From hearing perspective, not necessarily "buried", because masking happens only within relative narrow band around the fundamental. For example you can create -6 dBFS white noise and mix in -26 dBFS 1 kHz sine and I bet you have no trouble hearing the 1 kHz. You can drop the level even to -46 dBFS and still hear it.
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Review and Measurements of PS Audio PerfectWave DirectStream DAC
- Thread starter amirm
- Start date
From hearing perspective, not necessarily "buried", because masking happens only within relative narrow band around the fundamental. For example you can create -6 dBFS white noise and mix in -26 dBFS 1 kHz sine and I bet you have no trouble hearing the 1 kHz. You can drop the level even to -46 dBFS and still hear it.
I get it, but I hope you're not saying that the 30dB noise over fundamental is inconsequential.
Isn't there a limit to that ?
Last edited:
No, depends on noise profile around the fundamental whether it masks or not. Flat noise floor, or shaped one, or some spurious tones, or something else. It is quite common that the relation between fundamental and noise floor is frequency dependent. (and you can adapt that to the hearing sensitivity profile)
Let me ask you for your honest expert opinion, everything else being equal, would there be any reason for you to pick a DAC with 30dB higher noise floor?
dc655321
Major Contributor
- Joined
- Mar 4, 2018
- Messages
- 1,597
- Likes
- 2,236
My choice of DAC is result of many performance factors, not just noise floor. I rather have 20 dB higher white noise floor than 20 dB lower correlated noise floor. If there's a bit higher white a noise floor I cannot hear, it is fine. But if a DAC has very low noise floor, but has annoying sonic anomalies due to bad digital filter profile, it is not fine. Or if the noise floor for example consists of correlated tones (intermodulation or otherwise). So there's no single straightforward parameter to define pick / no-pick.
P.S. I don't have a PS DAC, or any plan to have one. They just don't fit my requirements, no hard feelings.
Last edited:
Fair enough. And to be on topic, do you see any of these good properties in the product on hand?
It is not my choice for DAC. I just see world a little less simple and straightforward regarding conclusions and what the conclusions are based on.
I personally prefer DACs that are just D/A converters, convert digital values to analog, without any DSP involved (bit-perfect). And even better, not boring COTS chips, but something someone has actually created on their own.
digicidal
Major Contributor
Ordinarily I would agree - however, I don't mind this kind of thing at all when the company clearly states it and offers an identical product without it. They're just saying for an additional up charge they'll "bedazzle" your speakers with jewelry so your audiophile "street cred" value is higher. Nowhere do they claim the AQ cables (nor even the Mundorf components in the XO) improve the SQ at all... just that you can get the HIFI version with those extras installed. So in that sense it's not much different than offering custom finishes/veneers for an increase in price - which a number of reputable companies offer.
Exactly. That's why I put them in the middle group... far less "budget" appearance than the Khadas board itself (or one with acrylic covers) IMO, while adding the backing of a larger product line with a single point of sale and support. Performance is less than the top listed products (as is cost), while aesthetics are an improvement over the look of most lower cost options.
Obviously the consumer must consider whether they themselves think it's worth the difference. To me, the cost in hours alone, for most DIY projects, makes a finished product attractive - for someone more handy, with a full set of tools, and an area to work on projects undisturbed... then likely rolling your own is a better proposition.
That definitely reduces the selection pool to a very few units. I feel similarly, but also realize that the odds of an entirely proprietary design which eclipses the capabilities of a dedicated production house like ESS, TI, etc. in any (let alone all) areas is getting more and more unlikely with each product revision. Even in the few cases where it might be the case, the odds of there being no other issues - and of my being able to justify the cost - are pretty small.
Do you have an example of what you consider "nearly perfect" by your criteria - at any cost?
Last edited:
Last edited:
OP
- Thread Starter
- #773
I have not. I need to pack it away. So give me one test to run and I will do so.
OP
OP
- Thread Starter
- #775
Well, the limit is your patience.
At 44.1 khz, 256K points is already 8 seconds. You can do the math on how long it takes if you keep increasing this value to one million or whatever. The FFT will also get slower to compute as the number of points keeps going up. It is painful to wait for the 1 million point for example but I have done it.
OP
- Thread Starter
- #776
Here is the dashboard for RCA output:
OH, now that's much better. LOL
GrimSurfer
Major Contributor
- Joined
- May 25, 2019
- Messages
- 1,238
- Likes
- 1,484
The interesting thing is that one could excuse the performance of the DAC if all we were talking about was the 1st-4th harmonics of the power supply (60, 120, 180, 240 & 300 Hz spikes to the left of the 1 kHz signal). Not ideal, or even warranted, but their sum of roughly -116 dB isn't criminal...
...though it is different than the "44.1 kHz noise spectrum (nothing playing)" dashboard initially featured in the review (THAT was awful).
But the even and odd order distortion (2k, 3k, 4K etc. spikes to the right of the 1 kHz signal) is...
...though it is different than the "44.1 kHz noise spectrum (nothing playing)" dashboard initially featured in the review (THAT was awful).
But the even and odd order distortion (2k, 3k, 4K etc. spikes to the right of the 1 kHz signal) is...
Last edited:
I think it would be easy to prove the ASR methodology wrong. Somebody just has to make a DAC that passes all of the ASR tests at a good or great level but that also obviously sounds bad and doesn't recreate the music successfully.
Similar threads
