• 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!

Why do NOS dacs sound different to oversampling designs?

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,280
Location
Gold Coast, Queensland, Australia
It is possible to get an 18 bit Burr Brown DAC to 95.5dB SNR, including ground isolation, output buffering, etc. Been there, done that. Never EVER want to do that again. Yes, that's 16 bits, but that was the actual spec on the device, so I don't feel too bad about that.

An untrimmed PCM-58P maybe, but not a BB PCM-58P selection S.

8x O/S (CXD-8003S), all top 4 bits individually trimmed for minimum THD. Tested at >117dB S/N and THD below 0.0015%. 1989.

Bottom view showing the four trimpots for each D/A.
1565567342349.png


Top view. This is how Sony built the Sony X7esd.
1565567464382.png


And yes, they are LC-OFC potted and canned line transformers for the balanced outputs.
 

THW

Senior Member
Forum Donor
Joined
Nov 21, 2018
Messages
412
Likes
630
if there is any delta sigma thinness, I’m not hearing it.

all of my audiophile recordings sound just fine to me on my DX3 Pro.
 

beefkabob

Major Contributor
Forum Donor
Joined
Apr 18, 2019
Messages
1,636
Likes
2,073
There always have been a problem with delta-sigma converters (my most of experience is with the first generation one-bit) that many people, including salesmen in the show rooms describe as "lack of the body" or "thin presentation". I describe it as follows: the tones with strong amplitude are presented clean on the expense of everything that happens in the background. The most affected are lower amplitude tones with frequencies that are very close to the dominant tone, but also the same for the harmonics. In that way any reverbations, interaction with wood of the instrument, overtones are filtered out, (partially) removed like an unwanted noise. The dominant tone is presented clean, but without body, like salesmen said. Our measurements do not cover such situation.

That would be EASILY measurable as distortion.

Throw away your perceptions. They're worthless. For thousands of years, philosophers have known that our perceptions lie. Just ask my glasses. Just ask my mom's tinnitus. Instead, find something measurable and concrete.

Biases are real. Yours are apparent. I came in here a few months ago, and my first post was asking if ASR was going about it all wrong. After much reading here and elsewhere, I've realized that ASR, if imperfect, is definitely going about things the right way.

I also haven't been able to figure out what gear I want to buy. It's just too damned confusing... once you factor speakers into the equation. Before that, it's pretty easy. The measurements show the truth. I know that this component has this level of distortion, power, and so on. There are still tricky bits, though, like:
1. What's actually perceivable, beyond the specs?
2. What weirdness is messing things up? Like a badly made speaker wire (been mentioned here and there), a DAC that needs a warm-up period to perform (Sabaj D5), or a DAC that runs at its best SNR when it's at full blast (all the newest Benchmark DACs), even though it has a super high quality digital volume knob.
3. What's the best value then?
4. Room EQ?
5. If I bi-amp or tri-amp, am I better off just buying a pre-made speaker where they made all the parts work together?

And so on, ad nauseum.
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,685
Likes
10,700
Location
Hampshire
There always have been a problem with delta-sigma converters (my most of experience is with the first generation one-bit) that many people, including salesmen in the show rooms describe as "lack of the body" or "thin presentation". I describe it as follows: the tones with strong amplitude are presented clean on the expense of everything that happens in the background. The most affected are lower amplitude tones with frequencies that are very close to the dominant tone, but also the same for the harmonics. In that way any reverbations, interaction with wood of the instrument, overtones are filtered out, (partially) removed like an unwanted noise. The dominant tone is presented clean, but without body, like salesmen said. Our measurements do not cover such situation.
You seem to be confusing delta-sigma with MP3.
 

BDWoody

Chief Cat Herder
Moderator
Forum Donor
Joined
Jan 9, 2019
Messages
6,948
Likes
22,625
Location
Mid-Atlantic, USA. (Maryland)
I did actually mention that I never had better resolution of details comparing to the current (D30) setup. In that subject delta-sigma wins, I agree. So maybe we talk about different things?

You're going to turn down the offer of an EASY $500?

Should be a no brainer to prove all these haters wrong...don't ya think?
 

sajunky

Active Member
Joined
Jul 14, 2019
Messages
186
Likes
68
Location
South Africa
That would be EASILY measurable as distortion.
That is simply not true. There are many DSP processing methods that lead to no perceived distortions. Dynamic volume compression was used in communication systems since the beginning of 20th century. In analog domain, now in production of CD's (loudness war). With DSP you have much more possibilities that were impossible with analog filters. DSP processing on images can bring details out of the noise. The same in audio noise removal. You do apply a noise convolution function and it is done. Change parameters and -60dB signal of the same properties as a dominant signal signal is exposed. But you can do it in reverse, so the dominant signal comes clean on the expense of others, which I suspect is happening.
Throw away your perceptions. They're worthless. For thousands of years, philosophers have known that our perceptions lie. Just ask my glasses. Just ask my mom's tinnitus. Instead, find something measurable and concrete.

Biases are real. Yours are apparent. I came in here a few months ago, and my first post was asking if ASR was going about it all wrong. After much reading here and elsewhere, I've realized that ASR, if imperfect, is definitely going about things the right way.
ASR is doing good job. I hope it will be improving over the time, it is why I submitted my proposal for a research new methods. Couple years ago nobody was testing intermodulation distortions. Manufacturers work hard on improving measurements, some do not bring audible benefits, ESS by example. The old, proven PCM1794 sounds better than any ESS flagship chip. I don't trust ESS guys, everything is kept secret. Even simplified block diagrams are cheated, people are finding out that not everything works as is on the picture, as an example DSD data path and conversion in the latest chips. It is why users should also research new methods to uncover cheating.
I also haven't been able to figure out what gear I want to buy. It's just too damned confusing... once you factor speakers into the equation. Before that, it's pretty easy. The measurements show the truth. I know that this component has this level of distortion, power, and so on. There are still tricky bits, though, like:
1. What's actually perceivable, beyond the specs?
2. What weirdness is messing things up? Like a badly made speaker wire (been mentioned here and there), a DAC that needs a warm-up period to perform (Sabaj D5), or a DAC that runs at its best SNR when it's at full blast (all the newest Benchmark DACs), even though it has a super high quality digital volume knob.
3. What's the best value then?
4. Room EQ?
5. If I bi-amp or tri-amp, am I better off just buying a pre-made speaker where they made all the parts work together?

And so on, ad nauseum.
This is relaxing. A time to have a beer.
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
Couple years ago nobody was testing intermodulation distortions.

There's so much wrong in your post that it's hard to know where to start. But this one is particularly hilarious unless by "a couple years ago" you mean "80 years ago."
 

sajunky

Active Member
Joined
Jul 14, 2019
Messages
186
Likes
68
Location
South Africa
There's so much wrong in your post that it's hard to know where to start. But this one is particularly hilarious unless by "a couple years ago" you mean "80 years ago."
No comment, just to add. Methods were changing too.
 

j_j

Major Contributor
Audio Luminary
Technical Expert
Joined
Oct 10, 2017
Messages
2,267
Likes
4,758
Location
My kitchen or my listening room.
That is simply not true. There are many DSP processing methods that lead to no perceived distortions.

Whoa, there, dude, what's this "no perceived distortions" qualification all of a sudden.

You claimed that missing spectral components could not be MEASURED.

Of course they can.

You're just trolling. Either you don't understand what you're saying, or you're trolling.

Or both.

Couple years ago nobody was testing intermodulation distortions.

And for ((*&(*& sake, error spectra, which are anything but new, capture everything. Noise, IMD, THD (btw, IMD and THD are effectively the same thing in different forms) and about those measurements for IMD and THD from the 1950's?

You HAVE to be trolling. It's also obvious you don't know the difference between an error spectrum and a single-frequency THD test.
 

sajunky

Active Member
Joined
Jul 14, 2019
Messages
186
Likes
68
Location
South Africa
And for ((*&(*& sake, error spectra, which are anything but new, capture everything. Noise, IMD, THD (btw, IMD and THD are effectively the same thing in different forms) and about those measurements for IMD and THD from the 1950's?
Are you trying to say that measuring method for Noise, IMD, THD are the same?

Not going to reply, just wondering...
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
That is not what he said.

Error spectra capture all noise and distortion (effectively a SINAD test). SINAD is also used to generate ENOB (effective number of bits) that includes all noise and distortion.

IMD can be mathematically derived from THD measurements.
 

30 Ounce

Active Member
Joined
Apr 9, 2019
Messages
155
Likes
217
PCM 1794 sounds better than ESS’s flagship 9038 PRO? I don’t think so...you need better equipment, more focus or your head examined.
 

sajunky

Active Member
Joined
Jul 14, 2019
Messages
186
Likes
68
Location
South Africa
That is not what he said.

Error spectra capture all noise and distortion (effectively a SINAD test). SINAD is also used to generate ENOB (effective number of bits) that includes all noise and distortion.

IMD can be mathematically derived from THD measurements.
Still, there are different measuriing methods and for a reason. Some problems are only shown when changing a method, Amir sometimes have to do to find out why is this and this... Didn't see that?
 

j_j

Major Contributor
Audio Luminary
Technical Expert
Joined
Oct 10, 2017
Messages
2,267
Likes
4,758
Location
My kitchen or my listening room.
Are you trying to say that measuring method for Noise, IMD, THD are the same?

Not going to reply, just wondering...

Why don't you learn something rather than attempt to "frame the opponent" into saying something ignorant.

First, find out what an error signal is.
Then, find out what an error spectrum is.

Until you can do that, and explain it clearly, you are UNQUALIFIED TO SPEAK ABOUT MEASUREMENTS OF ANY SORT.
 

Thomas savage

Grand Contributor
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
10,260
Likes
16,298
Location
uk, taunton
@sajunky , I don't feel there's a honest attempt to communicate and learn coming from you , that or your hopelessly out of your depth and don't seem to realise.

In order to curtail what is now a fruitless back and forth I am issuing you with a temporary reply ban. Please take this as an opportunity to go back on what has been written , look up any references and learn a little bit about what's being discussed.

Thank you.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
I suspect sajunky is one of the 'we can't measure everything yet' proponents.
These folks believe currently used methods of measuring aspects using tones does not say everything.
A bit like that cable manufacturer that measures every aspect and can't find any explaining differences and just continues saying that hearing tells him otherwise.
They believe is that SD 'removes' musical aspects that do not show up in tone measurements.
We have seen more of those here lately.
What they fail to see is that a DAC simply outputs a sample value and then (according to an algorithm) lowers or increases the output voltage to the next described sample value.
A DAC does not care what those values are at all. properly filtered R2R is not better than SD.

The theory these guys have can 'easily' be shown NOT to exist using Paul's software and sampling at say 192kHz or higher the same (analog out) music played from different DACs.
If there were differences in 'amplitude compression', or other 'smearing' effects then they would show in a difference file.

Even IF those tests were done (which will show certain differences anyway) they will simply claim that things like the cable manufacturer did and demand we need more evidence and other testing methods but won't do a properly controlled blind test to show their remarkable abilities.
The ability to not even need a blind test because the differences are so vast and everyone, but engineers, can seem to hear them.
 

BDWoody

Chief Cat Herder
Moderator
Forum Donor
Joined
Jan 9, 2019
Messages
6,948
Likes
22,625
Location
Mid-Atlantic, USA. (Maryland)
I suspect sajunky is one of the 'we can't measure everything yet' proponents.
These folks believe currently used methods of measuring aspects using tones does not say everything.
A bit like that cable manufacturer that measures every aspect and can't find any explaining differences and just continues saying that hearing tells him otherwise.
They believe is that SD 'removes' musical aspects that do not show up in tone measurements.
We have seen more of those here lately.
What they fail to see is that a DAC simply outputs a sample value and then (according to an algorithm) lowers or increases the output voltage to the next described sample value.
A DAC does not care what those values are at all. properly filtered R2R is not better than SD.

The theory these guys have can 'easily' be shown NOT to exist using Paul's software and sampling at say 192kHz or higher the same (analog out) music played from different DACs.
If there were differences in 'amplitude compression', or other 'smearing' effects then they would show in a difference file.

Even IF those tests were done (which will show certain differences anyway) they will simply claim that things like the cable manufacturer did and demand we need more evidence and other testing methods but won't do a properly controlled blind test to show their remarkable abilities.
The ability to not even need a blind test because the differences are so vast and everyone, but engineers, can seem to hear them.

It's a sign of the success of the site...more and more people are cruising through and reading information they've never come across before. It seems to me that there are a lot more of the newcomers that are relieved to read this kind of truth, where they can have a DAC with total transparency for...$10...(for an example of the egregious heresy) than there are haters.

Unfortunately, you are right though...too many others have bought so far into the cult that admitting cost and performance are not as closely related as they have always been assured and reassured is just not an option.

I just hope they keep reading...
 
Top Bottom