Came across this old 2012 statement from Benchmark VP John Siau regarding 96kHz and 192kHz. Is this info outdated and not applicable to most current DACs?
All of BenchmarkÂs A/D converters and D/A converters support sample rates up to 192kHz. However, we strongly recommend 96kHz for optimum performance. There is a performance penalty for operating at 192kHz. The problem is that all A/D and D/A converter chips operate at reduced oversampling ratios when converting at 192kHz. At the current time, the negative consequences of the reduced oversampling ratio far outweigh any benefits derived from the higher sample rates.
At 192kHz the stopband attenuation of the digital filters is usually much poorer than at 96kHz. Many converter ICs have 120dB of stopband attenuation at 96kHz, but only 80dB at 192kHz. This makes 192kHz converters very susceptible to aliasing and poor image rejection. These artifacts clutter the audible spectrum with low-level non-musical distortion.
It can be shown and demonstrated that there is no loss of time-domain accuracy when operating at 96kHz versus 192kHz. It is a myth that 192kHz gives better time-domain accuracy.
To date, Benchmark has no evidence that 192kHz performs better than 96kHz, but we have a substantial body of evidence that shows that 192kHz has defects that are not present at 96kHz. These issues are also shared openly by one of our competitors: Lavry Engineering. We suspect many other manufacturers are aware of these issues, but choose not to talk about them.
Bottom line: Be very careful about any claims that 192kHz sounds better than 96kHz. Our experience points in the opposite direction.
With my new DS being fed by my Oppo 205…I can set the 205’s digital output
to 44k, 96k, 192k or bitstream…
With my DS, all settings have a different sonic presentations…the higher the frequency
the more is resolved. The setting that the paddles on the side of my head most like
is the bitstream setting…Bitstream offers the best of 192k but with better definition for me.
One great thing I like about my Oppo 205 is that it will send the 192k in a 28bit format
to the DS…according to the DS screen readout.
Interesting. I have a plain 2.0 (no sub) channel system and have owned an Oppo 95 and for years now a 105D. Whether through boredom or the devil making work for idle hands I periodically revisit certain settings. On both models I’ve consistently returned to these as sounding the most “open” and just plain accurate.
Naw you are not hijacking anything…just adding more insights
to the topic.
We parallel the same …for me it is the ole tweeker just wanting to play
around with his toys…just like to revisit previous settings…even speaker
configurations… It pretty much seems we tend to like the same settings,
and end up with the same…great minds thinking along the same path…
Even speaker distance and trims…another toy to play with.
It’s like an icecream shop…what’t your flavor for today sort of thing…
Benchmark continues to support up to 192 kHz PCM with the latest DAC3 HGC. Siau has my deepest respect for not messing around. Benchmark measures and publishes the most thorough no nonsense data to back its’ DAC products in the industry. I take what he says about the strengths and weaknesses of the hardware seriously. I have a particular take on his comments and Benchmarks’ position on what the available hardware does well and not so well. My take is the consumer should be careful about succumbing to the hype from the industry that higher sample rates sound better simply because of higher sampling rate. Siau’s cautionary note is think twice before falling for it. Let’s be frank, sites like HDTracks and others want to sell you higher sample rate formats at a higher price (and profit margin). I have a number of recordings in my collection transferred at 96 kHz and 192 kHz and I have yet to convince myself, when I set aside my own confirmation bias, that 192 kHz is better when played back with the hardware manufacturers like Benchmark have access to. Which is very, very good. That is Siau’s point and I think it is right on point.
He is correct about many DAC chips of the era. His statements don’t apply to, say the DS which has the same stopband attenuation for all input rates.
He is correct that time resolution isn’t affected in principle by the sample rate. The sample rate primarily affects bandwidth. But bandwidth can affect our ability to sense clean impulses, tho I doubt that the difference between 96k and 192k is significant in the case of clean impulses.
I don’t know if some of the more modern DACs do higher internal upsampling which could help.
I guess I could add that (in the past anyway) on great systems I could relatively easily hear the difference between 44.1, 96, 192 and SACD. The differences on a great system weren’t in time resolution, sound stage, etc. They were in the amount of space surrounding each source. For example 44.1 just seemed more ghostly than SACD, Each performer was a little more diffuse. I still enjoyed listening to each different sample rate, it’s just that in back to back listening when changing to lower sample rates the positional precision of each performer suffered.
I don’t know about other DAC’s, but with the DS, 192k definitely sounds better than 96k. I’ve compared it many times with the exact same musics. 192k is simply more detailed and clearer sounding. I’ve even compared the same recording double rate DSD and quad rate DSD and quad rate has the edge. My brother was over here and he also heard the same thing.
I agree 100%. With my old PerfectWave MkII DAC, 96k & 192k didn’t sound that different to me, so I bought a lot of recordings back then in 96k, rather than spend the extra money for 192k. With the DirectStream DAC, it’s no contest…192k clearly sounds more natural to me, and tends to have the same general characteristics as DSD files. In fact…these days I often have very little preference for 24/96 vs. a good 16/44.1 version of the same album, but I always prefer the 24/192 vs. the 16/44.1.
It’s complicated. Many DAC chips now do things like 8x upsampling no matter what the input frequency is which is good. On the other hand the choices of output filters don’t necessarily present the best filter parameter choices:
Here are some plots out of the ES9038Q2M datasheet.
Note that some of the filters don’t get down past about -100dBFS and others stop around -120dBFS so perhaps all input frequencies are penalized the same…
Most of the digital streams and downloads I listen to are mastered in 24/96 PCM.
24/192 seems to be quite rarely used and more of an audiophile predilection used as a standard by the likes of Linn Records.
I’ve used devices with both commercial D/A and proprietary FPGA chips. I don’t think one is necessarily better than the other. Commercial chips have vastly more time invested in their development than any proprietary chip (except possibly the dCS RingDAC). I suspect it it more about how they are deployed.