Noise Isolation

[StudentSA]

Hi All!

I have a Raspberry Pi 3 that I am trying to use together with an FM module called the TEA5767 i.e. the 5V, SDA, SCL and GND are all from the Pi's GPIO.

The issue I am having is that there is a background ticking noise from the output. Sounds almost like a helicopter is the distance.

So, I have been trying to isolate the issue and these are my findings:
1. If I shutdown the pi, the noise goes away. i.e. the pi still supply 5V but the processor etc is off.
2. If the Pi is on and I disconnected the SDA and SCL lines, no effect, noise still persists with the module located near or far.
3. If I connect only Vcc and GND of the TEA to another isolated 5V supply. Noise is gone when more that 10cm away from the pi. As I move the board over the pi the ticking sound returns. the closer I get the louder it ticks
4. I tried wrapping the tea module in foil as much as I could but still noise.

Any advise is appreciated.

 

[dknguyen]

Sounds like noise from the switching supply of the Raspberry Pi. I'm not familiar with it but I assume it has one onboard somewhere. Not sure what you can do about that other than to shield the board.

It sounds like the noise is strong enough to be picked up even if the TEA is completely disconnected from the Pi so just galvanically isolating the power supplies and I2C lines won't be effective unless you do that and just keep the board separated from each other. Or it might be simpler to try filtering the power and I2C lines between the boards with some ferrite before galvanic isolation.

EDIT: Yeah, there are SMPS everywhere on the Rapsberry Pi 3. The SMPS all seem to be either 1.5MHz or 3MHz switching so that's probably your target for filtering.

 

[AnalogKid]

While you cannot hear a 1.5 MHz noise signal, you can hear it go on and off. If one of the power converter circuits is running in a "discontinuous" mode for better efficiency at low output currents, this radiates as an ultrasonic noise carrier that is amplitude modulated at an audible frequency like a few Hz of above. You have confirmed that the source is radiated noise from the pi, so your main option is to shield things.

ak

 

[dknguyen]

While you cannot hear a 1.5 MHz noise signal, you can hear it go on and off. If one of the power converter circuits is running in a "discontinuous" mode for better efficiency at low output currents, this radiates as an ultrasonic noise carrier that is amplitude modulated at an audible frequency like a few Hz of above. You have confirmed that the source is radiated noise from the pi, so your main option is to shield things.

ak

I suppose that does introduce the possibility of artificially loading the offending SMPS so it always stays in continuous mode...either with resistors or just programming the raspberry pi to work harder in the right way so more current is drawn from the appropriate regulator.

 

[dr pepper]

So what are you using for audio output, headphones would be a good quick check.
A choke in the supply to the module and a couple of caps might do the trick.

I built a receiver a while back, I wasnt happy with this module and I ended up with a Si4703, which for me was much better and it supports Rds.
I dont remember having trouble with noise on the '5767 though, but I was using a 'duino not a Pi.
I save old tuner cans from dead tv boards, they make nice little Rf cans for these kinda things.

 

[StudentSA]

Thanks guys, So yes I am convinced that the Raspberry Pi is somehow emitting this noise. Also the fact that the noise "sound" changes with computing load that the pi is handling.

dr pepper , would you mind sharing a rough schematic of a suitable choke circuit? I will also try to find an old tin jewelry box and see if placing the TEA in it helps.

I came across the Si4703 but could not find an "easy" ready to go solution. So opted for the TEA in order to save time... how that has worked out...

 

[unclejed613]

on the Raspberry Pi reddit page, the problem comes up a lot. the solution seems to be putting the Pi in a metal case. there's a lot of signals in the pi, and i'm sure some of them are in, or have harmonics in the FM band.

 

[StudentSA]

Found it!
The "helicopter/ticking" noise is being induced by the WiFi of the RPi. If i switch of the WiFi interface using:
sudo ifdown wlan0
The noise stops.
Still unsuccessful with shielding against this... I must say RF is black magic.

 

[unclejed613]

RF isn't black magic. RF follows the laws of physics. one of those is the inverse square law, which means the field strength of RF is proportional to the inverse square of the distance from the radiating surface. you can try using a long cable to the TEA board (be sure to use ferrite collars on the cable to keep the RF from being conducted through the cable)to get the board out of the near-field zone of the wifi interface.