Passive volume control with switchable inputs – DIY project
Passive volume control (sometimes incorrectly called “Passive preamplifier”) may be a good solution in case we are aware of all pros and cons of such design.
Pros:
Choice of volume potentiometer
As already mentioned, potentiometer impedance affects input impedance, output impedance and noise. In case that the potentiometer is driven from ideal voltage source (has zero output impedance), the worst case regarding noise and output impedance is when the pot taper is at 50% of pot total resistance R. Output impedance is then R/4 (two R/2 resistors in parallel) and R/4 is also an equivalent noise resistance.
Resistor thermal noise
Every resistor has its thermal noise voltage that depends only on temperature, resistance value and bandwidth of measurement. For noise voltage Vn we have:
Vn = sqrt(4*k*T*R*ΔF), where
k … Boltzmann constant, 1.380649×10−23
T … temperature in Kelvin
R … resistance
ΔF … frequency bandwidth
Next plot shows resistor thermal noise as a function of resistance at 27°C and 20kHz bandwidth.
My goal is to keep worst case noise (pot at 50%) below 1uV. This, together with the demand on reasonable input impedance leads to pot resistance value of 10 kohm. This makes 2.5 kohm worst case noise resistance (if driven from voltage source) and this would make 0.91 uV noise voltage. So I shall use a 10 kohm potentiometer with log taper, Alps RK27112. These parts are very reliable and have low gang error.
Now I want to have a choice from 3 single ended sound sources and possibly one balanced source. So I shall add an input switch and a balanced-to-SE converter. Balanced input path will be active, as I do not want to use a 4-gang potentiometer.
One of the key design goals to me is to keep low inter-channel crosstalk, meaning low crosstalk between selected and unused input sources. It is extremely annoying if I hear a sound from the channel which was not chosen by the input selector. This leads to a design with unused inputs shorted to the local ground via a resistor rather than left unconnected in the air. The later option results in audible capacitive crosstalk. The former option is much better with respect to inter-channel crosstalk, but keeps unused sound sources loaded, in my case with 680 ohm resistors.
Circuit design
This is the circuit schematics of my passive volume control:
We can see 3 SE inputs and one balanced input. The inputs are selected by means of AXICOM D2n 12Vdc relays. It is much better option than to use a mechanical switch and a bundle of quite long signal cables. Mechanical rotary switch is used here to turn on/off dc voltage of relay coils.
The balanced input uses instrumentation amplifier INA217. The input selector board PCB looks like this:
Some measurements
In case that the SE input is driven from sound source with 50 ohm output impedance, the worst case output impedance of this passive volume control was measured at 2.8 kohm. This makes 0.96 uV worst case output noise voltage. The best sound source with respect to output noise I have is the Topping D10s DAC, and it has 2uV output noise voltage (20kHz). So the volume control has lower noise as requested. Dynamic range of the volume control box would be 126dB ref 2V.
I have been measuring the volume control box with Topping D10s DAC and (modified) Creative USB X-Fi HD ADC. This combo is limited by the Cirrus CS5361 ADC, which has dynamic range of 114dB A-weighted and 111dB unweighted, at best. And such are the values that I have already measured. The combo is unable to measure the non-existent distortion of the SE inputs with relays + Alps pot.
Interchannel crosstalk and stereo crosstalk
Crosstalks would be the most interesting parameters to me. We can see them in the next plot:
CH1 – CH3 are single-ended inputs, CH4 is the balanced input. CH2 ==> CH1 plot (green) shows the crosstalk from CH2 channel to CH1 (selected) channel. CH1 input is left open. At 1kHz, the interchannel crosstalk is about -120dB. It rises to -96dB at 20kHz. CH2 ==> CH4 plot (blue), from SE to balanced input, gives even a bit better results. CH2L ==> CH2R plot (orange) is a standard stereo crosstalk from left to right channel of the same input and it makes -92.5dB at 1kHz and -67dB at 20kHz. It is affected mostly by capacitive coupling in volume potentiometer and input relays.
The next 2 plots show what I get either from digital volume control -60dBFS or analog volume control -60dBFS. The difference is negligible but the results and noise limits are affected by the already mentioned parameters of the CS5361 ADC circuit.
Thank you for reading and please feel free to ask technical questions.
Passive volume control (sometimes incorrectly called “Passive preamplifier”) may be a good solution in case we are aware of all pros and cons of such design.
Pros:
- non-linear distortion is eliminated,
- low noise in case of proper choice of volume potentiometer impedance,
- reliability in case of proper choice of components used,
- stability and no occurrence of self-oscillations.
- higher output impedance, depending on choice of volume potentiometer and taper position,
- noise depending on taper position
- input impedance may depend on load impedance,
- output impedance may depend on source impedance,
- load capacitance (length of output cable) to be reasonably limited.
Choice of volume potentiometer
As already mentioned, potentiometer impedance affects input impedance, output impedance and noise. In case that the potentiometer is driven from ideal voltage source (has zero output impedance), the worst case regarding noise and output impedance is when the pot taper is at 50% of pot total resistance R. Output impedance is then R/4 (two R/2 resistors in parallel) and R/4 is also an equivalent noise resistance.
Resistor thermal noise
Every resistor has its thermal noise voltage that depends only on temperature, resistance value and bandwidth of measurement. For noise voltage Vn we have:
Vn = sqrt(4*k*T*R*ΔF), where
k … Boltzmann constant, 1.380649×10−23
T … temperature in Kelvin
R … resistance
ΔF … frequency bandwidth
Next plot shows resistor thermal noise as a function of resistance at 27°C and 20kHz bandwidth.
My goal is to keep worst case noise (pot at 50%) below 1uV. This, together with the demand on reasonable input impedance leads to pot resistance value of 10 kohm. This makes 2.5 kohm worst case noise resistance (if driven from voltage source) and this would make 0.91 uV noise voltage. So I shall use a 10 kohm potentiometer with log taper, Alps RK27112. These parts are very reliable and have low gang error.
Now I want to have a choice from 3 single ended sound sources and possibly one balanced source. So I shall add an input switch and a balanced-to-SE converter. Balanced input path will be active, as I do not want to use a 4-gang potentiometer.
One of the key design goals to me is to keep low inter-channel crosstalk, meaning low crosstalk between selected and unused input sources. It is extremely annoying if I hear a sound from the channel which was not chosen by the input selector. This leads to a design with unused inputs shorted to the local ground via a resistor rather than left unconnected in the air. The later option results in audible capacitive crosstalk. The former option is much better with respect to inter-channel crosstalk, but keeps unused sound sources loaded, in my case with 680 ohm resistors.
Circuit design
This is the circuit schematics of my passive volume control:
We can see 3 SE inputs and one balanced input. The inputs are selected by means of AXICOM D2n 12Vdc relays. It is much better option than to use a mechanical switch and a bundle of quite long signal cables. Mechanical rotary switch is used here to turn on/off dc voltage of relay coils.
The balanced input uses instrumentation amplifier INA217. The input selector board PCB looks like this:
Some measurements
In case that the SE input is driven from sound source with 50 ohm output impedance, the worst case output impedance of this passive volume control was measured at 2.8 kohm. This makes 0.96 uV worst case output noise voltage. The best sound source with respect to output noise I have is the Topping D10s DAC, and it has 2uV output noise voltage (20kHz). So the volume control has lower noise as requested. Dynamic range of the volume control box would be 126dB ref 2V.
I have been measuring the volume control box with Topping D10s DAC and (modified) Creative USB X-Fi HD ADC. This combo is limited by the Cirrus CS5361 ADC, which has dynamic range of 114dB A-weighted and 111dB unweighted, at best. And such are the values that I have already measured. The combo is unable to measure the non-existent distortion of the SE inputs with relays + Alps pot.
Interchannel crosstalk and stereo crosstalk
Crosstalks would be the most interesting parameters to me. We can see them in the next plot:
CH1 – CH3 are single-ended inputs, CH4 is the balanced input. CH2 ==> CH1 plot (green) shows the crosstalk from CH2 channel to CH1 (selected) channel. CH1 input is left open. At 1kHz, the interchannel crosstalk is about -120dB. It rises to -96dB at 20kHz. CH2 ==> CH4 plot (blue), from SE to balanced input, gives even a bit better results. CH2L ==> CH2R plot (orange) is a standard stereo crosstalk from left to right channel of the same input and it makes -92.5dB at 1kHz and -67dB at 20kHz. It is affected mostly by capacitive coupling in volume potentiometer and input relays.
The next 2 plots show what I get either from digital volume control -60dBFS or analog volume control -60dBFS. The difference is negligible but the results and noise limits are affected by the already mentioned parameters of the CS5361 ADC circuit.
Thank you for reading and please feel free to ask technical questions.
