Analogue design of the pre-amplifier
The preamp design
Although the preamp is kept as straightforward as
possible, it feautres a comprehensive display to show the current state
of the amplifier. A few photos are shown, the green display is built
around a character LCD module. The blue-ish like display is the newest
prototype using a VFD display which can also be used as graphic display
for showing different types of fonts. Additionally this VFD display
illuminates, so also in the dark it can be seen.
All inputs of the preamp have gold plated busses to reduce contact resistance. The input selector for the four inputs devices is created with a high performance analogue audio switch (Analog devices). These switches feature outstanding channel separation (100dB@1kHz) ultra low noise (1nV/(Hz)½) and low distortion (0.003%). The OFF-isolation is 120dB. The selection is done digitally by the microcontroller.
bass output is created by summing both channels using two precision
opamps (Analog Devices).
These opamps are optimized for low voltage noise at the expense of
higher current noise. Generally, both voltage noise and current noise
contribute to the total noise in a system. Voltage noise dominates in
the low frequency range whereas current noise dominates at the high
frequencies (generally, the sum of both is minimum at 1kHz).
For the low frequency bass output, voltage noise is the dominating noise source. It makes sense to take a low noise type for voltage noise in this case. This opamp also has very low offset (10µV) which reduces the need for AC coupling with capacitors in the signal line. The slew rate of these opmaps is not very high (3V/µs) but a simple and straightforward calculation indicates that this is sufficient for low frequency signals. Besides, it makes the circuit less sensitive to capacitive loads and noise glitches.
After this summing an additional opamp is used to create a phase inverter on the bass response (180 degrees). This phase inverter can be beneficial for some acoustic situations. In the calculation of the filter this phase shift is accounted for.
Separate from the bass, two channels are available after
the input selector. These signals are further processed by high
performance buffer and amplifier stages (Burr-Brown).
The balance and volume settings are determined by a programmable gain
This chip uses a serial digital interface to read the code words for
the volume settings. The dynamic range of this chip is 120dB and volume
settings can be changed in 0.5dB steps. Furthermore, crosstalk is
-126dB and THD+N<0.001%. This chip is the central part in the
preamp. It processes all audio changes including the MUTE function.
Before driving the filters of the speakers, the signals are buffered through fast acting opamps. To some extent, voltage amplification is performed through these opamps. Here, the slew rate is 20V/µs, more than sufficient for any audio frequency to be amplified with a headroom of 24dB typical. A bandwidth limiting capacitor is used with this fast output stage to avoid the introduction of high frequency noise from digital sources. These opamps are chosen for their low voltage noise in combination with current noise since they amplify the full audio range. Offset is <2mV, generally small enough to avoid AC coupling techniques. Any existing DC voltage will utlimately be blocked by the active filters in the speakers. Finally, these opamps are able to drive high capacitive loads. This is important because a relatively long interconnect cable will be connected to this output stage (the active filters are build in the speakers). Finally, the output of the pre amp is protected for power up/down (DC) clicks by using a timer controlled switch.
Slew rate and headroom
What slew rate is required for an amplifier to perform
correctly for the whole audio range? A straightforward and simple
calculation provides some insight in slew rate.
A sinusoidal tone of 20kHz, will have its highest voltage after ¼ wavelength. In time, after 12.5µs. Take as reference a 1V input signal, then voltage must raise 13V to reach its maximum available headroom. This results in a slew rate of 13/12.5 ~ 1V/µs. For SACD, 40 kHz signals can be of interest but even then a slew rate of 2V/µs would suffice. For low frequencies such as bass, slew rates can be even very low. An input signal of 1V is generally enough to fully drive any amplifier. Purchasing an amplifier with a slew rate larger than 10V/µs is of no use for the normal audio bandwidth. In this calculation, the maximum available headroom is used. Headroom is here defined on the dBu-scale. This is a standardised scale for which Vin=0.775Vrms. But more important is at what distortion level this headroom is defined. Larger headroom values are often available but it is important to understand what THD+N is for that headroom. In this case it is defined at THD+N<0.01%.
When additional gain is applied some amount of voltage gain is needed. This would indicate that the amplifier should be faster to handle the complete voltage span. This situation occurs for power amplifiers where large voltage gains are needed to get the required power at the speaker terminals. Power amplifiers need therefore higher slew rate to match with the preamp but a slew rate of 10V/µs also suffice for any power amplifier.