Let me see if I can help explain some of this.
First, let’s start with the upsampling and why 512DSD, which is what the PMG runs at (if you want as it is selectable in the menu) is better than 256DSD, which the MKII ran at. Quite simply, noise. The higher the DSD frequency the easier it is in the analog filter to reduce the ultrasonic noise.
One of the tricks with an analog filter is to make sure it does not introduce phase shift in the audible band. To do that you need to start at a high frequency and not make it too steep. But, you also have to be effective. This is one of the reasons why upsamping really helps: a gentle filter is effective if the starting frequency is high enough.
Next, jitter. Ted was brilliant in his algorithms that worked on lowering jitter in the FPGA. The lower the jitter, the better the sound. In PMG, because we are not using an FPGA in the digital section, but rather dedicated chips and sophisticated reclockers, we are able to get the jitter lower than DSMKII by quite a lot through hardware and technique. No need for fancy programming. PMG enjoys jitter at an astonishingly low 2Ps (two Pico seconds) which is extraordinarily low.
The ladder DAC is in the analog board and here we use an FPGA and some fancy programming to build a precision 64 tap sequencer that shoves the noise and distortion out further in frequency so our gentle analog filter can lower noise and distortion to better than -100dB (DSMKII was around -80dB). That extra 20dB of better noise and distortion performance is significant. The reason we gave this system a fancy name, SpectraWave, is because the way engineer Bob Stadtherr pulled this off is unique and potentially patentable. Really cool ideas here that haven’t been done before.
Lastly, as mentioned in an earlier thread, PMG Signature is radically different in yet another way: the volume control. While many (most) will be able to implement the 512 properly, through a PMG Signature Preamplifier, for those wishing the go it alone, using the built in volume control of the PMG 512, we have eliminated the math approach of lowering the signal amplitude through programming, and instead implemented an analog stepped attenuator at the output. This is better because, while -100dB is quite low, as you turn down the volume, the noise floor remains and the signal level goes down, thus making the S/N ratio worse as level is decreased. PMG 512 solves that with the analog volume control because both signal AND noise are reduced at the same time. Simple in theory, not so easy to implement properly.
Hope this helps folks understand better.