Are the parameters real?
The technical parameters such as SNR and THD+N are based on the orignal manufacturers datasheets.
The schematics and the PCBs were designed to make use of these parameters and to guarantee that the parameters can be achieved really with the setup.
Why these parts were selected?
The intention was to create a DAC which has outstanding and state-of-the-art parameters.
The parts were selected due to best SNR, THD+N and noise parameters.
Also the power rails where optimized for best performance, lowest noise and 50/60Hz hum supression.
What is the difference between RPi-DAC SingleStereo and DualMono?
DualMono RPi-DAC uses two DAC boards, each with a separate DAC and LDOs.
One DAC is used for one channel. The Stereo-Outputs of the DAC are combined for one channel so that we can get a +3dB better SNR.
The DualMono would provide also a better channel separation so that a Stereo signal is much better processed without cross talk.
Why there is a headphone amplifier?
The main purpose was to use the RPi-DAC or T-DAC with a headphone.
Just with a headphone you can really use and enjoy the quality. Any other power amplifier will provide not the same quality or degrade the signal. A loudspeaker will not be able to perform in way to make use of the provided audio quality.
For the DAC output (PCM1794A) an OpAmp would be needed anyway (e.g for the I/V conversion). The headphone amp selected provides outstanding parameters so that even connecting to a power amplifier would benefit.
The headphone amp used is actually a line driver, offered for balanced-to-single-ended outputs. It works perfect with the TPS6120A to feed a power amp. This chip provides also excellent parameters such as Slew Rate, Noise and THD.
Can I connect a power amp with loudspeakers?
Yes, you can, if you attenuate the signal. The signal (voltage) generated for a headphone is way to too high and strong for a regular power amp input (e.g. 2Vrms as max. input for your amp).
My tests have shown (e.g. on a Woo WA3 tube headphone amp): you can use the headphone out directly and feed a power amp with it, it worsk. Your amp should have a pot for the loudness control and input volume level. Benefit: you can drive with high level and reduce the noise level.
You could use a direct cable from 3.5mm Stereo Headphone out to 2x RCA in, if your amplifier has an input potentiometer and you make sure not to turn to full loudness.
Is there a natice RCA out?
RPi-DAC does not have a native RCA output, for a power amplifier, because the signal cannot broken out to RCA directly.
Internally, the RPi-DAC uses symmetrical signal pathes (optimal for quality) and the DAC output is connected directly (via OpAmp) to the headphone amplifier (main purpose). For unsymetrical RCA an additional circuit would be needed anyway.
The used headphone amplifier is a current feedback amplifier: any breakout before it will not work and create trouble for the headphone amp if done directly.
It is quite easy to convert the headphone output into an RCA connection for a power amplifier compared to separate and process the signal for an RCA output without any feedback to other signal pathes. It would need more parts and costs for such an option.
The T-DAC-IF, RPi-DAC-IF module provides RCA output connectors. The line driver for it is still the headphone amplifier.
Can I connect my own I2S source, e.g. an FGPA?
The main intention was to use the T-DAC and RPi-DAC on an FPGA used for audio processing. For this purpose there is an option to have also an SCLK signal which is not provided by the Raspberry Pi but would be needed by the DAC.
The PCB has jumpers so that other formats can be selected even to use an external filter, e.g. in front of the RPi-DAC.
There is also an option to solder a 24.576MHz oscillator which can be used as master clock source, e.g. on an FPGA.
As long as your source provides an I2S signal which is accepted by the RPi-DAC, T-DAC - any other I2S source can be used.
What about the SCLK signal on DAC chip?
Actually, DAC chips have an SCLK signal. This is the system clock needed to be provided so that the DAC can work.
The Raspberry Pi (RPi) does not provide such a signal, it outputs just the other I2S signals: LRCK, DATA and BCK, but not a system clock.
The benefit of using a PCM1794 is: it has internal PLLs and can generate the needed internal clocks using the SCLK signal as reference for the internal PLLs.
The PCM1794A works well if the SCLK signal is connected with the BCK signal which is done on the RPi-DAC board via a jumper (SCLK is BCK).
Could I hear the 24bit?
Actually not: there is a technical fact which is: for every bit you want to hear you need approx. 6dB SNR.
Some DAC products or chips provide a parameter called Effective Nimber of Bits (EOB). This is the calculation backwards from SNR: how many bits can be resolved due to fact that 6dB are needed for 1 Bit on a given (system !!) SNR?.
Therefore, even with the DualMono, assuming 132dB SNR, you would be able to measure (and hear) just a 22bits, not 24 bit.
For real 24 bit you would need a complete audio processing, including your power amp, which would provide 144dB SNR. Any power amp or headphone will not be able to support 144 dB SNR (coming out of loudspeakers - no way!), therefore, not realistic to listen really to 24 bit on common end consumer systems.
But anyway, you would benefit from a 24bit DAC: the better the signal source, the better the results.
Or: you woud use an upsampling, a sample rate conversion, e.g. 16bit/44.1KHz (Audio CD) to 192KHz: the interpolation can create additional (artifical) samples which can make use of the 24bit value range. This can improve the signal.
Just: if thinking about upsampling and using the 24bit range: the DAC performance gets a bit worse on higher sample rates (SNR and THD+N decreases a bit).
The more DACs you would use in Mono Mode the better the SNR. If you would add the signals (of each channel) on separate, parallel DACs, you could increase the SNR and Effective Number of Bits. But using the PCM1794A and to hit 144 dB SNR for 24 bit you would need TEN (10) of those DACs.
And: your ears do not have a sensitivity and dynamic range, an SNR, covering 144dB. Think about: 120dB signal level is already a serious pain level, airport loudness. Even your DAC could cover 120dB, potentially you will use just a range of 60 - 80dB, above it will be so loud that you do not have fun anymore (therefore: Blue Book Audio CDs (16bit/44.1KHz) cover quite perfect the human capatilities).
Why transformer without common (center) tap?
The RPi-DAC uses OpAmp and Headphone Amp with symmetrical + and -15V.
In order to achieve a high performance also due to optimization on power supply - the same positive LDO is used in order to provide the -15V. This LDO has better parameters compared to a negative LDO.
The positive LDO is used to generate the negative -15V. This works only if no connection is there between the positive and negative rail, including the AC input. The positive output of the LDO is grounded and the ground used as negative rail. Any connection between positive and negative rail would shortcut the LDO and will not work.
It is mandatory to use a transformer which has two separate coils, no common tap or center tap between the two windings.