Requirements to the volume control circuit:
There exist a couple of different soloutions, which differ in quality and required effort.
In its simplest form the volume control is done with a potentiometer -short a Pot- as dual, or quad-channel type, with a logarithmic characteristic curvature (fig. A and B). Huge differences in quality as well as price exist.
Typically this type shows large tolerances of the nominal end resistance values as well as rather poor matching between the channels.
The temperature sensitivity of the parameters is quite high (e.g 200ppm).
In its usual wiring connection the input resistance is constant and is defined in the datasheets. The output resistance on the other hand varies with the volume setting between close to 0Ohms and 1/4 of the specified nominal value (input resistance).
To keep the influence on signal quality small and as a rule-of-thumb the input impedance of the following device should be 5-10 times higher.
If now additional capacitances occur in this high-ohmic surroundings, as it is the case with cabling and HF-blocking capacitors in amplifier inputs, the situation may become critical. The output resistance of the Pot and the sum of the capacitances form a lowpass-filter, whiches bandwidth limit varys with the position of the Pot-setting. Running long cables the bandwidth limit may even fall below 20kHz into the audible range.
The Pot-circuit is therefore only useable for short runs of cable or wire.
The usual solution to the problem is the use of a dedicated buffer stage that features a high and constant input impedance and a low and constant output impedance.
In the case of using switched discrete resistors instead, one can realize different circuit topologies which are superior to the simple Pot.
In the simplest form, the chain (fig. C), the resistive-track of the Pot is replaced by a chain of series connected dicrete resistors. The contacts of a switch replaces the wiper contact of the Pot. The contact may come in the form of a mechanical switch, a relais or a solidstate switching device.
The effort, parts number count and cost can quickly increase up to impracticality. So in most cases the number of volume steps is limited to 12-24. In IC-form of the so called digital Pots devices allowing up to 256 steps may be found.
Apart from the useage of resistors with improved characteristics and tighter tolerances for lower deviations with regard to overall resistance and channel matching, the chain doesn´t offer decisive advantages against a turning or sliding Pot.
Another variant of the Pot-circuit is the ladder (fig. D). Only single pairs of resistors are switched. The signal passes many less solder joints and each resistor pair may be designed and tolerated individually in value.
Because of the high effort needed, one will seldomly find ladders with more than 24 volume steps.
To overcome the inherent problems of the Pot-circuit, the varying output resistance, a different circuitry of resistor arrays are required.
The circuit shown could be implemented with equal switching stages, but the effort would quickly become very high, or the number of volume steps needs to be rather small. If the single stages are weighted Bit-wise, e.g. -1dB, -2dB, -4dB and -8dB, the number of required stages remains low while the number of possible steps still is sufficiently high.
The 4 stage circuit already allows for 15 -1dB-steps.
A 7-stage circuit would allow for a huge dynamic range of -127dB.
Designed properly the output resistance stays constant and may be kept below 1kOhm. Even long runs of cabling of 30yards don´t play a role any more and one may omit with a dedicated active buffer stage alltogether.
The input resistance of the circuit varies though, but it never falls below a couple of kOhms. This is sufficiently high as long as the driving sources output impedance remains low enough (1/5 - 1/10 of Rout).
Weighted Bit-wise the circuit asks for tightly tolerated resistors, the more so the more stages are involved, because the tolerances of each stage add up. The most critical steps are those where a complete Bit is changed, e.g from step 7 to step 8, step 15 to step 16, or step 31 to step 32, and so on.
This circuit allows for highest playback quality, because it allows to realize a pure passive signal path without any active devices or parts in the signal path.
Besides the passive, resistive variants one can find active volume control solutions. A thought that immideately springs into mind when active amplification stages are already in use. One could directly vary the amplification factor, instead of the variant of variable damping followed by a fixed amplification factor. It remains quite difficult though to achieve a sufficiently large range of volume on one side and not to spoil decisive quality parameters of the amplifying stage on the other side.
In IC-form combinations of this technique and passive resistor chains are on market (Crystal CS3310, TI PGA2310).
The possible dynamic range is huge and the precision of the volume steps is remarkably high. These ICs allow for simple, compact and cost-efficient circuit implementations.
The devices are controlled via Microcontrollers for great comfort and easy implementation of multi-channel useage.
It combines though three inferior techniques at the same, the chain-structure, the material the resistors are manufactured from and the variable gain stages. Nonetheless these ICs are favoured by many and they are quite hyped in DIY audio forums.
As very precise form of Bit-wise weighted arrays so called multiplying DACs come to mind (R2R-DAC). These DACs feature current outputs.
The current level beeing defined by the steering Bits which may be generated by a Microcontroller. They require a current-to-voltage converter at their outputs. The analog input signal is fed into the reference-voltage pin of the DAC. Typically this voltage is limited to +5V, which translates to a signal voltage of just 1,75Vrms. The input signal voltage must not surpass this value or severe clipping distortions result.
The clock input of the DAC only gets pulses when the volume is changed, the rest of the time it remains unused.
The low input voltage limit is basically the only restriction of this technique,
makes it really a bit astounding that this solution is implemented so seldomly.