Discussion in 'Digital Sources' started by Bill Ferris, Jun 5, 2018.
No, I am NOT. The data (the purely digital bit stream) is treated identically, regardless of the media being transported.
The bits are not the problem; provided your link isn't so bad that unrecoverable errors occur, the received data stream is identical to the transmitted data stream.
So, we recover the perfect digital bitstream from the bearer medium at the receiver, and use a DAC to convert to the desired analogue music signal.
The problem comes due to the fact that those bits carried over the bearer medium have an analogue representation (of the bit, or data symbol); a real, physical signal. That signal can carry extraneous analogue noise, as well as the noise inherent in the fast edges of our digital data stream (Fourier series).
Unfortunately, if we don't take care, the bearer noise signals can couple into the analogue side of our DAC, both those extraneous noise signals, and the digital edge noise. This is what we must try to eliminate when designing an interface and the DAC, and there are a number of techniques to do this.
These techniques involve the transmitter (the interface data driver, its power supply, the line impedance control/filtering), the cable (preferably a good, impedance controlled, shielded cable), the receiver (again, power supply, input impedance matching, receiver buffer, filtering), and in subsequent processing stages, using isolation of the receiver from the digital processing, then isolation of the digital processing from the DAC digital input, and an isolation barrier across the DAC, which provides the digital/analogue barrier. Get these isolation barriers right, and you should kill any input coupling to such an extent that it becomes inaudible.
Then there's time domain noise to consider, especially if the DAC is data source clocked... That's why a destination-clocked system is best (e.g the 'asynchronous' USB transfer mode); the clock can be propagate back from the DAC, to all the isolation reclocking stages. Using a FIFO buffer, we can isolate the DAC output sample timing from the source data supply.
Source-clocked transfer mechanisms require the DAC to be driven by a clock recovered from the data stream. Recovered clocks are harder to produce with low jitter than those generated by an independent clock. Jitter (time domain noise) causes audible effects in the DAC output. Like 'grittiness'.
When you're designing to a budget, you take shortcuts. These might be considered 'design failings'. Engineering is always a compromise...
So it's possible that a USB regenerator acts like one of the isolation stages I mentioned, reducing the amount of noise carried from source to DAC. In a DAC with poor input isolation, it will improve the audio performance.
At the price of the regenerator I'd personally rather buy a DAC that didn't need it's help. $200 is steep to correct something that shouldn't need correcting.
What DAC do you use?
Currently an Audient pro-audio interface as I need the mic pres/phantom power. Feeding my Millett hybrid and Fostex cans primarily.
On the stereo I've mainly been doing vinyl for serious listening lately. That system is in a state of flux. Planning to grab a Topping D30 or D50 soon.
I consider the Audient to be well into the realm of diminishing returns DAC wise. I've been perfectly satisfied with it's performance.
Either of those would lend itself to upgrading the OPA2134/OPA1612s with Burson FET modules.
I replaced the OPA2134s/OPA2604 in a Music Hall DAC 25.3 with Burson V5s for a marked improvement in transparency.
Is that the iD14?
Thoughts on the OPA627 pair on riser boards?
Yes, iD14. Some sort of discreet Burr brown DAC chip. Running line out to the Millett through 1/4" jacks. One of the reasons I want to grab the topping unit is to do a comparison with it's pre-outs. I've always found pro-audio gear to have pretty decent performance for the price and the mic pres in the thing are gorgeous.
Some say that's the best of the TI op amps. I'd still rather go with discrete FETs from companies like Burson (tall red modules). The Music Hall still can't match the Audio Research DAC8, but it is closer.
Nice! I personally think that the biggest gains to be had in a DAC are actually in what you're doing with the signal AFTER it's been converted to analog, hence my interest in the opamp choices. My natural skepticism leads me to believe that any DAC chip with sufficient SN ratios and dynamic range capability, referenced by a sufficiently stable Vref system and sufficiently buffered inputs/outputs will have similar performance to any other. Conversely, it is quite possible to screw up an analog signal with poor quality components/bad buffering/poor impedance matching/etc.
The analog stage that follows is as important as any other preamplifier in the signal chain.
Separate names with a comma.