mixing balanced and unbalanced connections- problem?

[if6was9]

Alright pardon my noobness but is there an issue with mixing balanced and unbalanced connections?
As I understood it it meant an increased chance of noise and a drop of 6 db in signal strength.

How much of a problem is this? Never really had this issue before but I've been reading up on the FMR RNC and people complain about it alot. Some go so far as to cite it as a reason to NOT buy some units. Is it that much of a problem?

 

[Omega_Void]

IMO you should worry more about the general performance of the gear than whether the IO are balanced or not. You can always get external line drivers and DI boxes if necessary.

If your cable runs are nice and short, and there's not too much electrical noise in the environment, unbalanced outputs are fine. Better to connect them to unbalanced input if possible.

If you use long cable runs, and/or there's a fuck-ton of electrical noise where you are, balanced is necessary.

It also depends on how well the grounding is designed in the gear. If a dipshit designed the circuit-board, noise picked up by the cable shield will get into the audio, balanced connections or not.

Stupidly technical elaborations coming next...

Well I couldn't just leave it at that, could I?

 

[Omega_Void]

As far as the extra 6dB of signal, even if the inverted ("cold") output stage has only as much noise as the non-inverted output, you still get an extra 3dB of noise as well.

Actually, it's often achieved with an inverting unity-gain opamp buffer. In that case the stage gain is -1, but noise gain is 2, so the SNR isn't improved at all.

So often all you get is 6dB extra headroom, which only matters if an unbalanced out can't give enough level. IME what actually happens is you have top turn the balanced out down to prevent overdriving the downstream input.

Also, not all balanced outputs drive both the hot and cold wires. Some only put signal on the hot. As long as the output impedance is the same on both, it's still a balanced connection, and common-mode noise rejection still happens.

So as far as the extra 6dB of signal goes, IMHO it's not worth caring about.

 

[Omega_Void]

And then there's the noise rejection.

Balanced connections only cancel out the noise picked up by the cable. Short cable runs pick up less noise.

Active differential circuits (i.e. electronically-balanced inputs) unavoidably have at least 10dB greater input-circuit noise than a similar single-ended (i.e. unbalanced) input circuit.

So the question in that case is... does the noise picked up by the cable outweigh the extra noise from the input?

Professionals always prefer balanced lines for both studio and live work, because they invariably have to use long cable runs, which pick up lots of noise. Also, if you have cables carrying different signals running along together, balanced lines cancel out crosstalk between them.

With short cable runs in a fairly noise-free environment unbalanced IO may actually give lower noise.

Probably the worst-case however, is plugging an unbalanced output into a balanced input. The cable noise isn't cancelled, and the cold input stage still contributes noise to the circuit even though it isn't doing anything useful.

In that case a 1:1 DI box on the unbalanced out will at least give you balanced noise rejection.

 

[Omega_Void]

Transformer-balanced inputs are a different kettle of fish.

The active circuitry is single-ended, so no noise penalty there,and the transformer contributes very little noise. (Yes, passive parts have noise too!) But they're expensive, and need to be well (i.e. expensively) shielded to avoid the trafo itself picking up noise.

Transformers do have limited bandwidth, and may alter the frequency response audibly. They also add low-frequency distortion. Sometimes it's cool mojo. Sometimes it's just naff. Transformers can be had with very low distortion and wide bandwith. But they are very fucking expensive.

And of course, transformers allow full galvanic isolation of the upstream from the downstream devices. Nice if you like breaking ground-loops.

 

[Omega_Void]

Oh, and going back to active balanced inputs. Some folks make a lot of noise about the fact that the bog-standard circuit has very different input impedances on the hot and cold lines, claiming this screws up the noise rejection completely.

Actually, to a common mode input (which the noise is), it presents the same impedance on both hot and cold inputs. Noise rejection is as good as the precision of the parts.

And while I was googling up that image, I found this page, which has more information on balanced circuitry than even I want to know. Tech-heads; enjoy.