Cell phone noise on my speakers...

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
High everyone... I need some help filtering some chirp I am getting from a cellular modem I have built into a product. When the model transmits I get a chirping on my speakers.

I have tried shielding the cables. It isn't on the power supply line (used a 12V battery and REALLY short cable) or the Audio input (grounded the input) to the power amp. Its on the output and so far I have been unable to shield it to make it any better. Could it really be in the coils or maybe I should try using a class a/b amp instead of the class D I have now.

any help here would be appreciated.

Tony
 
This is difficult to advise on given that there is very little given in the way of details of the system.

Possibly you could pipe the output RF (using a co-ax) to a remote antenna in a less sensitive location? If you can get it outside the building (or vehicle) this can be a big help as not only does it decrease the signal seen from the antenna it often results in the cellular system using less transmit/receive power. It helps to know the relative location of the base station. You don't say if the installation is mobile. An antenna with a beam has a real advantage when fixed.

The absolutely conventional approach to this is to apply ferrites to the cables, but this is often unsuccessful. More strenuous efforts involve the use of metal boxes for shielding and pass-thru capacitors.

A switch to a different amp may work, since we have no clue as to how the signal is appearing at AF.
 
I get the same thing, zzitt zzitt from the flat above me, turns out its the lasses Samsung, though her bf's I-phone has no effect.
I asked her if she could turn it off the one day to see IF it was her phone, she did and the RFI went away, two mins later she turned it back on and zzitt zzitt....... now when she's home, she turns it off, not that there's much point in having a mobile in my village as we live in a reception black hole.
 
This is difficult to advise on given that there is very little given in the way of details of the system.

Possibly you could pipe the output RF (using a co-ax) to a remote antenna in a less sensitive location? If you can get it outside the building (or vehicle) this can be a big help as not only does it decrease the signal seen from the antenna it often results in the cellular system using less transmit/receive power. It helps to know the relative location of the base station. You don't say if the installation is mobile. An antenna with a beam has a real advantage when fixed.

The absolutely conventional approach to this is to apply ferrites to the cables, but this is often unsuccessful. More strenuous efforts involve the use of metal boxes for shielding and pass-thru capacitors.

A switch to a different amp may work, since we have no clue as to how the signal is appearing at AF.

I did pipe the RF out. I have a redundant sytem (GSM Modem and WCDMA Modem) and both are coaxed. One is silent the other is noisy. One is in an all metal enclosure and the other is in a plastic enclosure but the RF module is shielded inside. Both use a modified patch style antenna and I have a fairly good farraday cage around the system.

In a former life I was an RF engineer and audio is king of a hobby. I have put a lot of systems through qualification testing FCC/IEC but this one is stumping me.

A 20W PA should not be giving me this much grief.

Tony
 
Can't turn the system off... Guess I'll try and build a RF filter and try and roll everything off above 100Khz and see if that works.

I appreciate everyone taking the time to reply.

What I really am hoping for is for someone to say... I know exactly how to fix this ;)

Tony
 
RF is funny stuff, sometimes you send it outside and it crawls right back in the shack on the outside of the coax. This is particularly prevalent with balanced-fed antennas. You end up with undesired feeder radiation.

At a lower frequency we'd try a choke balun (not really a balun) made up by coiling up a load of coax (20 feet or so) on a plastic pipe just below the feed point. Single layer closewound. You might get away with something like this at higher frequency if the coax is thin and flexible.
 
You sound like you should know better than most of us. For RF, everything is an input, including output, power in, and input, plus all of the antennas formed by the circuitry.

The RF will get rectified by any PN semiconductor junction that it gets to. Stray (or not) capacitances can then make its DC-equivalent level higher, or act as an envelope detector or demodulator. Depending on the RF and the circuit, it could do anything from subtly changing a DC operating point of a transistor circuit inside of a chip somewhere, to producing audible output pulses (or intelligible audio, if the RF was AM).

The bad offenders are often the antennas formed by the conductor pairs with "enclosed loop area" in the audio circuitry. (And grounding the input doesn't necessarily remove the possibility that it's happening there. Maybe the ground had a high-enough impedance at RF frequencies.)

Basically, EVERY audio input should have an RF lowpass filter (200-300 kHz f-3dB should be low-enough).

BUT, you ALSO need to be sure that there is absolutely as little physical gap as possible between the signal input and signal ground conductors, ALL the way from the input jack to the input pins of the first active device(s), and then throughout the system. Same goes for power and power ground pairs, and AC mains input pair, and for the pair between the rectifiers and the smoothing caps, and for the output pair. If wires, they should be tightly twisted together, and then also shielded if possible. But that means twisted pair inside a shield, NOT running the ground on the shield! If PCB traces, use a plane for one of them, on one side of the board. Otherwise, run them right next to each other, ALWAYS (i.e. everywhere).

Speaker wires are a nice big antenna. Are yours twisted pair? If not, try that. But also make sure the pairs going to the jacks from the circuitry are twisted. Also, is there a low-value output resistor in parallel with a low-value air-core inductor at the output? That can help to not let the RF come in through the output. RF that goes in through the audio output will typically go right around the feedback loop and get into the negative input pin of the active amplifier portion of the circut.

You can look in chapter 6, here: ADI - Analog Dialogue | Op Amp Applications Handbook .

Wavelenth in meters is 300 divided by frequency in MHz. So 1800-1900 MHz has a wavelength on the order of six inches. So 1/4 wavelength would be about 1.5 inches. That would be an "ideal" antenna length, but much shorter could still work pretty well. If you don't have power and ground planes and no wires anywhere, you might want to try adding a low-value resistor in series with the power and ground conductors, just before each of your decoupling capacitor networks, to increase the RF attenuation where the power enters each chip pin.

Also, any component that connects to an IC pin should connect right at the pin, or extremely close to it.

Good luck.
 
Last edited:
Got it fixed... It was two different offenders. First there was a coating on the sheet metal preventing a good RF ground for the antenna (biggest contributor). Then We had to reroute the output cables in some channels (Cu tape foil over bare sheetmetal for a good ground).

Before we did this I tried most everything EMI filters on the input, put a battery on it to remove power supply noise, grounded the audio inpt and turned the gain up (still heard the noise), CM and differential chokes on the output (LC PI low pass filter with a 10MHz corner) at the speaker and then tried moving it to the output of the audio amp.

None of these worked... That's when we noticed the powder coat issue on the antenna mount.

I would like to thank everyone here for taking the time to respond. It was deeply appreciated.

Tony
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.