A balanced interface using symmetrical and inverted output signals.
Could you please explain why audio gear uses balanced signals and how they work? Are impedance‑balanced signals the same thing and if not, what advantages or disadvantages do they offer?
SOS Forum post
SOS Technical Editor Hugh Robjohns replies: A balanced audio system (it is the whole output‑to‑input ‘system’, not just the cable) is primarily used to minimise unwanted external interference making its way into an audio circuit, but it also has the advantage of being ground‑free, by which I mean the audio signals passing across it are not referenced to the system ground, and that helps to avoid ground‑loop noise problems.
A balanced interface comprises two signal wires and, usually, an overall (grounded) screen, although the last is not technically essential. The balanced input receiver, whether in the form of a transformer or active electronics, works ‘differentially’. This simply means that the signal voltage on one input ‘leg’ is subtracted from that on the other. If the signal voltages are exactly the same on both legs (what we call a ‘common mode signal’) they cancel each other out, so that common‑mode signal is rejected. In contrast, the voltage difference between the two legs will pass through. So, for the interface to work as intended, the wanted audio has to be sent as a differential signal, and unwanted interference has to be made common‑mode.
A lot of balanced connection explanations wrongly assert that it must use symmetrical and inverted audio signals on the two wires.
To act as a common‑mode signal, external interference must generate identical voltages on both connection wires, and this requires the two lines to have exactly the same impedances to ground. That’s what makes the interface ‘balanced’: it’s the impedance balance that matters. A lot of balanced interface explanations wrongly assert that there must be symmetrical and inverted audio signals on the two wires, but actually the wanted signal just has to be presented differentially (different signals on each wire), and there are a couple of equally valid ways of achieving that.
Arguably the most common approach is to split the input signal voltage (V) to send half on one wire and half on the other, but with the latter polarity inverted: the classic symmetrical‑and‑inverted format. Mathematically, that’s V/2 on one wire and ‑V/2 on the other. When subtracted at the differential receiver (balanced input), the output is: V/2‑(‑V/2). Simplifying that equation, you get (V+V)/2 = 2V/2 = V. In other words, you get out exactly what you put in! Any interference, though, will appear as +U on both wires and because +U‑U = 0 the interference doesn’t appear at the output at all. Hurrah!
The other approach, known as the ‘impedance‑balanced’ format, sends the whole input signal on one wire and nothing on the other (it is effectively just grounded). For this technique, we have V on one wire and zero on the other, so the receiver gets (V‑0) = V. Again, you get out exactly what you put in, and interference is still rejected (since U‑U=0).
An impedance‑balanced interface puts all the wanted signal on one wire, and nothing on the other.
So, if both approaches deliver the same result why have the two formats? Impedance‑balanced outputs are less costly to implement and have the useful advantage of being fully compatible with both balanced and unbalanced destinations, which is why they’re popular in semi‑pro equipment. However, in professional applications the symmetrical‑and‑inverted signal format offers other advantages.
- First, lots of audio equipment uses transformers, and transformer‑coupled outputs inherently generate that format, making it convenient.
- Second, it offers a 6dB headroom and 3dB system noise advantage.
- The third advantage is that the output circuitry (or transformer) feeding each signal wire is identical, so there are inherently identical impedances to ground and this ensures excellent rejection of interference.
In contrast, an impedance‑balanced output has only one line driven actively, while the other is simply tied to ground, typically through a resistor whose value is selected to match the notional output impedance of the active driver. However, the output driver’s impedance might vary with frequency, especially at radio frequencies, and this can result in the source impedances of the two lines being imperfectly balanced so interference may not be rejected quite so effectively. Of course, most impedance‑balanced outputs are carefully optimised to minimise this potential problem, but it is a fact that there’s a risk of being slightly less effective at rejecting interference than symmetrical‑and‑inverted outputs.