Digital noise in battery powered system

Hello,

while building a Bluetooth speaker based on this instructable, I ran into some problems regarding noise. There already exists a question regarding this problem here, but it is a few years old, and it gave me no solution.

The power source for this project is a self constructed battery pack consisting of 4 18650 battery's soldered to a battery protection circuit (just like in the instructable). The PAM8610 (black version) amplifier board, however, can't take 16.8 volts so a voltage regulator is placed in between set to 12 volts. Same story for the Bluetooth receiver board (KRC-86B V4.0), but with a voltage of 5 volts.

All fine and dandy till there. The KRC-86B has an input for a 3.5mm jack connection and an output. The output is a pass-trough of the 3.5mm input unless a Bluetooth connection is present, then the Bluetooth signal will overwrite.

Here's the problem: whilst the connection using the 3.5mm jack works fine, creating a Bluetooth connection results in a very annoying digital noise. This noise is there instantaneously and is the same on all volumes of the host/transmitter. If no music is being played by the host for around 15-20 seconds the device seems to enter a power saving mode and the noise changes to a regular beep. This is way less bad, but of no use because it is replaced again by the bad noise when the music starts again.

I have tried a diversity of things, listed here with their effects:

Action: wrap the Bluetooth receiver in electrical tape and aluminium foil.
Result: nothing much, it didn't even block the music. Signal was only blocked by three layers, a few meters distance and my body in between my phone and the receiver.

Action: lift ground in signal between Bluetooth receiver and amplifier.
Result: noise gets slightly worse, no change in tone, only amplitude.

Action: moved away from my computer, as to avoid interference.
Result: nothing at all.

Action: tried another Bluetooth transmitter, PC instead of phone.
Result: -

Action: broke the previous board at the start of this project by heating it too much in the attempt to change a wire.
Result: noise was also present at the first board which I did test by hooking it up to the line-in of my computer. Though that could also have been by the ghetto set up I used. The 3.5mm cable was poorly connected, trough a breadboard.

Taking my multimeter and measuring the difference between negative of the battery and GND of the voltage regulator for the amp gave a 6.0 mV reading. The same result is achieved when measuring between battery and GND of the regulator for the KRC-86B. The potential between GND of the signal and the GND of one of the regulators was 0.1 mV.

I understand that you might want to hear the noise, so I recorded it: link. It is very soft, but audacity doesn't read .m4a files, and turning up the volume works just as well.

Please let me know if you have any idea's on what could solve it and/or if you need more info.

Thanks in advance.

Picture of Digital noise in battery powered system
IMG_20160318_232648257.jpg
IMG_20160319_205217216.jpg
sort by: active | newest | oldest
-max-1 year ago

Ummm, I would first suspect those switchmode buck converters. They can be pretty noisy electrically speaking. They work by basically turning on and off the battery voltage really fast, and they use a LC filter and a feedback loop to smooth that switching out some and regulate the output voltage accordingly. However the LC filter following that switching transistor cannot remove all the noise, just attenuate it. There might be something like 100mv of peak to peak ripple on the output. To fix it, add some decoupling capacitors to each board, as well as additional bulk filter capacitors, or better yet, LC filters following the power supply modules or maybe some LDO regulators.

When it comes to filter and bypass capacitors, watch this video:

Different capacitors respond to different frequencies so for the lower frequency stuff, electrolytic capacitors are fine. However, because of their equivalent series inductance (ESL) and equivalent series resistance (ESR) they are not well suited for higher frequency filtering, so you would use something like a film capacitor or large multilayer ceramic capacitor for that.

Downunder35m provided lots of good advice. Long wires act like antennas, you need to put put some distance between signal wires and power wires, and you need to electrically shield the amplifier board and the signal wires, (using aluminum foil like that should help, but it's pretty cringeworthy! Proper coax and shielding would be much better.) An osciloscope and a frequency analyzer is a priceless tool / instrument to have in this case, you could use it to track down where that noise is coming from. When measuring small analog signals with a scope, it is best to get rid of that "antenna earth lead" and use that little spring that slips onto the end of the probe to get the lowest inductance connection possible to the signal you care about.

arduinefier (author)  -max-1 year ago

That was very helpful, thanks! I now have a general plan of things I will try to fix the problem. Including properly shielded wires (Downunder35m, yes, no more foil :-)). Today I have had a conversation about this subject with a professor of my university, for noise and interference was actually the subject of the practicum. Together with the video I now fully understand what he was talking about.

I tore some capacitors of different sizes of an old PSU where I also got my wires from. The thing was defect anyway.

However, my amplifier somehow burned up a few minutes ago. The positive power connection had come loose, so I refastened it, turned it on, and SWOOSH. Because it was very inexpensive (€1.17, though pretty good still) I will order 3 more as to avoid this in the future. In short: my project is on hold for two weeks or so. I might do some testing with the line-in of my pc but for today I'm done.

Yeah, this is definitely a practical designing problem and not so much a theory. Understanding both the practical considerations are equally important to theory.

Getting old electrolytic capacitors from salvaged parts is generally not a good idea. Electrolytic capacitors dry out over time and their ESR increases. Other types of capacitors are normally fine.
arduinefier (author)  -max-1 year ago

Luckily those were (mostly) ceramic capacitors, but good to know. I looked up he phenomenon, and these could be too old. Since I was experimenting with the capacitors, I did connect an old electrolytic capacitor before the amplifier fried itself, though it was at the terminal of the Bluetooth receiver, and not the amplifier itself ¯\_(ツ)_/¯. Anyway, I will avoid those in the future and order some new ones with the amplifier.

They mainly teach about the theory at a university. This project is getting more and more useful.

Connecting bad electrolytic capacitors into the circuit should not cause additional problems, unless you are dealing with one of those really sensitive LDO regulators that will lose regulation unless you keep the output capacitance within a tight range and the load current within range, or whatever other kind of inherently unstable regulator

Here is a good video for understanding why I think that those switch mode converters are producing the noise:

arduinefier (author)  -max-1 year ago

I tried the new amplifier and connected it to the Bluetooth module with RCA cables with the connectors cut off. Situation same as first, which was nice because it meant the rest wasn't broken. Tried the ground loop isolators which I also ordered just in case, but that did not appear to be the problem. Thus I think you are right in suspecting the power delivery. Then the Bluetooth module decided to quit |-(. I now have my capacitors so when I get a replacement I will connect everything the right way, first go (and maybe try an oscilloscope if the problem persists). Anyway, thanks for the help, I'll select it as the best answer and maybe post an update if I find anything useful.

"Ummm, I would first suspect those switchmode buck converters"

+1

One thing I learned back in the day when playing with vinyl is that everything that can interfere with low power audio signals will interfere with them once they have a chance to do it.
What I see in your pics is a big mess - no offense as I start my projects in a similar manner ;)
But you deal with signals here that do more than switching a LED on and off, you want good music.
My first order of business would be:
Replace all wires with something as short as possible.
Use shielded chables for all audio signals, judging by your lengths the ones from old CD rom drives in old computers might be enough (if the shield is like a coax cable).
The shield should be on ground, battery negative and all ground connections should be joined together in the system.
The BT module will have the antenna on the circuit board or just a tiny wire for it poking out.
There is not much you can shield here but you add a small inductor and capacitor to the power input to filter out any EM interference from the module on the power lines.
My best suggestion is to locate the player and BT module close to the outputs of your box to keep the power cables as short as possible.
The speaker wires should be no problem, just the sound input and all other audio lines should use shielded cables.

arduinefier (author)  Downunder35m1 year ago

It is/seems a big mess because a photo shows almost no depth, it was an older photo and it actually was a mess :-). I have since improved the situation.

I tried insulating the wire and the Bluetooth module and did some more testing. The noise grew slightly less loud when I connected the negative power terminal of the BT module directly to the battery (though still audible, even at more than 30% volume). Though as you can read it didn't quite disappear. I can't really shorten the signal connection because I planned on gluing the module to the right side.

There already is a capacitor between the power cables of the module (470 uF). I have another one, but no change comes forth when I connect it as well.

All the grounds in the system are connected, except the ground of the signal. When That wire is connected to the 'common ground' as well, nothing happens.

I know that aluminium foil might not be as good as real shielded wires, so if you think that I should try them as well, just say so. I still have the cable from my first test which is shielded. It was ripped from a 2.0 PC speaker set.

IMG_20160321_140249107.jpgIMG_20160321_135035573_HDR.jpg
arduinefier (author)  arduinefier1 year ago

I decided to change plans, and move the receiver to the left, thus
shortening the wires to 7 cm or so. A reduction of 40%. This still did
not make a difference. Also soldered the connection from the Bluetooth receiver directly to the battery.

IMG_20160321_160758913.jpg

You still have a nice collection of antennas in the box...
If you really need such long cables and can't go with direct connections then get some CAT5 cable or similar.
At least this way all wires would end up in a single strand with some shielding around it.
The BT module might be crappy but even a top range one will struglle if you add so many antennas to the layout.

+1

iceng1 year ago

We live in an FM fringe area and enjoy the kitchen radio that is barely possible to make out talk shows, what with the new fluorescent track flood light hiss and basement hobby tumbling machines.

Just today, I put a 2A Corcom EMI filter on the tumblers and a line to line 35VA isolation transformer (low pass conduction filter) on the kitchen radio resulting in a good clean audio !

If you can't find an isolation transformer use two small 12v xfmrs wired line - 12v to 12v -line, this may be a better low pass filter because inter-winding capacitance does not come into play...

Good luck keeping it short....

CORCOM.JPG