The Subwoofer DIY Page v1.1 - Projects
Alternator Whine Exercise
last updated: 30 March 2011
By: Jason Cuadra

Here's an interesting case study on alternator whine. I don't have much experience in car audio, but my perspective is that of a practicing electronics engineer with an interest in it.

A friend of mine spent some money and an entire day having a car sound tech troubleshoot annoying whine in her car stereo, to no avail. Being a switching power supply designer who deals with electrical noise everyday, I confidently bragged I could solve her noise problem in 2 hours, including breaks for pizza and beer. Well, it took me 3 - so much for my braggadocio.

Symptom: whine volume doesn't change with music volume, but increases with load on alternator. Probably very common, so it would be educational for me.

Setup: amps and active xo in the trunk.

Procedure: I bypassed the xo, and I tried using a 12V motorcycle battery to power the head unit and amp in the trunk, to no effect. This is to check for conducted noise ("ground loop" is technically a misnomer) in the ground wiring of the components. Note, there are 3 ways of noise getting in a system - conducted, capacitive, and inductive.

Most curiously, while fiddling with the shielded cable connecting the head unit and the amp, noise appears only if both left and right channels are connected. If only one channel is connected, neither channel has noise.

If I plug in one channel (so of course only that channel is playing music), then connect just the ground of the other channel's jack, noise now appears in the channel playing music.

If again I plug in one channel, then connect just the "hot" lead of the other channel's jack (so now both channels play music), the 2nd channel has noise big time and the other not.

Then it dawned on me that the noise was coupling into the cables magnetically, which I first dismissed because magnetic coupling drops at 6dB/oct as you go down in frequency, and alternator charging pulses are in the kHz range, not the 100 kHz range where you'd typically start seeing magnetic noise coming in. Then again, in a car, noise at 4kHz, 50 dB down from full power, would be very audible when the music is playing softly.

An alternators output current is not smooth DC but a rectified 3 phase current waveform - a DC current with a ripple riding on it. This current flows in a loop, and if this loop forms a large area, it will radiate magnetic (flux) noise.

The concept is of loop area. Any circuit's current flows in a loop. Any current flowing in a loop radiates a magnetic field proportional to the physical area of the loop. A 2nd circuit nearby (the receiver) will pick up that noise, in proportion to its loop area and the derivative of current (time rate of change) of the 1st current, and inversely proportional to distance. The equivalent circuit is a noise voltage in series, in the 2nd loop. The loops are actually one-turn transformer windings.

When I checked the amp's input cct with my ohmmeter, I found out the following: The RCA jacks' ground was isolated from the amp's power ground. The 2 channels' RCA jacks, however, were connected to each other. I suspect this is typical of most amps.

Now the signal current from the on channel of the head unit flows down the "hot" lead of the shielded wire, into the amp's input stage, and back down the shield of the cable (this is the "return current"). Remember current flows in loops. What makes shielded cable work against magnetic noise is that the shield and the center conductor are concentric and so it has effectively no loop area in any particular plane. This is why a twisted wire pair works against magnetic noise too. So, with only one channel connected, there was no noise. Now, the shielded cable is a pair of cables, running side by side, because the cable is actually a stereo cable, which looks like zip cord, except that each side is a shielded cable.

When I connected only one channel, but connected the ground of the other channel the return current split betweent the 2 shields, and thus created a significant loop area. Voila, noise. When I connected just the "hot" of the other channel, its return current flowed down the other shield so it developed noise, but the fully connected channel did not.

If both channels are fully connected, each channel's return current split between the 2 channels, creating loop area, and so picked up magnetic noise. I thought then, to prevent the return current from flowing in the other shield, the input jacks' grounds should have been separated by the amp designer. But of course the solution was to replace the amp.

I realized then that if I used a single shielded cable, with 2 "hot" conductors, to carry both channels, then the loop area of both channels would be small. Voila, no more noise, even with the active xover reconnected.

Couple of points. The system design is bad. The car is an electrically noisy environment. Her head unit's output was about 0.2V. Her amp's inputs were neither "fully differential" nor "balanced". The former is evidenced by the non-isolated "grounds" on the input RCA pair of jacks. The above circuitry, which any decent engineer would put in, would cost less than a dollar in mass production. It seems they are only available in the more expensive amps. The penny pinchers and creative marketers are at it again. Incidentally only the devices (whether xover or amp) to which the long cables are connected need "full differential" inputs.

BTW The offending cable was sold to her by a salesman at a comparatavely hi price, whereas the cable I had at home which I gave her, I'd bought for 30 cents a foot, and the jacks were 50 cents each.

I wonder how many people replaced an amp when changing the cable would work.

Hope this is educational,

-- A dude named Lincoln replied:

For $20 you could have made a "sniffer" outta of a cassette recorder and headphones and found the entry point of the noise in the system and then you would have decided to replace (or re-route) the cables within 15 or 20 minutes instead of wasting 3 hours!!!

To make your own sniffer: Any cassette player with a record function (play/record head) and any headphones. Extend the tape head wires to be a couple of feet long. Start car, press record, put headphones on, adjust volume so you can hear, hold tape head near all connections, equipment, and wire routes...find the loudest places and decide what to do based on your findings.

That's an excellent tip, using a sniffer. However, I think it should be in "play" mode and not "record". For my own education I'll probably go back to my friend's car and use one and find out where the car is radiating the noise. My guess is it's either in the engine bay or the wiring to the taillights which may happen to be running where the audio cables are running.

--I replied again:

One point I was making was that the cable I described (one shield, two hot wires inside, one for each channel) is superior to two shielded cables (one per channel) when the power amp's (or xover) (in the trunk) RCA jacks' ground terminals are connected to each other, which I suspect is typical of many power amps and processors. The physics of it can't be denied.