Probably the internal OPA in AD1860 was not up to the task. That OPA in PCM56 is not something either. Only internal OPA in AD1865 is acceptable.
Denial. You think the people at Analog Devices produced a DAC chip with defective opamp? That the first CD players by Sony and Philips were designed by daft engineers that implemented output filtering (let alone oversampling) just because they felt like it?
We don't know anything as information is give partially but in a normal/average chain having no filtering at all is just awful to most ears. Waveform does only resemble what it should be. Only in the tube guys universe with everything odd, outside standards and inductive parts in abundance such experiments have a chance to succeed.
Non OS is another matter.
We don't know anything as information is give partially but in a normal/average chain having no filtering at all is just awful to most ears. Waveform does only resemble what it should be. Only in the tube guys universe with everything odd, outside standards and inductive parts in abundance such experiments have a chance to succeed.
Non OS is another matter.
Last edited:
I did not say that it is defective, but that it is not good enough. I tried the PCM56 with an internal OPA, it sounds worse than with an external one. Too soft, lacks a lot of detail in the sound.
I go way back to the days of Radio Shack catalogs which had 2x, and 4x oversampling digital filters on their CD players.
All digital audio goes through a DAC of some kind.
Those early CD players had a metallic harshness to them which was solved with better filtering hardware, software and faster silicon.
I remember a JVC 10-disc player I installed in a number of cars. We at work thought it was crap because compared to the pioneer it sounded like a tin can. Harsh, annoying and unlistenable for long periods. When installed in cars with factory stereos, it played through the FM tuner, which rolled off the highs naturally and the annoyance went away, albeit with the cost of lowering fidelity to that of FM radio.
Things have evolved much since then, but my answer is "of course dacs can sound different."
My latest experimentation was with a cambridge audio dac that had switchable settings. My bro could hear the differences, I could not although I consider myself more experienced in audio dissection.
All digital audio goes through a DAC of some kind.
Those early CD players had a metallic harshness to them which was solved with better filtering hardware, software and faster silicon.
I remember a JVC 10-disc player I installed in a number of cars. We at work thought it was crap because compared to the pioneer it sounded like a tin can. Harsh, annoying and unlistenable for long periods. When installed in cars with factory stereos, it played through the FM tuner, which rolled off the highs naturally and the annoyance went away, albeit with the cost of lowering fidelity to that of FM radio.
Things have evolved much since then, but my answer is "of course dacs can sound different."
My latest experimentation was with a cambridge audio dac that had switchable settings. My bro could hear the differences, I could not although I consider myself more experienced in audio dissection.
That's a separate story. We don't all hear the same, so it happens that something sounds great to someone, while to another it doesn't. Everyone is looking for a source of joy for themselves. 🤣
I recall building/finishing a class A JLH amplifier with a guy and testing various digital sources connected to that JLH. It all sounded pretty good but he was not satisfied at all. It was already noticed that opinions on devices quality to be about each others opposite. There was IMHO however no source not up to standard and the amplifier was breaking in nicely. Still he was not satisfied and looked for a tube device with tone control to insert.
He turned knobs, manipulated to a few extremes and this made the sound not exactly better (to the level that I would switch it off) but he was satisfied. Afterwards I understood he suffered from severe hearing loss in one ear and had tinnitus. Audio then seems a non optimal hobby 🙂
He turned knobs, manipulated to a few extremes and this made the sound not exactly better (to the level that I would switch it off) but he was satisfied. Afterwards I understood he suffered from severe hearing loss in one ear and had tinnitus. Audio then seems a non optimal hobby 🙂
Last edited:
I don't know, it's a fun hobby for me. I don't use any tone controls or anything like that. I can hear up to 14kHz without any anomalies, both ears approximately the same. I'm turning 62 soon, I think it's OK for my age.
In my place and in a couple of other places, in completely different systems and lots of other local DIY forum members. Price range of DACs up to 1000eur, to be compatible in price with these DIY machines. The point of our little research was whether it is worth making something in general, if there are factory devices of similar price and sound quality. They don't exist of course. I'm not interested in much more expensive devices, I'm not such an audio maniac.
Last edited:
Can you elaborate that?
Are you saying the effect of ~500kHz switching from class D has the same effect as 30kHz (if presented) signal feed to the speaker drivers?
The HF output is negligible compared to, DSD64. Also, some filtering ist always in place even for "filterless"NOS.
It's a complete non-issue.
You can be assured all the newly generated harmonics are above human hearing (and there's no intermod), OTOH this looks like a slew rate torture test (all these vertical lines), and that CAN affect reproduction in the audio range of an otherwise good amplifier.
Depending on edge risetimes, there could be frequency components up to around several hundred MHz. Some opamp inputs are not sufficiently protected against RFI up around there. If not, then demodulation of RF can occur in semiconductor junctions and produce DC offsets at the opamp outputs, if not worse.
Sorry it is a broken design. Doesn't meet the minimum requirements for the sampling theorem. It may sound different, it may sound to your preference, but without a filter is doesn't meet the requirements for being able to reconstruct the signal it is sent.
@931410: But NOS does, near enough for the intended application of human music listening, satisfy the Sampling Theorem. Why? Well, let’s take a closer look at some concepts. The term ‘unreconstructed’ is, to my mind, misleading. The raw, unfiltered, output of a DAC already contains the entire, fully completed and assembled analog audio base-band signal. The base-band doesn’t require ‘reconstruction’, or creation, as such. It only falsely appears to require that. To be incompleted. To just appears to still require recreation when viewed in the time-domain. Its full presence becomes obviously true when viewed in the frequency-domain. The audio base-band exists from, DC to the system Nyquist frequency, 22kHz for CD. The audio base-band exists along side the many ultrasonic image-bands, which are located mirror-folded above and below multiples of the sample rate. The base-band is merely falsely obscured, by the repeating image-bands, when viewed in the time-domain.
What reconstruction filtering does is to remove all the ultrasonic image-bands. Removal of the image-bands leaves only the signal base-band, and is what makes the signal now appear continuous rather than discrete, when viewed in the time-domain.
Back to NOS. The human ear is a biological reconstruction filter, regarding the application of digital audio. Because it filters away the ultrasonic image-bands, leaving the pass-band as ‘reconstructed’, per the Sampling Theorem. You can’t directly view the biological ‘reconstruction’ of an NOS signal, as far as I know, because it happens only after the ear converts the acoustic signal to a bio-electrical one. Again, the pass-band was always fully complete and present, it just that this fact is obscured by the additional presence of the image-bands.
What reconstruction filtering does is to remove all the ultrasonic image-bands. Removal of the image-bands leaves only the signal base-band, and is what makes the signal now appear continuous rather than discrete, when viewed in the time-domain.
Back to NOS. The human ear is a biological reconstruction filter, regarding the application of digital audio. Because it filters away the ultrasonic image-bands, leaving the pass-band as ‘reconstructed’, per the Sampling Theorem. You can’t directly view the biological ‘reconstruction’ of an NOS signal, as far as I know, because it happens only after the ear converts the acoustic signal to a bio-electrical one. Again, the pass-band was always fully complete and present, it just that this fact is obscured by the additional presence of the image-bands.
Last edited:
So the ear does some filtering, the speaker does some filtering other parts of the chain may do some filtering. It may be enough for the reconstruction to be accurate. If it is, then it would present to the ear/brain, the same signal as if you did the anti-imaging filtering at the DAC output and what the rest of the chain does make no difference. There certainly is no inherent reason it sound better to not filter, and let happenstance do the work.
I did not use un-reconstructed in my post. Only that you cannot properly reconstruct the signal without the filter. Maybe all the accidental filtering does the job to allow an accurate reconstruction at the ear compared to what the original recorded signal was or maybe it doesn't. We don't have to leave it to chance.
- Status
- Not open for further replies.