Tobes,
You describe the most common way of deriving a balanced signal from an unbalanced. We do this differently. In the topology you describe, the diffamp becomes the limiting factor from a noise standpoint. Our technique has a 9 dB to 19 dB SNR advantage over this common technique.
In the HPA4 we do the following:
- Send the + input to two independent voltage buffers.
- Send the output of one buffer to a low-impedance inverter that uses the isolated RCA ground as its signal reference (providing common mode rejection).
- Send the output of the second buffer to the inputs of two differential amplifiers that are wired in opposite polarity.
- Send the output of the inverter to the remaining inputs on the two differential amplifiers.
- One differential amplifier drives the +, the other drives the -, this balanced signal feeds the headphone amp or line output buffers.
- Placing the inverter in front of the diffamp gives us about a 6 dB SNR improvement.
- Using 2 diffamps gives us an additional 3 dB SNR improvement.
- The outputs of the diffamps are voltage balanced and common-mode interference has been removed from both balanced and unbalanced inputs.
- We use a relay to remove the inverter when feeding from a balanced XLR input.
- The dual diffamps are integrated into our active relay-controlled stepped gain/attenuation volume control stage to optimize the noise performance of the diffamps.
- The fully-balanced stepped gain/attenuation controls the gain of two buffer amplifiers that are in front of the diff amps in order to optimize the balanced signal level into the dual diffamps.
- This optimization is important because balanced inputs are often 16 dB higher than unbalanced inputs.
- This optimization can provide a 10 dB SNR improvement for unbalanced inputs (above and beyond the 9 dB described above).
- The result is that the system SNR is largely independent of the input signal level.
- The SNR for a 2 Vrms (8.2 dBu) unbalanced input is almost as good as it is with a +24 dBu (12.28 Vrms) studio-level balanced input.
- The line and headphone outputs have entirely separate relay-controlled gain/attenuation stages.
- The line and headphone outputs are derived from two separate sets of dual diffamps.
In total there are 8 diffamps in the HPA4 allocated as follows:
- 2 for L input driving balanced line bus
- 2 for R input driving balanced line bus
- 2 for L input driving balanced headphone bus
- 2 for R input driving balanced headphone bus
I am using "bus" to refer to the internal balanced signal distribution paths that follow the fully balanced volume controls.
The line bus drives the balanced and unbalanced output buffers.
The headphone bus drives the balanced inputs on the THX-888 headphone amplifier.
While this may sound complicated on paper, the parallel structure shortens the individual signal paths and allows the use of lower impedances in critical parts of the signal path.