Details: The Story of Gozinda & Gozouta
Balanced inputs are a thing of beauty. Magic, really. They take two signal lines and amplify only the difference between them while rejecting anything that is common. Neat. The common stuff is all the garbage induced in the lines as they dutifully connect two pieces of equipment. Hum, noise, your un-favorite radio station, local police and CB radio signals, radar, fluorescent lights, life itself - all competing to violate your pure signal as it tries heroically to traverse from one piece of gear to another.
Many solutions exist to accommodate balanced inputs. Transformers are the most common, and the most expensive, and the most troublesome. So much for transformers.
Heading the list of transformerless balanced inputs is the single op amp difference amplifier of figure 1. From here the solutions expand (as does the cost) to two and three op amp instrumentation amplifiers. The main advantage of the more complex circuits is their equal input impedance characteristics. A characteristic not exhibited by the single op amp difference circuit, with one exception. Patience, I'm getting to it!
Back to figure 1. If R1 = R3 and R2 = R4, then the circuit amplifies the difference between the inputs by the factor R1/R2, i.e.,
Vout = R1/R2 (Vin+ - Vin-)
Any signal common to both inputs will not be amplified at all, providing all resistors are perfectly matched and you are using a decent op amp. More on this later.
For unity gain applications, all four resistors will be equal, and the circuit performs the important task of converting line level balanced inputs to single ended outputs, and is probably the most commonly used circuit for signal processors. So what's the problem? Unequal input impedances, that's what.
Looking into the negative port, the input impedance seen by Vin (-) is R2, while Vin ( + ) "sees" R3 + R4 as its input impedance. Typical numbers would be for all resistors to equal 20k Ohms, therefore the negative input impedance is 20k Ohms to ground, while the positive input impedance is 40k Ohms to ground. Why is this a problem? It isn't, for some. For others, it creates a system full of humming birds.
The difference is the output impedance of the driving circuits. If it is zero (wistful thinking) then an unbalanced input impedance is not a problem, since the impedance of the connected lines is zero and nothing will be induced. As the output impedance is increased, then so is the impedance of the connected lines. If it is equal, then what is induced will also be equal and the common mode rejection of the circuit will not allow the garbage to pass; however, if the output impedances are fairly large (several hundred Ohms is plenty) then the mismatch in input impedances will now cause the connected lines to have unequal impedances and the induced garbage will not be common, but different. This is then converted as signal and everything goes to Hell. Hence, the popularity of instrumentation amplifiers with their equal input impedances. But, all is not lost - remember the one exception?
It turns out, in general, that if R1 = R2 (not R3 as previously stated) and R3 = R4 then you also get a unity gain difference amplifier, and, specifically, if R2 = R3 + R4 then you get a truly balanced input amplifier with unity gain (it doesn't work for gains greater than unity - sorry). Letting R1 = R2 = 20k Ohms and R3 = R4 = 10k Ohms satisfies our requirements and yields equal input impedances of 20k Ohms to ground for each leg. The answer to the not-so-balanced input stage is so obvious that most designers have missed it (or ignored the problem) for years. Now, you have the heart of a general purpose balanced input stage that, with a few embellishments, will serve for any projects you may have requiring balanced operation.
Figure 2 shows a complete balanced stage with a stereo 1/4" phone jack input. There is a bit of cleverness going on here that comes for free. Using a stereo phone plug into the stereo jack gives you the balanced line input - obvious. Not so obvious is that by using a mono phone plug, the circuit automatically switches to unbalanced single line operation. Nice, huh? What happens is that the ring of the jack gets shorted to ground by the mono plug and turns the difference amp into a non-inverting stage with a gain of two, but there is a 6 dB pad hung on its positive input, so the net result is a gain of one, unbalanced input. The stage becomes even more universal now, being either balanced on unbalanced, depending only on whether the input plug is stereo or mono.
Trimpot R3b is added as a "tolerance soak", since this one adjustment will soak up all of the resistor mismatches. A common mode signal is applied to the inputs (e.g., tie both inputs together and drive with a signal generator set for 2 Volts output and 1 KHz) and R3b is then adjusted for minimum output. One percent resistors are required if any kind of decent common mode rejection is to be gained, since any mismatch will cause a common mode input signal at the resistors to appear as a difference signal at the actual op amp input terminals.
So much for down and dirty balanced inputs; now how about an equally dirty "balanced" output that will blow the socks off more elaborate versions?
Balanced outputs have evolved in a similar fashion to balanced inputs; starting out with everything using transformers and gradually moving toward transformerless solutions. The most commonly seen solid-state circuit involves two op amps as shown in figure 3 (or some variation of this configuration). There are other, more complicated schemes involving reverse feedback circuitry to guarantee that the differential output signal is virtually immune to loading effects (using at least four op amps), but each of these circuits share the same problem - they all reference their outputs to the amplifier ground - a definite no-no for hum-free systems.
The best solution, as is so often true, is the simplest. It is no more complex than figure 4. By taking a regular single-ended (unbalanced) line driver and floating its output and ground, you create a compatible system for driving differential (balanced) inputs that is trouble-free. Figure 5 shows the interface between the two systems. Note that while the chassis of each unit may be at the same potential, their signal grounds are allowed to be at different potentials. This is very important in keeping hum common mode, and not differential. Any difference of potential existing between the two grounds is seen as a common mode signal and is rejected.
Like the balanced input stage of figure 2, the "balanced" output stage of figure 4 automatically switches to a conventional unbalanced output stage if a mono phone jack is used for interconnection. The ring of the jack gets shorted to the sleeve, thus grounding the sleeve as you would in a normal balanced system.
So, there you have it: A universal input/out-put scheme that is balanced or unbalanced, depending upon the required application... just like apple pie and cheese.
Feature by Dennis Bohn
Previous article in this issue:
Next article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!