Stubborn noise issue

[zzattack]

Having spent many hours on an in-car noise issue without being able to resolve it I'm hoping someone can point me toward a solution.

The issue I'm trying to solve is a buzzing noise in a car audio install present only while the normal lights are on and the brake pedal isn't being pressed. This eventually led me to measure on the rear lights. When the headlights turn on, the rear side lights receive a 100Hz PWM signal, ~25% duty cycle and battery level amplitude (so the bulbs share a common ground). When pressing the brake pedal this signal changes to 100% duty cycle and the buzzing noise is gone. In my unlearned opinion this either means this noise is picked up and amplified, or it causes some kind of noise to be generated inside the amplifier.

I 'checked' the ground point quality for the rear lights because every car audio/snakeoil expert seems to claim 'you have bad grounds'. Now I don't know how to check ground point quality or even what kind of measurements would be considered within tolerance, but alas. With the 25% duty cycle signal to the bulbs there's some 10mV and <0.2? between the chassis and bulb grounds. Does this imply the ground is 'good'? Or would I need to actually measure some disturbance in the supply lines to conclude this?

Back the the origin. This PWM signal is generated separately for the left- and right-hand sides of the car. Both are generated by the body control module all the way in the front of the vehicle. RCA cables with the audio signal run along the right side of the car. This led me to believe the buzz is picked up here, but it is probably not. I have cut the wires at the body control module, and rerouted them with new wire, far apart from the audio signal wiring. Hooked them up directly to the bulbs in the trunk but the buzz is absolutely identical - so I guess it's not picked up by the RCA cables and amplified after all.

I'm unsure how to proceed so any tips are helpful. Hopefully someone has a methodical or theoretically supported approach to this kind of issue, I'd love to learn! By now I've come across so much youtube advise including 'picofuse replacement', powercaps, grounding directly to the battery, noise level filters, ground loop isolators, using an amp with high-level inputs and more. For what it's worth: yes, the RCA grounds are fine. Ground loop filters hardly affect the buzzing. My RCA cables are not low quality, and have both a shield as well as drain wire. And I don't want to use high-level converters.

[DaJMasta]

Don't know a ton about car power and audio wiring.... but car power supplies are nasty in terms of noise and generally the devices using them are designed to be really tolerant of noise and big changes in voltage... but it's probably still a likely source of noise.  Stuff coupled in through the power supply would also be amplified, so it would take a smaller signal to be heard on the output.

Maybe the thing generating the PWM signal is not well enough isolated from the rest of the system (or just the switching of the lights isn't), and that's throwing that noise into the main power rail which is just not being filtered out of the audio power system well enough.  Rather than checking the grounds, I would try looking at the power rails right on the input to the stereo and see if that tone you're hearing is present, then probing the power rail at one of the later output stages (maybe right before the speaker amp?) in the system and see if it's also there.

If that's the case, maybe you've got some caps going bad or some filtering which just isn't adequate on the power supply for the sound system.  If you've got an amp module driving the speakers that is separate from the thing generating the sound (radio, cd player, whatever), then maybe there can be some grounding issues.... but if it's all one module like in many cars, the grounding on the PCB should be taken care of and there should be no external analog grounds to fight with in the audio signal path.

[fourtytwo42]

OMG more sophistication and cost cutting with car electronics I was not aware of this new wheeze and I don't expect it's uncommon in new cars. Sadly I don't expect the body electronics guy's even thought how there clever tech might effect others so didn't bother controlling there edge rates, result something like my broadband problems see https://www.eevblog.com/forum/renewable-energy/unusual-method-of-measuring-pv-system-emi/

Unfortunately most car audio stuff I have seen are unbalanced coax systems so not ideal for rejecting EMI whereas pro-audio used to working in electrically noisy environments uses balanced transformer or psuedo-transformer coupled circuits to eliminate it. 

Your best bet is either physical separation if possible (move the signal coaxes as far as possible away from the interfering source) and/or liberal use of clip on ferrite EMI filters both on the source power cables and signal coaxes.

If the problem still persists a disruptive EMI filter will be required!

[zzattack]

car power supplies are nasty in terms of noise

I should have emphasized that the noise is just as present regardless of whether the engine is running. So wouldn't the power be pretty much noise-free if it's just coming from the battery?

Maybe the thing generating the PWM signal is not well enough isolated from the rest of the system (or just the switching of the lights isn't), and that's throwing that noise into the main power rail which is just not being filtered out of the audio power system well enough.

This body control module is hooked up to virtually everything in the car. It has 7 plugs each containign 10-25 pins. FWIW, replacing this module has no influence on the noise.

Rather than checking the grounds, I would try looking at the power rails right on the input to the stereo and see if that tone you're hearing is present, then probing the power rail at one of the later output stages (maybe right before the speaker amp?) in the system and see if it's also there.

Good plan. I could see no variation in the power supply signal of the headunit when toggling on the lights. Measuring at the amplifier's power inputs in the trunk revealed some ripple, see attached screenshots. First one with lights off, second with lights on. So there's some influence there. Maybe the amplifier's noise filtering isn't what it used to be?

If that's the case, maybe you've got some caps going bad or some filtering which just isn't adequate on the power supply for the sound system.

How would one go about tackling this? Add some external filtering or ditch the amp?

If you've got an amp module driving the speakers that is separate from the thing generating the sound (radio, cd player, whatever), then maybe there can be some grounding issues.... but if it's all one module like in many cars, the grounding on the PCB should be taken care of and there should be no external analog grounds to fight with in the audio signal path.

I do have an equalizer in the chain but even when removing/bypassing this the noise is unaffected.

[David Hess]

Whether the grounds are good or not does not matter if you have a ground loop.

Is the audio system all one unit?  If there are separate units and they use different ground points, then single ended audio signals between them may be corrupted.

If an antenna connection includes a ground path to the metal chassis, then this can also cause a ground loop.

[DimitriP]

Make model year country ?

[DaJMasta]

Having an amp that isn't part of the head unit as well as having an EQ in between could certainly be ground loop related, isolating at the input to the amp could be an option, and if that doesn't do the trick, moving it to the input of the EQ instead.

If the output amp where you measured the ripple is easy to access, maybe see if you can check ESR on the electrolytic caps for it?  You can always bodge in some extras, but if it isn't failed parts and is design related, it you'd probably need an inductor between the local regulation and the amp, and that could be tricky to splice in.

Less invasive options which could still do work is maybe throw a ferrite on the output line and try to get a few turns through it - just cutting down on the high frequency from the hard edges of a PWM signal may make a notable difference, and it involves no modification of boards.


I think a good place to start is an isolation transformer, though.  Since you've got at least three major elements in the signal path before it gets to the speakers, all with separate ground paths, it's very likely that what you're seeing is at least partly a ground loop issue.  Ferrites or filtering may still help some, but it could be that the problem is small enough that until some extra noise gets injected by those PWM signals, it's low enough not to be noticeable, but the presence of that added noise is actually just revealing the ground loop.

[zzattack]

Whether the grounds are good or not does not matter if you have a ground loop.

Each component has it's own ground connection to the chassis, but there's a few runs of long wire connected to the same ground. One being the FM antenna, others being the RCAs - but their signal low reference does not internally connect to the chassis ground.
Other than using isolation transfers, I don't think there's much more we can do to improve the situation here.

Is the audio system all one unit?  If there are separate units and they use different ground points, then single ended audio signals between them may be corrupted.

There's the headunit, right next to that a simple equalizer, and then there's an external amp in the trunk. The headunit and equalizer indeed output unbalanced signals.

Having an amp that isn't part of the head unit as well as having an EQ in between could certainly be ground loop related, isolating at the input to the amp could be an option, and if that doesn't do the trick, moving it to the input of the EQ instead.

I've done a test powering up the headunit and equalizer from a separate battery. This greatly reduces the buzz to a point where it's inaudible unless I turn up the gains to near-maximum. Does this mean the filtering in the headunit is poor?

If the output amp where you measured the ripple is easy to access, maybe see if you can check ESR on the electrolytic caps for it?  You can always bodge in some extras, but if it isn't failed parts and is design related, it you'd probably need an inductor between the local regulation and the amp, and that could be tricky to splice in.

I will follow up on this. The amp is accessible but I can't tell whether it is easily serviced.

Less invasive options which could still do work is maybe throw a ferrite on the output line and try to get a few turns through it - just cutting down on the high frequency from the hard edges of a PWM signal may make a notable difference, and it involves no modification of boards.

Around the output wire carrying the PWM signal to the bulb? At which end, where it's being generated or closer toward the actual bulb?

I think a good place to start is an isolation transformer, though.

Placed where?

Since you've got at least three major elements in the signal path before it gets to the speakers, all with separate ground paths, it's very likely that what you're seeing is at least partly a ground loop issue.  Ferrites or filtering may still help some, but it could be that the problem is small enough that until some extra noise gets injected by those PWM signals, it's low enough not to be noticeable, but the presence of that added noise is actually just revealing the ground loop.

Sounds reasonable. I

[David Hess]

Each component has it's own ground connection to the chassis, but there's a few runs of long wire connected to the same ground. One being the FM antenna, others being the RCAs - but their signal low reference does not internally connect to the chassis ground.
Other than using isolation transfers, I don't think there's much more we can do to improve the situation here.

It is important that if there are multiple grounds, that they all use the same ground.

The antenna connection can be a big problem although I assume they know this and take steps.  You might try disconnecting the antenna temporarily to see if it has any effect on the noise.  Back when I had problems with my mobile VHF amateur transceiver where the ground connection directly to the battery and the antenna ground allowed significant alternator whine into the audio, I built a special coaxial adapter which used an array of low value ceramic capacitors on the *shield* side of the coaxial transmission line to isolate the antenna's ground connection at low frequencies; it worked great.

Is the audio system all one unit?  If there are separate units and they use different ground points, then single ended audio signals between them may be corrupted.

There's the headunit, right next to that a simple equalizer, and then there's an external amp in the trunk. The headunit and equalizer indeed output unbalanced signals.

I have designed audio electronics which had to operate with multiple grounds and used the same trick with the singled ended low references not connected to chassis ground.  I used instrumentation amplifiers or currents instead of voltages but isolation using audio transformers works well also.  I assume they take similar steps.

I've done a test powering up the headunit and equalizer from a separate battery. This greatly reduces the buzz to a point where it's inaudible unless I turn up the gains to near-maximum. Does this mean the filtering in the headunit is poor?

The filtering might be poor or it could be because you removed the ground loop.  You might try powering them with the separate battery while leaving the power ground connected to find out which it is.