Odd Problem

[pjanda1]

I just got back from a road trip with my chip amp (stereo single chassis LM3875, BrianGT NI board, with passive volume control and input switching, not snubberized). It emits FM radio stations at low levels from the speakers. The odd thing is that it only comes from one channel at a time, and the noise alternates from channel to channel based on the input selected. It does not matter whether the input jacks are unused, connected or a source is playing. I swear it didn't do this before the trip.

For a little history, there has always been some noise through both channels with the pot (1M linear PEC law faked with 33k resistors) completely open or closed (outside of useable positions.) The noise was not audible as an FM station and the new noise is. It was previously pretty quiet in most of the pots range, though not silent. The input selector (grayhill, just switching +'s, common -'s) has always had some bleadthrough, I just assumed that was part of the design. None of these things was a problem for me.

The amp was very noisy at my friends house (on the trip), but I blamed that on dirty power. (Multiple audio systems, computers, an aquarium, appliances, etc) In retrospect I'm guessing that whatever happened occured on the way there.

It's possible that I've always had this problem, but highly unlikely. I've recently started using much more effient speakers and the ambiant noise in the house is low this time of year. I'm still awfully sure the problem didn't exist before. I'm also sure I have the selector wired correctly, so I can't explain how the noise would switch from L-R based on the input.

Any thoughts?

Paul

[wes-ninja250]

Are you SURE it's FM, not AM?

I can't imagine how FM could possbily get in, except from an FM tuner.

AM can leak into an amp with high gain and long, floating input leads.

Actually, the easiest AM radio to build is an LM386 with the gain pins shorted (200x gain) and a ~16" length of copper wire hanging off the input. The only problem with THAT radio is it only gets the strong station in the area.

Wes

[Russ White]

Hmmm let me see if I can help.

A 1M pot is going to inject a lot of resistor noise, and help make the amp prone to RF noise. I would use either a 10K log pot or a 50K law faked pot.

You coulld add a 330pf cap to signal ground right after a 1K resistor in series with your input. This will shunt RF frequenties to GND by injecting a low pass filter on your input.

Cheers,
Russ

[pjanda1]

It is most definately FM. I can hear the station faintly, either 99. something or 100. something. Russ, I don't mean to doubt you, but I'm awfully sure that this problem began suddenly. When I was first testing the amp I spent some time with my ear right in front of the speakers and I couldn't hear anything like this. Then after it's little road trip this happened out of the blue. If it were the fault of the pot value, would it not have been there all along?

thanks for your help,

Paul

[Russ White]

You might be surprised how quickly a pot which seemed fine can go south.

Honestly a 1M pot is very large for the application. Even if it were an excellent pot you would get quite a lot of resistor noise. Also, by using sucha large pot you decrease the current going to the input stage, which makes picking up stray signals by the opamp much easier. I speak from experience, as I have had issues very similar to yours. They can seem to spring up suddenly, but I can tell you that your configuration (if I understand things correctly) was never optimal.

Cheers!
Russ

[carlosfm]

Originally posted by pjanda1
It is most definately FM. I can hear the station faintly, either 99. something or 100. something. Russ, I don't mean to doubt you, but I'm awfully sure that this problem began suddenly. When I was first testing the amp I spent some time with my ear right in front of the speakers and I couldn't hear anything like this. Then after it's little road trip this happened out of the blue. If it were the fault of the pot value, would it not have been there all along?

Russ pointed you in the right direction.
By coincidence, you have been on places where radio signals are strong.
Again by coincidence, maybe some radio station on your area ingreased the power.
What you can be sure is that you should not use a 1M pot, long (and probably unshielded) signal wires and (maybe?) incorrect grounding arrangements.

[wes-ninja250]

Sort of O/T -- sorry -- but can anyone explain how an amp could pick up and demodulate FM into a recognizable signal?

AM is easy, current induced in wire, chopped by diodes or transistors, mega gain, boom, speaker reproduces it. But FM demodulation is a little more complicated!

Is it possible that maybe the noise is coming from a leaky radio nearby?

Wes

[pjanda1]

I'm using 1M shieled (Belden 89259) IC's, shielded RCA connectors, and only about 4 inches of wire between the outside world and the pot. The noise can't be picked up in the IC's, because it happens on unused inputs too. The chassis is solid aluminum. The grounding is per BrianGT's recommendations, but I was planning on trying Carlos's scheme the next time I had it apart.

How is it possible that with the pot in the same location, the noise is always from one channel only, and that channel changes based on the input switch? Surely were it the pot the noise would either be from one or both channels at any given pot setting. You guys know far more than I; I just want to be sure I understand the problem.

I am willing to replace the pot, I just think it is cool. I appreciate all of your advice, I was just hoping the answer would involve keeping my pot. If there is a chance just switching grouding schemes would fix it, I'm all ears.


Paul
I assume this still describes the prefered grounding arrangment:
Carlos grounding Scheme

[Russ White]

Unless you are getting humm or some other indication of ground loop(nothing you have described so far would indicate that) changing your grounding scheme alone will not make any change at all.

You cannot look at the pot in isolation. Notice my second tip. Some places just have a lot of RF and EMI, the low pass filter I described will shunt that to GND.

Give it a try. Its easy to do/undo.

Cheers,
Russ

[carlosfm]

Originally posted by pjanda1
I'm using 1M shieled (Belden 89259) IC's, shielded RCA connectors, and only about 4 inches of wire between the outside world and the pot. The noise can't be picked up in the IC's, because it happens on unused inputs too.

The unused inputs will serve as antennas unless you terminate each input with a resistor to ground, on each RCA plug.
The length of wire is as much important AFTER the pot. Specially with a 1M pot.

Originally posted by pjanda1
I am willing to replace the pot, I just think it is cool.

It is not cool.
Besides the problems pointed out by using high impedance pots, the lower impedance ones (10~20k) also SOUND much better.
Sometimes it's not a question of picking the 'best' (or fanciest) parts, it is just a question of using the ones with the right values for the right place.