0
\$\begingroup\$

Or in short: was cost the only motivation? Because the major reason against this practice must have been already plain obvious in the mid of last century.

Namely, using the shield of a cable as signal return conductor makes ground loops between devices a problem. For all I know, microphone cables have always been equipped with a dedicated signal return conductor, i.e. 2 conductors plus shield. But for some reason, this hasn't been adopted for all audio.

The second order consequence, that may not have been obvious, is the various fairytales and voodoo industry that surrounds quieting audio equipment today. I believe audio electronics is the one subfield of electronics that has the most misinformation floating around.

So how did we get there? Maybe one could formulate some kind of community wiki around this topic.

Example:

A computer connects to an audio amplifier some meters away using a 3.5mm phone plug cable. Line-level. Both devices are mains powered and both have grounded shells. A large ground loop forms through the room wiring. The ground loop current offsets the reference voltage of both devices by a 50Hz disturbance, which consequently looks like "signal" to the amplifier. Instead, when using headphones with the computer or ungrounded phones/notebooks to provide the signal to the amplifier, there is no added hum because no groundloop is present.

If the connection between computer and amplifier would be made with a shielded differential cable with 1 extra return conductor, the ground loop would be of course still present. But the devices don't care about slight ground voltage mismatches because the signal level is transmitted differentially. No additional hum is caused.

\$\endgroup\$
16
  • 2
    \$\begingroup\$ Shield as signal return is literally how coaxial cables work, I'm pretty sure it has some major advantages in terms of loss and external noise rejection over using a twisted pair, even shielded twisted pair. I do not think any of those advantages are very relevant at audio frequencies, though. \$\endgroup\$
    – Hearth
    Jan 24, 2022 at 6:14
  • \$\begingroup\$ Generally, using shield as a signal return is the best way to get good signal integrity in cables. Oscilloscopes use the shield as the signal return. So do all coaxial cables. I am sure you have a good point, but I don't get what it is. I don't think using shields as signal return is the problem. I think using shields for anything other than signal return is likely a problem. \$\endgroup\$
    – user57037
    Jan 24, 2022 at 6:21
  • \$\begingroup\$ @Hearth I clarified that I mean specifically why 2-conductor cables were adopted. A twisted pair cable would suffer the same ground loop problem, if one cable would connect to a metallic connector case. In contrast, mic cables with 2 conductors + additional shield existed already then. \$\endgroup\$
    – tobalt
    Jan 24, 2022 at 6:22
  • \$\begingroup\$ I guess microphones are particularly problematic because the signal is at a very low level with high source impedance, and unlike RF applications, line frequency (50 Hz or 60 Hz) is in band. The phonograph output also. But line level stuff is much less susceptible to problems. Using the shield for return seems to largely work fine. \$\endgroup\$
    – user57037
    Jan 24, 2022 at 6:37
  • \$\begingroup\$ @mkeith This is certainly true, but instrument connection also have problematically high impedances in a range of maybe 10 kΩ - 1 MΩ, and don't use dedicated return and shield conductors. Even for line level, it still becomes a problem when forming a several meter large ground loops with cables and a busy electronics room. \$\endgroup\$
    – tobalt
    Jan 24, 2022 at 6:54

4 Answers 4

2
\$\begingroup\$

the major reason against this practice must have been already plain obvious in the mid of last century.

Must it?

In 1950 all audio equipment used vacuum tubes. Amplifiers had input impedances in the 100 kΩ to 1 MΩ range. At these high impedances magnetically induced EMI was less of a problem than electric fields, which shielded cable does a good job of blocking.

Record turntables had the piezoelectric cartridge 'grounded' through the shielded signal cable to the amp, with mains earth going only to the mechanism. Low impedance devices such as dynamic microphones needed a step-up transformer to match the higher amplifier input impedance, which provided balanced to unbalanced conversion for free. Where ground loops did occur they were usually the result of unusually long cable runs or poor grounding inside the equipment, not merely 'using the shield as signal return'.

But indirectly heated tubes had heater-cathode leakage, and mains supplies were unregulated with minimal smoothing, so users expected to hear some 'hum' in all but the highest quality equipment. It was how you could tell the amplifier was turned on!

\$\endgroup\$
2
  • \$\begingroup\$ Thanks for these examples of how hum was dealt with. I can't understand the first paragraph though. The ground loop current flows in the shield which has an impedance of Ohms at most. The input impedance doesn't matter. Even if you assume that the signal wire is totally unaffected by EMI, it is the offset in the reference (i.e. ground) voltage between the two devices that makes the signal look different although it is in fact at the same potential in both devices. \$\endgroup\$
    – tobalt
    Jan 24, 2022 at 11:03
  • 1
    \$\begingroup\$ Higher impedance means larger voltage at the same power level. Therefore a higher impedance system tends to be less sensitive to small interfering voltages. eg. 600 ohm microphone vs 20k ohm microphone the signal voltage is ~15dB larger at 20k. \$\endgroup\$ Jan 24, 2022 at 19:07
2
\$\begingroup\$

If the connection between computer and amplifier would be made with a shielded differential cable with 1 extra return conductor ...

That works if the output and following input stages are differential. This adds cost for additional circuitry or transformers, more complex connectors and more expensive cabling. The technique is employed on professional audio equipment but it's generally not required on consumer electronics.

For all I know, microphone cables have always been equipped with a dedicated signal return conductor, i.e. 2 conductors plus shield.

No, high-impedance microphones including those used for stage performance used single core and screen.

Maybe one could formulate some kind of community wiki around this topic.

I don't see what purpose this would serve.

If the connection between computer and amplifier would be made with a shielded differential cable with 1 extra return conductor, the ground loop would be of course still present.

You can insert a transformer or a differential buffer to break the ground loop.

\$\endgroup\$
1
  • \$\begingroup\$ Thanks. I don't see what purpose this would serve. At the risk of displaying my blatant ignorance of the purpose of community wikis: If there were a good universal solution, it will be a good general answer to the frequent questions on proper treatment of audio ground loops. \$\endgroup\$
    – tobalt
    Jan 24, 2022 at 11:53
1
\$\begingroup\$

Simple installations : to connect your gramophone pickup to the "pick-up" input on your AM radio. As such it was commonplace by the mid 1930s.

The screened cable didn't pick up mains hum from the AC motor (shaded pole induction motor) driving the thing. Perfectly good enough for installations at the time, more complex installations in the home were unthinkable. There were no computers - outside Alan Turing's brain - let alone multimedia installations.

\$\endgroup\$
0
\$\begingroup\$

My two cents.

The best way to preserve integrity of a signal is to use differential mode, but this requires electronics at both ends of the cable. In case of a microphone, for example, that would involve a battery in the microphone, or a fourth wire for power supply. Both add cost, and audio eleectronics is a very "consumer" field where cost is sensitive. Going to professional field, actually the standard is a balanced signal, with power supply (+48V) and shield (3 wires + shiled). And it works well: a magnetic microphone has a very weak signal but, with that system, can travel various meters keeping good quality.

About the ground loop, common problem in audio, it makes little difference whether the ground is a normal wire or a shield, the loop arises anyway and, in fact, many professional devices has a switch called "ground lift" which, essentially, disconnects the ground. Ground and signal return should be separate things, at least in theory, and I don't know how the internal schematics of many audio devices is.

That said, if we want to have low cost and simple things, we have to use only two wires. Using two wires, coaxial cables work better, probably because the impedance of the ground or common is, by definition, much lower than that of the signal. So the noise arrives to the cable, but it can not reach the inner wire because it is shielded by a low-impedance shield. This explanation can be plainly wrong, I am not sure, but my personal experience confirms the fact. The last case was a simple serial transmission (digital, frequency about 1 kHz) between two noisy devices with big DC motors about 10 meters apart. There were a lot of transmission errors and, using a scope, I could see a lot of spikes even when the line was driven low (directly to GND); the noise was caused by the motors (up to 15 amps @24V). Using a shielded cable solved the issue.

Also, I play in a band. When preparing a stage, we lay wires for 230V power supply, shielded cables for microphones, and jack (coaxial cables) for other electric instruments. In the vast majority of the cases, problems arise with jack cables because defective contacts in the jack itself. Microphone cables (we call them Cannon cables) are less likely to give problems and they are more robust - especially the connector. Sometimes ground loops or whatever happen, especially with monitor speakers and guitar amplifiers but, even when no big problems are present, there is almost always some noise when nobody plays or sings. The biggest problems happen when the jack cables work badly. If the shield has not a good contact, not only the signal has no return path, but also the core is exposed to every noise running around. There is even the risk to damage the audio mixer, amplifiers or speakers, because the signals can reach big voltages.

\$\endgroup\$
2
  • \$\begingroup\$ My view on things like ground lift switches (or even those where you can select the impedance to ground) is that they create new problems such as EMI especially in newer devices with switch mode power supplies and digital circuits. Which is part of the problem those techs are viewed as a sort of modern evil in some audiophile circles. And about cost, I actually think that one could save many strange gimmicks that is used to "remedy" single-ended audio signalling if balanced transmission was default even for instruments etc. Mics shows how well stuff can work and isn't insanely expensive. \$\endgroup\$
    – tobalt
    Jan 24, 2022 at 7:22
  • \$\begingroup\$ @tobalt I agree that it isn't expensive anymore. But it once was. Hence where we are currently at. It would be nice to establish a new baseline reality. A new floor beneath which we cannot sink. But I'm not sure how quickly it would be adopted. But good points, all the same. (Electret microphone elements would all be 3-wire IC-based today, as the cost is about the same and the quality is very much better for it. But we still find mostly then same 2-wire systems need to be supported, whether JFET or IC. Oh, well.) \$\endgroup\$
    – jonk
    Jan 24, 2022 at 7:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.