I want to thank the Hugo Designer Rob Watts for his comments. While I don't necessarily agree with all his points, I am a believer in the product and own one. I do have some questions/comments that I would love to hear his input. I will say the recent innovative approaches from him and others is really exciting. Personally I look forward to testing the Hugo against some of the other more innovative approaches I can afford to test against (PS Audio DirectStream, Lampizator) and hopefully some I can't afford (MSB).
1) In its most simplistic form: Is the Hugo DAC still a multi Delta Sigma based design (looks like you are using the PCM2706 chip for this) while applying a bunch of custom designed digital filters (WTA, noise shaping, etc.) to clean up the digital signal? I have heard reference to the Hugo to being a delta sigma dac and it not being one.
2) Master Clock: Is the master clock running at 12mhz and is it the one in the PCM2706? There are many high end dac makers who would consider the resulting nanosecond accuracy not to be accurate enough. MSB upgrades clocks are down to sub pico second accuracy. However, given a big part of MSB's secret sauce is reclocking the data stream for their R2R implementation, I guess it would be more sensitive to clock jitter. Assuming I am accurate in understand the specs of the master clock, I am curious why an even higher end clock wasn't used. Is it one of or some combination of the following reasons:
a) Basic design of Hugo DAC implementation is inherently more resistant to clock based jitter given your use of digital PLL as an example?
b) Belief that sub nanosecond accuracy doesn't make a difference in SQ
3) USB Implementation: Any details on USB implementation and how resistant it is to USB related issues? Personally, I find the coaxial digital input to sound cleaner than the USB. If we look at something like the Berkeley Alpha USB, we see extensive efforts to separate possible USB related issues from the DAC. A lot of those "solutions" would be difficult to take advantage in the box the size of the Hugo. I'm curious what measures were taken however.
4) With your basic design implementation, you have stated that an infinite number of FIR taps to mimic a true audio wave which makes perfect sense. With your previous designs, you have also stated that the WTA filter reduces the need to have a huge amount of FIR taps and that you reach essentially significant diminishing returns in the 1-2k. Given that the new FPGA allowed you to almost exponentially increase the number of FIR taps to 26kish using limited power, where are your thoughts on the number of FIR taps before significant diminishing returns lie? If tomorrow a new low power FPGA came out that allowed you to increase the number of FIR taps to 256k and nothing else changed, do you think it would be audibly noticeable?
Thanks in advance!