A little history of the Directstream DAC.
Paul McGowan, the leading man at PS Audio of Denver/Colorado, was introduced to Ted Smith via a mutual friend, the SACD maven Gus Skinas. Ted is a software developer with an immense track record. He is also a hardcore music lover who has been tinkering with SACD and its underlying DSD format for many years after encountering his first SACD disc.
His then Sony SACD player became the first victim of Ted’s experiments and was altered substantially versus its DAC’s original layout. This became the start of a serious project to design and build the best DSD DAC possible. Cost was no object nor was the size of the thing under development. From the beginning Ted was clear that the use of a commercial DAC chip wouldn’t be an option. What he had in mind was not possible with a single chip nor a massively paralleled array as what many manufacturers revert to. Being a software guy, he decided to do all conversions to proper DSD in software. This meant a field-programmable gate array or FPGA.
Now he coded a true 1-bit converter into his gate array. Initially Ted meant to use the DAC only for DSD and process that specialized signal for the very best analog equivalent on the other end of the DAC. But whilst working on the project he discovered a software solution that would accept any kind of digital input signal. In true software guy fashion, he merely looks at what enters his FPGA. By sampling that signal at very high speed, the resultant pattern determines the signal’s format. This welcomes PCM at all going rates just as it does DSD. In fact it converts all PCM to DSD in the process.
An additional advantage of this process is that the incoming clock signal is discarded. Once the signal type is recognized, there’s upsampling to 30 bits at 10 x the standard DSD rate or 28.224MHz to push noise deeply into the ultrasonic sphere. Before the signal hits the low-pass filter to convert to analog, it gets downsampled to DSD128 at 5.6448MHz. Ted’s gate array has plenty of bit headroom to dedicate 20 bits to just volume control over 100 x 0.5dB steps across 120dB. In total his DSP environment can process 50 bits of data density.
According to Ted Smith, "DSD is whatever bitstream gives you after running through a low-pass filter". A bitstream with a high signal results in a lot of high pulses, a bitstream with a low signal results in a lot of low pulses. Pulses represent energy, hence high pulses mean high energy. Sounds simple, doesn’t it? In PCM encoding the bits give you a number. That number has to be translated to an electric level at very high precision. The electric level is set by little resistors inside the DAC chip. This inherently limits precision and the power may fluctuate a little.
