PAM8403 & BT Module - Noisy!

eamest

Member
2019-10-25 7:33 pm
Hi all,

I have been working on crafting together a portable speaker for my bike and have picked up dev boards for the PAM8403 Class D amplifier and a Bluetooth audio module from Drok.

Both of these boards work great independently. The system even sounds decent when the two are powered by their own separate power supplies.

As soon as I power both the amplifier and the BT module from the same supply (a lithium battery) I hear audible noise (in the 1-2kHz range from what i read on the scope). Whether there is music playing or not this noise is constant.

I have tried adding bypass capacitors and L-C filters to the inputs of both modules but can not get the noise to attenuate.

What am I missing here?

Rough schematics of system. Top is w/ noise and Bottom is w/o noise.
IMG-20191025-142811 — imgbb.com

Thanks,
-Tim
 
What kind of noise: hum, hiss, shoots, white??? Perhaps a small CLC in the power lines to decouple both between them can help. Use, say, 100µH and 2 * (1000µF and 0.1µF ceramic) at both ends of the filter, like a π filter. Voltage of caps and current of inductors ratings according to the parameters available in the amp.
 
Last edited:

jean-paul

diyAudio Moderator Emeritus
2002-09-20 7:20 am
Germany
Sidenote: PAM8406 chips are better. There is even a board that is very small, has film caps at the inputs, power line filtering (AFAIK CLC which might be beneficial in your setup just like Osvaldo pointed out) and full output filtering. The brand is LOSC. Sounds quite OK. Drawback: 3.5 mm input Jack....

Bluetooth has no place in serious audio but it is convenient. Less convenient is that many BT boards emit a lot of noise and stuff on the PSU lines. Can be hard to get rid of. A metal case for the amplifier board or shielding probably is necessary.
 
Last edited:

eamest

Member
2019-10-25 7:33 pm
What kind of noise: hum, hiss, shoots, white??? Perhaps a small CLC in the power lines to decouple both between them can help. Use, say, 100µH and 2 * (1000µF and 0.1µF ceramic) at both ends of the filter, like a π filter. Voltage of caps and current of inductors ratings according to the parameters available in the amp.

The noise is not a hiss, but rather a whine. Sounds to be focused around a specific frequency. I measured the ripple on my 5V power line and noticed it to be 1.6kHz. Assuming this is what's being coupled to my amplifier input and thus speaker output.

I will try out a pi filter and see if it helps.
 

29285

Disabled Account
2010-01-17 9:20 pm
This noise has been observed many times now and I had this problem, too. To make a long story short: The BT modules have pulsating current consumption with frequencies in the most audible range about 1~2kHz. This adds some mV of ac-ripple on the GND return line and easily forms a gnd-loop with the effect you described. You options are
-separate supply for BT (which worked for you already)
-low impedance GND plane
-blocking cap directly placed at 5V/GND of BT module. >470uF, LowESR type.
-using symmetric audio coupling between BT and amp input.
 
I found the mistake in my set-up with Bluetooth almost instandly.
If it works well (silent) with different power supplies, you must have one or two ground lines. Stereo, left and right.
Now you connect to a common PS, with the 2nd or 3rd ground line.
This has to be noisy! If not, something is wrong...
Even symmetric signal will not work, if you connect the ground of the XLR connectors screen.

So, lift the signal ground at one side (which side may have different results), only use the fat ground (minus, - ) of the PS. If, on the Bluetooth side, power ground and signal ground are separated, connect them at the BT board.

If still noisy, throw the junk in the garbage can and don´t use BT. By the way, to talk about HIFI and amp quality, capacitor and coil "sound", while hearing with BT is a very stupid idea. Fine for data reduced MP3, if you are to lazy to insert some plug.

If you still have some kind of frequency "down mix", take a small piece of sheet metal, ground it and try to shade different positions between the BT module and the amp. Play around while you listen to the noise. Maybe you get an idea where it emits and where the antenna is. Very educational, too.

If you are serious about such stuff, you place the BT in a metal box and lead the antenna out of your case, away from the amp section. To put tuner in little cages has worked for ages: Just forgotten by computer infected folks.

Usually this will reduce noise to zero, on "good" modules. As these are rare, all the above CRC or CLC filtering and decoupling will do the rest. Fixing only one problem might not give the "real silence" that would be OK for a serious audio device.
 
Last edited:

eamest

Member
2019-10-25 7:33 pm
This noise has been observed many times now and I had this problem, too. To make a long story short: The BT modules have pulsating current consumption with frequencies in the most audible range about 1~2kHz. This adds some mV of ac-ripple on the GND return line and easily forms a gnd-loop with the effect you described. You options are
-separate supply for BT (which worked for you already)
-low impedance GND plane
-blocking cap directly placed at 5V/GND of BT module. >470uF, LowESR type.
-using symmetric audio coupling between BT and amp input.

Shunted a 1000uF cap at power input of BT module totally cleared it up! Thanks!

When you say BT modules have pulsating current, this is something that all modules exhibit or the specific Drok one? I would expect the controllers on the board to be switching at higher frequencies than 1-2kHz.
 
If you have enough capacity near a current consuming part, the pulsating demand may be buffered. If the part has been erased or made to small, because any fraction of a Cent counts, these pulses are modulating the power supply.
You are lucky your cap did the job. Sometimes such a capacitor does not reduce fast transitions, you may need some smaller ones too.
Any module can tolerate some dirt on the supply line, how much depends on the construction /dimensions.
With our cheap modules we use, the work of integrating them into a system is always the challenge.
In most cases we reverse the cost savings the producer has driven to the edge of functionality.
Many people ask why someone should use these modules, instead of buying something better. The answer is that "better" is not easily available and in most cases "more expensive" parts mean only more profit for the seller, but are constructed just as cheap.
The quest is to take a cheap basic part and turn it into something of better quality than the price would allow.
 
@eamest
in the plan you posted, if you take a red marker and follow only your ground connections, you will see that the upper scheme has a perfect loop between BT-module and amp. This has to be noisy. The lower scheme has no loop and is silent.
So, if you think for a second, can the capacitor solve a problem that does not exist in the lower way of connecting. Yes, to some extend, as it evens out the pulsation the BT induces. But the cause is still there!

Break the loop and it will be even more silent.