PA picks up AM radio

[Peter Tryndoch]

Hello,
I have a customer where I installed a PA interface.
It consists of 2 devices located in a data cabinet:
An IP interface that connects to the LAN via cat5 cable to a switch. The Output is analogue audio via a 3.5mm plug. The device is an Algo if that helps.
This connects to audio-in of the PA amp (Redback) and the output is via fig 8 cable to transformer speakers.
It picks up AM radio. I was able to fix it by running a cable from the earth pin on a GPO below the cabinet to the cabinet itself.
So I asked the customer to get the cabinet earthed (I'm not an A-grade). So this was done but apparently it still picks up radio!
So my first thought is to check if the earth cable is making proper connection. But since I need to make a site visit, I need a plan B.
I did try a ferrite core albeit a small one, so I will take a beefier one with me, but I don't hold out much hope for that doing anything.
One of my associates suggested replacing the fig 8 speaker cable with shielded cat6 cable, but that's a major undertaking and I want to avoid it if possible.
I was thinking about some capacitors on the PA output, each leg to ground.
Will that work and what value of cap would be the best.
Cheers

[DaJMasta]

Maybe a bit tricky to pick a single solution without knowing the circuit, but there are definitely things to try and you're definitely trying some of them.  Do you know if they're in proximity of an AM radio antenna?  It could be that the fix you've tried is fairly effective, but the signal strength near them is much stronger and it just isn't sufficient.

If you're hearing the radio, somehow the signal is finding its way into the amplification path, and because all you need is an envelope detector to demodulate AM, the parts required are often already part of amplifier circuits.  A low pass filter (ferrite with a few turns) on the input signal line, you can make it fairly resistant to picking up radio signals, but since AM is a lower frequency than FM or TV signals, you will need more turns (more inductance) to have the same effect.  A bunch of turns around a ferrite on the signal input, close to the amplifier device should be the most effective in terms of keeping the radio signals out of the signal going into the device.

Good grounding is also a critical step, making sure that both device's grounds are at the same potential and have a low impedance path to earth will go a long way towards helping and preventing chassis from acting as their own antenna or whatnot.  Shielding the signal cable is a good reaction, but rather than cat5 or similar, I'd probably look at shielded XLR cables or something - something with a shielding braid or foil with a good connection to ground should dramatically reduce what's picked up on the signal cable, and a lot of 3.5mm connector cables aren't shielded.  If it's a mono signal, almost any kind of coax could do the job so long as the connectors are easy to deal with.

I doubt caps at the PA output will help.  While caps (effectively a low pass filter) could be potentially a benefit on the input, by the time the AM signal is in the audio band (demodulated) and is audible, the frequency content is already too low to be filtered out without effecting sound quality - so any filtering of the audio signal of the AM would have the same effect of the intended audio signal of the PA.  It's going to be much easier to deal with the spurious signal before it's demodulated, when it's up around 1MHz and well away from the audio band.  To that end, a small cap between the signal and ground of the PA's input could actually be useful for filtering, but if there's enough signal picked up on the ground wire as well (if the grounding isn't good), the signal won't be filtered out because it will be differential across the two connections.  The actual value of the cap would depend on the source impedance of the input signal, but something in the ballpark of 10s of nF is where I'd start if I were to try.

Maybe there are some more audio/radio types that can offer more specific solutions that have worked for them.

[rcbuck]

I installed PA systems for over 20 years and have seen the problem on some occasions. I will make a couple of suggestions.

As DaJMasta said, caps at the PA output will not help.

Monitor the output of the IP unit with a butt set in monitor mode. If you do not hear the AM station at the IP output, you know the problem is wiring related.

I assume you are using something like the 8301 Algo unit. Make sure the unit is placed as close to the PA amplifier as possible. The output of those units is a balanced output. Are you using a balanced 600 ohm input circuit for the amplifier or connecting it directly to the AUX input of the amp? If the latter, use a 600 ohm to 10K transformer between the IP output and the AUX input. Make sure the transformer is within a couple of inches of the AUX input. If the transformer has to be more than a couple of inches from the amp, use shielded cable between the transformer and the amp input.

If you are feeding the IP unit output into a balanced 600 ohm amplifier input, connect a 10 nF capacitor from each side of the line to ground right at the amplifier. If the IP unit is any distance from the amp you may have to do the same at the IP output termincals.

Let us know if you are able to get rid of the RF pickup.

Edit: I just re-read your post and saw you were using the 3.5mm output. That is not a balanced output so only put the capacitor from the center pin to ground of the jack. Same thing applies at the amp input. Also, make sure the connection from the jack to the amp uses shielded cable. It would probably be better to use the balanced XLR output jack to feed the amp instead of the 3.5mm output.

[Peter Tryndoch]

The more I think about it the more I question my memory.
I'm pretty sure that the output from the Algo is 3.5mm, but the input to the amp might be rca (mono)
I need to find out the exact models of amp and Algo and see what i/p & o/p options I have to play with.
The 2 devices are right next to each other and the lead between them is probably 1/2m if that.
And they are located in the data cabinet which should provide some shielding earthed or not.
I have 3 of these setup on this site and only one of them is misbehaving.
I'm guessing that it's AM because it is a talk back station and at no time did I hear any music. That's so AM!
The lead between the Algo and amp is most likely centre conductor with shielded outer, is it any point in trying to wrap a ferrite core around it? If so I have a number of large ferrite cylinders (2 parts clip together), would they be as effective? I will try to monitor with my butt (haha setup for a joke there) My trusty Telecom Mk2.
I once told a trainee to stick his butt on "that" pair. The customer got a good chuckle out of that one.
The switch is also on the data cabinet, so is the cat5 patch lead. And the feed to it is fibre.
So really the only things poking out are the 240V and speaker cables.
However the front door of the cabinet is smoked perspex. And the whole building is metal clad.
Go figure -  why this building out of all of them!
PS: thank you for your suggestions, I'll take them all in.

[johnkenyon]

Sorry to ping off at a tangent...

Many moons ago I had a (neolithic) 1970's Rotel transistor hifi amplifier which was picking up an AM transmission - loud enough to hear, but not loud enough to determine who or what station.
The AM signal was there regardless of which input was chosen (so not the phono pre-amp, or anywhere before the input selector.)
You could only hear it when there was no input - the "wanted" signal otherwise swamped it. The volume control didn't affect the level. (so it was being demodulated somewhere inside the amplifier between the volume control and the power amp input.)

Being a lazy student, and not wanting to reflow every single solder joint and not being that good at identifying dry joints I thought "Hang on - a radio needs an aerial", and then bought two ferrite rings and wrapped the speaker cables around them close to where they connected to the amplifier.

This fixed the problem - my conclusion was that there was a HF path through the speaker wires through the amplifier back to the component/duff solder joint that was demodulating the signal and then feeding a very low level audio signal into the power amp which was audible when there was nothing else there to swamp it.

I was using vanilla 8 ohm speakers, but the same idea would work on the wires from the amp to the loudspeaker drive transformer.
Note that clamping a ferrite core on the cable won't work, and looping the cable around a clamp on ferrite core probably won't work either - look at the diagram at the bottom of https://rsgb.org/main/technical/emc/using-emc-filters-and-ferrites/

HTH
/John

[amyk]

I've seen installations receive AM radio when off. If you're that close to the transmitter, the only solution is lots of shielding.

[GeorgeOfTheJungle]

I'd put a low pass filter at the PA input, it only takes one R and one C, can even fit inside the RCA. 1µF and 16Ω for example.

[Peter Tryndoch]

Today I was able to have a go and fix the interference.
The electrician wired the cabinet to an earth stake as the building does not have a distribution panel with an earth bar. It’s a portable of sorts with a mains connection on the side of the building which is fed from an adjacent permanent building. Interestingly when I ran a lead from the GPO earth to the cabinet, it no longer suppressed the radio! So I was off to a frustrating start.
First I disconnected the speaker cable from the amp and the radio was still there,
So I knew it wasn’t the speaker cabling.
I swapped the amp and Algo with another onsite and the interference was still there, so I knew it wasn't faulty equipment.
The Redback Amp had a good earth. I tried earthing the Algo to no avail.
I tried both a ferrite rod and toroid on the audio lead between the Algo and Amp – no change.
A 10nf (then 20nf) on the input to the amp reduced the interference a little but not enough.
The next step required some surgery so I pulled the PA amp apart.
I must say it was nicely constructed inside. Made in OZ (not Austria!), and before turning on the soldering iron I had to tape a plastic bag around the smoke detector.
I connected the 10k to 600ohm transformer with the 10k towards the PA amp = nope!
But then I connected the 10k winding towards the Algo (source) and hey presto – the radio was gone.
It hard to express how happy/relieved I was. Reading it her sounds like a fairly simple step by step process of elimination, but the reality is far from that, it took me hours to get to this point.

So I mounted the transformer inside the amp with:
10K A-leg to amp input
10K B-leg to amp ground
600 ohm A-leg to source (Algo)
600 ohm B-leg to ground

Interestingly the 600ohm winding had a centre tap which I also grounded.
I disconnected it and the interference came back, so it went straight back on.
At home and in hindsight I realised that this shorted half the 600 ohm winding.
I could have connected the shield of the source to the 600 ohm B-leg as it was affectively floating with no reference to earth until you connect it to the AMP. The rca sockets on the AMP were this pretty standard block formation, so I would have had to cut ground planes on the circuit board or drill and mount a separate isolated rca socket which I did not have on me at the time.
If I had this setup at home and all the time in the world, I could have probably done that.
But I was at a customer’s premises and time is a luxury, and I was burning plenty of it.
All of it a loss to my company.


A big thank you to everyone who chipped in with ideas. I could not have done it without you.

As a side note: Does anyone know if you can set the output gain on an Algo 8301 in software?
The volume could be slightly higher and the amp is already at max.

[drknob]

I highly recommend the Jensen Transformer Application Notes: https://www.jensen-transformers.com/application-notes/ especially note Jensen System Troubleshooting Guide.
Most notes are written by Bill Whitlock, my go to resource for audio power and grounding.

[David Hess]

I was thinking about some capacitors on the PA output, each leg to ground.
Will that work and what value of cap would be the best.

I would start with the capacitors unless I had something easier to try which might work.  100s to 1000s of picofarads is typical.  Keep the lead lengths as short as possible.

[Peter Tryndoch]

Well I thought all was well until I got an email from the customer telling me that the volume was too low.
I guess the isolation transformer with 1/2 the input winding shorted wasn't going to cut the mustard.
So another visit today and...
This time I tried the balanced output from the Algo into the other input channel of the amp which I had not connected the isolation transformer to.
Straight away - no radio. I dialed into the Algo from a nearby phone and feedback!
Hm, lots of lovely gain there.
So I turned down the amp from 9 (full) down to 7.
Audio is good, and some headroom for turning up the amp for quietly spoken people. And no radio!!!
So then I decided to do one of the other amps just for the extra gain.
I would have done the 3rd one (3 of 3) had my boss not moved the additional connectors and I could not find them.
Later I found them at his desk. He's a Magpie!
So maybe another visit to do the 3rd amp. Who's pocket is that coming out of, clever boy.