# Why does GSM cause speakers to buzz?

Based on numerous internet resources, speaker wire acts like an antenna which picks up the transmitted signal of nearby cellphones and causes the speakers to buzz. But I'm not really buying that...

A 3.5 mm speaker cable is designed to carry 1 V. I've seen old setups where PC speakers are powered directly from the 3.5 mm jack (and I've tested playing unamplified sound directly from a PC through the jack, although the volume in my setup was not very high at all). How can the tiny bit of EM emitted by a cellphone radio cause a speaker system, designed to operate off of a fluctuating 1 v signal, produce such a loud buzzing noise? I couldn't imagine the EM generating more than a few micro-volts in a receiving antenna. Am I wrong?

Thanks.

Updated - corrected voltage of line out to 1 V (see comments)

Update I looked it up, and yes it seems GSM transmits at 2 W. I'd like to do a sanity check with that figure to verify some of the answers which state that the transmitted power is significant. My physics is quite rusty, but I'll try...

We know that the intensity of EM radiation around a source is:

$$I = \frac{P}{4\pi r^2}$$

So let's say we have a wire 2 m long and 0.2 mm wide (I hope this is a valid approximation for the wire) that is approximately 2 m away from a transmitting GSM module.

Then for $P = 2 W, I = 39 \frac{mW}{m^{2}}$

Multiply that by the surface area of the wire (0.2 mm * 2 m)

The total EM power along the wire is then 16 $\mu W$.

Like I said I'm quite rusty, but is this not correct? Is this really significant enough to produce that sound without being amplified somehow? Perhaps the signal resonates? Or interferes directly with sound cards?

-
First standard line-out is 1 V RMS. Next, there isn't a tiny bit of EM. When the phone is establishing connection, the radiation is pretty strong. –  AndrejaKo May 28 '12 at 19:52
@AndrejaKo Yes, I could be wrong about the 5 V... - now that you mention 1 V sounds correct. I'd still like to know just how much voltage gets induced by GSM, as I could not imagine it being comparable to even 1 V. I am wondering whether the GSM signal is somehow being amplified within the the speaker itself. –  Bizorke May 28 '12 at 20:02
The length of the coil windings on the speaker may well be quite a bit longer than 2 meters. A air core transformer equation might give a closer approximation to the energy transferred, if the GSM phone is oriented appropriately. –  hotpaw2 May 28 '12 at 22:25

The buzzing is AM detected signal.

The reason of audio amplifiers being hit by GSM signal is that contemporary audio semiconductor parts are actually very functional up to high GHz range. For GSM-800-900MHz range any 80mm copper trace works like 1/4 wave antenna, or stripline resonator. The signal is AM detected on any non-linearity (transistors or diode structures in chips) on multiple points of amplifier simultaneously, also including power regulator chips and so on.

It is translated into audio range as tiny but very sharp and periodic dips or pops of averaged conductivity of non-linear parts (AM detection), which are DC powered.

Think of low speed oscilloscope trace showing straight line with beads of UHF flashes. Simple sharp spikes of consumed DC current will become audible with amplifier.

-
This actually ties into the underappreciated fact that even if the output of an amplifier will ultimately be filtered to audio frequencies (a typical speaker isn't going to vibrate appreciably at frequencies over 1MHz, after all), the presence of higher or lower frequencies on the input may cause very-audible distortion. A 869.0000MHz signal amplitude-modulated at 500Hz will contain components at 868.9995MHz and 869.0005Mhz, neither of which is audible, but distortion may produce frequency components at any sum or difference of any multiples of the original frequencies. –  supercat May 29 '12 at 16:22
An easy way to envision the problem is to imagine an amp that clips the signals, before any filtering, at 1.2 volts on the high side and -1.1 volts on the low side. A strong 868MHz signal (e.g. +/- 2 volts or greater) would effectively force the input stage of the amp to see 0.05 volts when it's present; if no such signal is present and the input is otherwise silent, it would see 0.00 volts. Thus, a strong 868MHz signal which appears and disappears at an audio rate would cause an apparent 0.05 volt peak-to-peak audio-frequency wave on the input. –  supercat May 29 '12 at 16:26
why does the effect only happen when a phone in nearby? Is the phone transmitting when it gets a call and then only the transmission is being picked up? –  Nate Nov 23 '12 at 3:40

In my work on hearing implants I found that the buzzing is caused by magnetic radiation from the battery wires in the phone which carry pulses of maybe 2A. This magnetic field is relatively intense and can couple into low frequency EMI susceptible circuitry of nearby devices. The noise in my case wasn't from the RF and antenna at all. This is why RF shielding may not work.

-

First of all, it is the power that matters in this situation. Any small speakers are going to be only a few watts, if not loss.

The power out from the handset in cell applications can be as large as 33dBm (or 2 Watts). This is the case for both UMTS and GSM; however, for GSM there are bursts that are around 217 Hz (which is in the audible range) This 2 watts of power can be very strong compared to audio signals. In UMTS, the protocol changed and was specifically designed to avoid bursts that would be in the audible range.

-
Kellenjb, I've done a sanity check and updated the results to my question. Could you please take a look and tell me if I am in error? –  Bizorke May 28 '12 at 20:59
Yes. Thje GSM protocol design was brain dead and the implications were not spotted until too late in the process. They knew what was going to happen but there was enough \$ invested by all concerned to allow them to ride roughshod over regulatory authorities. Any one company who had this sort of issue usually would never get past type approval. GSM was a 6000 pound gorilla (that's 10 x the usually overwhelming 600 pound ones). The interference is an accidental subharmonic related to how signalling frames are built. –  Russell McMahon May 29 '12 at 1:51

The buzzing you hear is interference. Put ferrit beads on your cable to get rid of it. Here is how. As AndrejaKo pointed out, the radiation is significant since it can peak at about 2 Watts (for headsets).

-
Hi Suha, I understand what the sound is and how to get rid of it. However I don't understand how it's possible that 2 W is being induced into the cable when a GSM device is near-by. Can you see my edits to the question and let me know if you see any errors with my calculation? –  Bizorke May 28 '12 at 20:58
One up for the neat formulas :). –  Bizorke May 29 '12 at 13:56
@Bizorke I don't understand, they didn't answer your question but you gave them an upvote for neat formulas? I don't even see any formulas. What am I missing? –  Kellenjb May 29 '12 at 17:15
@Bizorke, the 2 watts is peak power on antenna, not on the cable itself. A ferrite is not going to absorb 2 watts. It gets loud because it couples into the signal lines previous to the amplification. –  Kortuk May 29 '12 at 17:16
@Kellenjb i edited the formulas in the original post. If Bizorke wants to thank me for that, i'm fine with it. If you don't, i'm also fine with that too. –  suha May 29 '12 at 19:10

If your speakers buzz every few minutes when the radio is off, then it is the auto-sync to the nearest cell tower that is the cause. This burst is to ensure you stay connected to the strongest tower to receive calls.

If the above is true, then the solution is to add a small RF cap, 100 pF ceramic may work nicely, check ESR and choose the value that gives lowest impedance at 850MHz.

Reason: The speaker coil happens to make a nice antenna and the magnetics act in a non-linear fashion with eddy currents being rectified and causes EM force on the coil to be heard. Normally >3m distance should not be detected. The protocol for offline sync uses baseband signals when AM detected by the speaker coil. but offers better SNR and reliable connection. ADSL has the same issue and the bandsplitter is designed to filter that out. Otherwise you would hear the protocol to connect using autobaud at broadband speeds. The GSM problem is not evident with online connections as the modulation avoids baseband content with continuous carrier.

I do not know the Q of your speaker coil, but if you can hear it in a passive mode, it works pretty well.

If my assumption is wrong then it may be the other sources as mentioned above.

As I said the TDMA is AM detected by the resonant coil and non-linear properties of the magnet at those freq. Can you respond to my Q? The coil length is not as important as the radius.

-
Bizorke< GSM Edge 2G gets this interference from its TDMA connection protocol. Whereas CDMA bursts used by Verizon do not. I believe you hear the burst rate in the beat of "dit-ditty-dit-ditty..." for a few seconds. The TDMA frame rate during "off-hook" is 216.7 Hz frame rate is the "dit" burst, when it sends "Can you hear me?" The 1/4 wave of GSM (@850MHz) carrier is 89mm or 3.5" which puts speaker coils in the range of reception. Can you please confirm my assumption on your test conditions with audio power off? Loud audio may kill the noise by disturbing the resonance. –  Tony Stewart May 29 '12 at 5:39
@Bizorke et al, the clue to the accuracy of my answer is the sound of the buzz which may be quite loud and annoying vs a clean 200 Hz tone. This proves the coil is not just moving axially as if driven by the electronics, but radially rubbing on the magnet and making a spring buzz. That is why I asked if it buzzes with the power off and speaker disconnected. –  Tony Stewart May 29 '12 at 8:06
I have not observed this with the speakers off or disconnected. One of my hypotheses was that GSM somehow interfered directly with sound cards. I also do not have the specifications for my speaker available, but I agree that the speaker coil itself will contribute considerably to the overall energy absorbed by the system. –  Bizorke May 29 '12 at 15:44