach a point where the noise is low, but still noticeably there, unless you're prepared to change things a lot, e.g. say switch from USB power to battery, or not use the LEDs/microcontroller if not required, and so on.

Hi, 

please take a look in this datasheet.

https://www.alldatasheet.com/datasheet-pdf/view/246505/PAM/PAM8403.html

-> page 10, figure 2: here is a description about an EMI filter, which helps to reduce EMI.

It seems your amp_PCB is not RF stabile; its layout is not optimized as well. Its GND trace is very small.

As well your PS_PCB please reroute for low noise and low impedance.
-> short traces as you can do, separate in-going from out-going, signal traces need pairing (parallel going) with GND not with V+.

3W output at 5V supply produce a relatively high and fast impulse current into the speakers, separate is it from input traces.

Do you have an oscilloscope.

It will be much easier to fix the problem if you know what it is causing it.

Since the noise happens when the LED is on it should be easy  to see what is changing on the amplifier when that happens.

MK

Decided to quickly try that amp board, my results won't necessarily help at all since my setup is completely different and I don't have LEDs/microcontroller attached. I'm not near 'scope so I couldn't grab any traces.

I'm using a very high quality audio cable (double shield), but there is about 3 cm bare at the board end. Speaker is 8 ohm (I only wired to the right channel).

For the power supply, I plugged into an Anker mains-to-USB adapter.

It's quite a noisy amp, there's reasonably noticeable hiss. With the 3.5 mm plug unplugged, I hear mains buzz; it needs to be shorted for that to go away, leaving just the hiss.

Then I plugged the audio cable into a battery-powered MP3 player. It sounded fine.

However, then I attached the audio cable to my laptop instead, which was powered from another port on the same Anker adapter. It was intolerable, lots of electronic noise sounds. If I unplugged the laptop and let it run on batteries, then the sound became acceptable again.  

I tried a few things; additional decoupling (220uF electrolytic cap on the power rails, no difference. Then added a choke preceding that (a 1 mH common-mode choke with the windings wired together to be a normal single choke), still no difference. Then removed the electrolytic and re-soldered the choke as common-mode, then tried two of them : ) and also tried adding it to the audio connections too. Still no difference.

Then I remembered I was not using the official laptop charger - I attached the laptop to the correct charger, and the amplifier remained on the Anker one. Noises gone. The official laptop one has 1k resistance from the USB shield to the earth pin. The Anker one doesn't have an earth connection (i.e. SELV).

I'm guessing perhaps the laptop puts enough noise into the charger, that it egresses on all the other ports (common-mode or differential or both - don't know, guessing both!), and the filtering I was doing wasn't sufficient - I should have tried both differential and common-mode simultaneously but I didn't get a chance.

Hi  shabaz 

Here are the datasheet and an app-note. The PAM8403 is obsolete, the PAM8406 is recommended.

As I mentioned you will need filtering on the output to the speakers. Also, the PCB needs improved routing.

In the PAM8406 schematics BOM list you have the values of the components.

 A huge problem is a GND loop, -ringing and -bouncing.

-> Poweranalog is now part of Diodes:

https://www.diodes.com/search?q=pam&action_results=Go&start=20

Regards Gerald