GND loop noise

[Saimoun]

Hey there

I have a noise issue on my device, and I would love to hear your thoughts
The device is 5V USB powered, and the noise comes from an LED display (which is PWM'ed).

I think the issue is coming from a GND loop, or more exactly ground currents not going where they are supposed to. It's my analysis but I am not 100% sure.

Basically the display creates voltage noise on the 5V supply, which is fully sorted, so that's not an issue. But it makes sense that the return currents (i.e. GND) have the same noise, which is not a problem either when these currents go through the USB connector's GND.

Problems start to arise when I have another connector on the board, which gives another path for these return currents. See image:


(note: I guess the red arrow on the top right is either very small or not existent, since most currents will go back to the source which is the laptop / wall plug on the left)

There is a big difference from one audio equipment to another, whether this noise on GND is hearable or not. On some you cannot hear a single thing, on others it is very obvious.

The solutions I found so far are:

• Telling the users to avoid this type of wiring (i.e. use a separate wall plug for the device)
• Removing the noise by not PWM'ing the display -> a big change on PCB/electronics, and it would require another display
• Taming the noise by putting a 470uF cap in front of it -> difficult to do because of the USB in-rush current limitation, also that does not get rid of the noise completely
• Somehow avoiding connecting the ground on the Jacks -> unfortunately this is not possible on one of the jacks since the signal is single-ended and needs a reference

As usual I am using this forum both to hear your opinions and ideas, but also as a self-reflexion board (also this could be useful to people in the future having similar issues)

Thanks!
Simon

[Saimoun]

Just answering myself here, I think I have another solution:

5. adding a small resistor between GND and the gnd connection of the Jack connector. I think just 10ohm already would reduce the noise by quite a bit.

[jonpaul]

Suggest to do a bit of research on grounding, and audio noise reduction
See SYN AUD COM and many AES papers on this issue.

https://www.prosoundtraining.com/

Best refs is Henry OTT Noise reduction techniques in electronic systems
https://www.amazon.com/Noise-reduction-techniques-electronic-systems/dp/0471657263/

See my AES papers on digital audio transmission and transformers.

Use transformers in digital audio links

 break ground loops with audio transformer for baseband audio

Neutrik and any others make inline adapters.





Jon

[pqass]

Do you hear the same noise if you plug the laptop into the same wall receptacle as the audio amp?

[Saimoun]

Thank you jonpaul I will definitely look into all of these 

pqass -> on my example the audio equipment is also USB powered. I think there is no noise at all if it has its own power supply, especially if the laptop is in a different wall plug.

I actually just tested now adding a 10 ohm resistor from Sleeve to GND on the Jack and most of the noise disappears. Do you guys see any issues with the 10 ohm resistor? The Jack has a quite high output impendance (about 1k).

[pqass]

"...10 ohm resistor from Sleeve to GND on the Jack..."  bad connecton?    I'm not sure I understand.

Is there normally a physical connection between the display and the jack?  like a USB powered audio player (with stereo outputs)?   Or, could the jack wiring just be picking up stray EMI?   If the jack and the display aren't connected, what happens if you move the jack and it's wire away from the display?

[Saimoun]

"...10 ohm resistor from Sleeve to GND on the Jack..."  bad connecton?    I'm not sure I understand.

I mean if I add that 10 ohm series resistor from Sleeve pin (i.e. ground pin) on the connector, then the problem disappears.

Is there normally a physical connection between the display and the jack?  like a USB powered audio player (with stereo outputs)?   Or, could the jack wiring just be picking up stray EMI?   If the jack and the display aren't connected, what happens if you move the jack and it's wire away from the display?

Well the display is on the same PCB as the Jack. But the only Physical connection is Ground. Moving the wires makes no difference.

The way I understand it is that
1) the ground current makes the reference voltage move a bit
2) if somewhere inside the Audio equipment there is a single ended signal that uses this "moving" reference voltage as a... well, reference, then it "looks" like the audio voltage is moving, i.e. the amplified audio signal (sent to the speaker) has the noise.

Same as 2) if instead of having 1 audio equipment you have 2 connected together with a single ended audio cable, where that red GND arrow is going through. Then the cable picks up a buzz sound even tho the audio voltage is 0V (because the GND is "moving"). That case creates the most noise especially with a long cable.

Sorry if this is not clear it's difficult to explain And feel free to disagree

[jonpaul]

currents flow in a closed loop.

Sneak/stray paths develop eg mains PSU/transformer capacitances, or flow of current in ground systems

There is no such thing as a "ground"!! every conductor has L and R however small.

This the stray CM and DM currents induce a voltage drop which may be summed with the desired signal.

The key is

a/ Breaking unwanted current loops

b/ Use a transformer preferable  a shielded to break a loop

c/ Avoid copper TX eg Bluetooth, or TOSlink Optical

d/ Use shielded transformer



Kind Regards

Jon

[wasedadoc]

Reducing the noise source could also help.  Improve the decoupling at the display electronics.

[pqass]

"...10 ohm resistor from Sleeve to GND on the Jack..."  bad connecton?    I'm not sure I understand.

I mean if I add that 10 ohm series resistor from Sleeve pin (i.e. ground pin) on the connector, then the problem disappears.

Is there normally a physical connection between the display and the jack?  like a USB powered audio player (with stereo outputs)?   Or, could the jack wiring just be picking up stray EMI?   If the jack and the display aren't connected, what happens if you move the jack and it's wire away from the display?

Well the display is on the same PCB as the Jack. But the only Physical connection is Ground. Moving the wires makes no difference.

Are you sure the tip or ring pins of the jack aren't connected to something on the PCB?
I sounds like you've got a missing GND until you've added the 10 ohm resistor to the jack sleeve pin.
If so, as Dave would say, the inputs to the amp are "flapping around in the breeze."

[Saimoun]

@Jon: yes I understand. In my example (diagram), my "unwanted current loop" goes:

wall plug -> laptop -> USB cable -> USB plug on PCB -> display -> Jack plug on PCB -> Audio equipment -> Laptop -> wall plug

while it should go

wall plug -> laptop -> USB cable -> USB plug on PCB -> display -> USB plug on PCB -> USB cable -> laptop -> wall plug


Adding that 10 ohm resistor does not break the loop but it increases its resistance so forces most of the current to take the other path.
Yes a transformer would be even better you're right


@wasedadoc: agreed! I put a LC filter which cuts at 3.3kHz in front of the display. That removed the high frequency stuff (probably good for EMC), but did not really make any difference sound-wise. Which makes sense since the display's PWM frequency is at about 120Hz.


@pqass: good point! But no unfortunately, everything is connected to where it should be, the tip/ring are connected to the signal lines and the sleeve is connected to ground. The board is a proper PCB and the connectors are soldered onto it. (to test adding that 10 ohm resistor I soldered it in a cable, so I can try with a "normal" cable and then with my "10 ohm on sleeve" cable to compare)

[magic]

@Jon: yes I understand. In my example (diagram), my "unwanted current loop" goes:

wall plug -> laptop -> USB cable -> USB plug on PCB -> display -> Jack plug on PCB -> Audio equipment -> Laptop -> wall plug

while it should go

wall plug -> laptop -> USB cable -> USB plug on PCB -> display -> USB plug on PCB -> USB cable -> laptop -> wall plug

That was my first thought too. Strictly, your problem is voltage drop developed across the laptop-audio ground conductor when the display's current flows there.

Increasing resistance of the display-audio ground connection reduces this problem.
OTOH, the difference between display and audio grounds increases and now this link will suffer more noise.

The ideal solution is for the display to have its own PSU or at least some internal separation of USB and audio grounds.


So that's your X problem, and what is the Y actually?

[wasedadoc]

@wasedadoc: agreed! I put a LC filter which cuts at 3.3kHz in front of the display. That removed the high frequency stuff (probably good for EMC), but did not really make any difference sound-wise. Which makes sense since the display's PWM frequency is at about 120Hz.

To deal with 120Hz you need large capacitance electrolytic(s).

[Saimoun]

@magic: oh yes very good point - if the buzz noise comes from that Jack cable then yes the 10ohm resistor makes no difference => there is less noisy current going through it but the voltage noise is the same (less current going through more resistance = same voltage like you said).

But in most use cases, the buzz audio noise is coming from further, usually somewhere inside the audio equipment. And reducing the noisy current going that way helps a lot.

What do you mean with the X problem Y thing? is that an expression?

@wasedadoc: yep I tried that - even 470uF does not remove it completely (but it tames it quite a bit). Then I would also need to handle the inrush current.

[JohanH]

This is the issue with single ended audio signals. A similar issue exists with audio gear that have grounded chassis. I've overcome this issue by not connecting amplifier signal ground directly to chassis. A usual configuration is to connect the incoming audio signal ground to amplifier chassis at the incoming RCA jacks. This caused 50 Hz noise in speaker output. By lifting the audio signal ground from chassis with a 100 ohm resistor and 100 nF capacitor in parallel, the noise disappeared.

[Saimoun]

Yep that makes sense

[jonpaul]

120 Hz buzz:

1/ Show the noise on a scope to see wave shape.

2/ Audio signal level noise/buzz vs peak/avg/min signal?

3/ PSU is SMPS or linear?

Buzz is a 120 Hz pulse created by the mains rectifiers. Sometimes a RC snubber across the rectifier can reduce.

4/ Identify is the a CM or DM mode noise?

Good luck


Jon

[wasedadoc]

Buzz is a 120 Hz pulse created by the mains rectifiers. Sometimes a RC snubber across the rectifier can reduce.

The OP appears to be in Denmark so rather unlikely the 120Hz is created by mains rectifiers.

[Saimoun]

@Jon: the noise is from the display, not from anything else. The display is basically a bunch of LEDs which are PWM'ed, and the return current of these LEDs goes through the "wrong" GND path instead of going back the USB path. If there is any single-ended audio on that path which will use this GND as reference, then you can hear the noise.
If I turn the display off the noise completely disappears.

As for whether that is CM or DM mode I am not sure   

I solved the problem adding those 10ohm resistors between Jack's sleeve and GND. I understand this is not ideal, especially for EMC and I also understand that it means any external noise (i.e. not the noise from the display but any other noise coming from somewhere else) going through the Jack will be amplified because of this. Thank you magic for pointing that out   But the device can handle quite a bit of noise on its inputs without being affected.

Maybe I'll make a new version with a huge cap in front of the display with an in-rush current limiter in the future.

Thank you everybody for helping me with this in any case 

[jonpaul]

Look at the DC bus, see if any pulses at the video rate. If yes, try à large ELCO cap on the LED display DC bus.


Better than the  resistance 10R0, an inductor, perhaps 100 mhy 1A 1 ohm or less.

Jon

[Saimoun]

yes a large cap on the display DC bus - what I said in my last post 
(I'll just have to manage the in-rush current).

As for the resistor, are you saying I should put a 1 or 0.1 ohm instead of 10 ohm? And why 1A?

[jonpaul]

resistance same impedance regardless of frequency

inductor impedance proportion to frequency

120 Hz buzz implies a pulse at 120 rate, thus many harmonics.

Suggested replace the 10 ohm R with a,carefully selected inductor

Jon

[Saimoun]

Ha ha ok, I just had to understand your "100 mhy 1A 1 ohm", you mean an inductor of 100mH 1A 1ohm?

It's an interesting point! 100mH sounds like a LOT though.

[jonpaul]

at audio frequency and mains  frequency inductors of 100s mhy..henries are common.

We have small,axial 1 Hy

Old valve PS series chokes were 10s of Hy

Reactance is Xl= 2 pi F L ( Hz, Hy)

jon