OK If it is that good, I will try it. Thank you for the info.
But really the dac and analog/HPA all inside the computer probably isnt as good as an outboard dac like the BenchMark. anyway......
THx-Richard
Last edited:
You beat me to it, I was tempted to make a comment that it was also funny how folks still scramble around wondering if their audio is re-sampled, really 24 bit, etc.
It sounds like they have good reason to wonder. I never use a computer for listening to music... so how would i know. But, I know what bad sound is and so I went in and opened things up and reset them. I still intend to only store wave files on the computer.
Am I going to have problems off-ing those, too?
THx-RNMarsh
Benchmark has selected the ES9018 filters which provide the lowest pass-band ripple. We then frequency-shift the filter transition band upward so that it is centered at 110.5 kHz. We do this by operating the ES9018 at an input sample rate of 211 kHz. This means that the entire transition band of the ES9018 filter is always above the highest audio frequency contained in the incoming audio. At a 192 kHz incoming sample rate, the highest incoming frequency is 96 kHz. This is completely below the lower limit of the transition band that is centered at 110.5 kHz. Benchmark's system effectively eliminates the filters in the ES9018 by frequency shifting the filters out of the audio band. It also completely eliminates all traces of image fold-back. The Nyquist frequency of the D/A converter exceeds the Nyquist frequency of the incoming digital audio.
We used this same frequency-shifting technique in our DAC1 converters. When the DAC1 was designed, the available technology limited us to a D/A input sample rate of 110 kHz. In the DAC1, the D/A filters were out of band for sample rates up to 96 kHz. The DAC2 extends this unique technology to all sample rates up to 192 kHz. The goal for all Benchmark converters has always been to make the digital filters as transparent as possible. The accuracy and precision of the filters is a function of the oversampling ratio used in the filters. Benchmark moves the low-pass filter out of the D/A converter so that it can be executed at a much higher oversampling ratio.
We used this same frequency-shifting technique in our DAC1 converters. When the DAC1 was designed, the available technology limited us to a D/A input sample rate of 110 kHz. In the DAC1, the D/A filters were out of band for sample rates up to 96 kHz. The DAC2 extends this unique technology to all sample rates up to 192 kHz. The goal for all Benchmark converters has always been to make the digital filters as transparent as possible. The accuracy and precision of the filters is a function of the oversampling ratio used in the filters. Benchmark moves the low-pass filter out of the D/A converter so that it can be executed at a much higher oversampling ratio.
They stuck a TI SRC4392 before the DAC and choose an odd rate to convert to. Nothing wrong with it, but don’t think it’s the way I would prefer to do it.
I'm not good enough at the maths but with 211kHz not being a multiple of any known sampling frequency would suggest no input rate will be ideal. Certainly the marketing bumph you quoted doesn't help understand why.
Too many uses of the word 'unique' in the benchmark user manual.
Too many uses of the word 'unique' in the benchmark user manual.
They use that odd rate because some ASRCs had worse performance at 1:1 ratios and it’s just above 192 kHz. It also allows them to use a non audio frequency oscillator.
Benchmark is a cheater DAC!
Instead of using separate switchable oscillators for each Fs, proper PLL etc they just stuck one ASRC IC at the input and, strangely, got away with it. Must be either marketing power, consumer stupidity or both. All other manufacturers were playing fair at the time of its introduction.
Each time you play anything into it, the data output of the SRC IC is different. Yes, every playback. Plus, any jitter present in the stream is encoded in the output.
IMO, the ASRC process has no place in a DAC.
Last edited:
Hi Richard, thanks for putting up that info (that I probably already have, but never read) and reminding me of how serious Benchmark takes each digital problem. I didn't think about it like this before.
I agree with you, that computers 'suck' when it comes to high fidelity or even consistent playback of music. I never realized the problems with both Apple and Windows for handling these snippets of music. Next to useless.
Like we have known for decades, if possible just extend the usable recorded bandwidth to 40KHz or more, and the vast majority of these problems go away. What a concept!
Is it possible? Is it practical? Can anybody do it? WOW!
I agree with you, that computers 'suck' when it comes to high fidelity or even consistent playback of music. I never realized the problems with both Apple and Windows for handling these snippets of music. Next to useless.
Like we have known for decades, if possible just extend the usable recorded bandwidth to 40KHz or more, and the vast majority of these problems go away. What a concept!
Is it possible? Is it practical? Can anybody do it? WOW!
I wouldn’t go that far. Their approach is legitimate, if not my preference, especially given that good ASRCs are probably audibly transparent. It’s especially fine for a SPDIF input.
How do you know that's the reason? 😕 Seems to me there is likely more than that behind the decision.
The architecture is ASRC > FPGA (PCM interpolation filter & volume control) > DAC (with ASRC enabled and set to narrow bandwidth).
Clocking the ASRC a bit above 192kHz allows more frequency range for the FPGA filter to roll off. They might have gone higher, but the maximum frame rate for SRC4392 is 215kHz. 221kHz works with easy to obtain 27MHz clocks (that part is in agreement with your explanation).
Last edited:
John, your statement is similar to if one would have said "the Wurtlitzer jukebox is no good for vinyl playback" 😉
Computers have been reliably used in SOTA music playback since late '80s /early '90s. Just get the proper additional hardware and application(s). Not cheap though.